Abstract
Electroacupuncture (EA) has emerged as a biologically plausible adjunctive strategy for modulating neuroimmune dysregulation in sepsis. However, several recent reviews have already summarized anti-inflammatory mechanisms, ST36-centered animal evidence, organ-protective effects, and pooled clinical signals of acupuncture or EA in experimental and clinical sepsis. The unmet need is therefore not another catalog of positive pathways, but a critical translational framework that asks when, in whom, and how EA-inspired neuromodulation should be evaluated within the clinical complexity of septic intensive care. This review integrates mechanistic, preclinical, clinical, safety, and trial-design literature to appraise the evidence hierarchy, clarify sepsis immunophenotype-based patient-selection logic, and identify ICU-specific barriers to clinical translation. We emphasize somato-autonomic reflexes, inflammatory reflex signaling, vagal-adrenal and cholinergic anti-inflammatory pathways, macrophage and T-cell regulation, gut barrier immunity, immune suppression, and biomarker-guided stratification. Current clinical evidence suggests possible adjunctive signals, particularly in sepsis-associated gastrointestinal dysfunction and inflammatory biomarker modulation, but available trials are small, heterogeneous, frequently unblinded, and often use usual-care rather than sham comparators. Accordingly, reported mortality signals should be interpreted as hypothesis-generating rather than established clinical efficacy. We propose that future research should prioritize phenotype-specific, sham-controlled, multicenter trials with standardized stimulation protocols, organ-specific endpoints, immune and barrier biomarkers, and rigorous ICU safety governance, with gastrointestinal dysfunction serving as the most actionable initial indication. Closed-loop or artificial intelligence-guided EA should be regarded only as a future, physician-supervised engineering perspective. EA remains a biologically plausible but clinically unproven adjunctive neuromodulatory strategy in sepsis; its value for hard outcomes such as mortality, durable organ protection, and long-term recovery requires confirmation in rigorously designed ICU trials.
1 Introduction
Sepsis is no longer conceptualized as a simple systemic inflammatory response. The Sepsis-3 definition emphasizes life-threatening organ dysfunction caused by a dysregulated host response to infection (). Global estimates indicate that sepsis contributed to approximately 48.9 million cases and 11.0 million sepsis-related deaths in 2017, corresponding to nearly one fifth of all global deaths (). Contemporary international guidelines prioritize time-sensitive antimicrobial therapy, source control, hemodynamic optimization, vasopressor support, corticosteroids in selected shock phenotypes, mechanical ventilation, renal replacement therapy, nutrition, and other organ-support strategies (). These measures remain the therapeutic foundation of sepsis care.
Nevertheless, sepsis-associated organ dysfunction frequently progresses despite technically adequate source control and hemodynamic support. One explanation is that sepsis is immunologically dynamic. Early inflammatory amplification, endothelial activation, complement and coagulation dysregulation, mitochondrial dysfunction, lymphocyte apoptosis, impaired antigen presentation, and macrophage exhaustion may overlap within the same patient (, ). This complexity has undermined many single-target anti-inflammatory and immunostimulatory strategies. A rational adjunctive therapy would therefore need to be titratable, context-sensitive, and less likely to abolish host antimicrobial defense.
Acupuncture has historically been used in inflammatory and gastrointestinal disorders. Bioelectronic medicine and neuroimmunology provide a mechanistic framework for understanding how peripheral neural stimulation may influence immune responses (). The anatomical discovery of hardwired neuroimmune circuits, such as the vagal-splenic axis and the cholinergic anti-inflammatory reflex, demonstrates that the central nervous system continuously monitors and intricately calibrates peripheral innate immunity (, ). Within this framework, electroacupuncture (EA) can be viewed as a reproducible form of somatosensory electrical stimulation rather than as a conventional anti-infective therapy. By applying precise electrical parameters to specific peripheral somatosensory coordinates (), EA may engage autonomic and neuroendocrine reflexes that modulate inflammatory responses in experimental settings. This allows for the precise, dose-dependent dampening of localized hyperinflammation and the preservation of critical organ barriers, circumventing the systemic toxicity of conventional pharmacological agents (). Recent reviews have summarized the anti-inflammatory and organ-protective effects of EA in sepsis, including a systematic review of excessive inflammatory responses and organ damage () and a 2025 rodent-literature review focused on ST36-mediated neuro-endocrine-immune regulation, exosome biology, and multi-organ protection (). A previous ST36-focused systematic review of experimental sepsis also highlighted substantial laboratory evidence but emphasized methodological heterogeneity and limited translatability (). Therefore, the present review shifts the focus from whether EA is biologically plausible to how it should be critically evaluated, safely implemented, and rigorously tested as an adjunctive neuroimmune intervention in sepsis. We integrate selected concepts from newer mechanism-oriented literature, including local acupoint biology, afferent recruitment, gut microbiota-metabolite axes, and stimulation-dose standardization, while deliberately avoiding overextension from non-sepsis EA fields. The central argument is that EA can be considered only as an adjunct to guideline-based sepsis care and that its potential clinical value should be tested through ICU-compatible, phenotype-specific, sham-controlled trials with organ-specific endpoints and robust safety governance. As illustrated in Figure 1, sepsis-associated organ dysfunction arises from reciprocal interactions among innate immune activation, autonomic imbalance, endothelial-microvascular injury, barrier failure, and mitochondrial-metabolic dysfunction.
Figure 1
2 Scope, evidence audit, and distinction from previous reviews
This article is a critical translational review rather than a PRISMA-style systematic review; therefore, the search strategy was designed to support transparent evidence mapping rather than quantitative evidence synthesis. It uses previous systematic reviews, rodent-literature reviews, mechanistic studies, randomized trials, ICU safety evidence, clinical guidelines, and reporting standards to identify translational gaps and trial-design priorities. Because recent reviews have already cataloged ST36-centered animal evidence and broad anti-inflammatory mechanisms in sepsis (, ), the present review does not reproduce a complete inventory of all ST36 animal studies. Instead, it asks which mechanistic signals are sufficiently credible to justify clinical testing, which ICU factors may invalidate simple extrapolation, and which endpoints and safety criteria are needed for rigorous evaluation.
To improve transparency, a targeted literature search was performed in PubMed/MEDLINE, Embase, Web of Science, Cochrane Library, and ClinicalTrials.gov. The search covered studies published from database inception to May 2026. Search terms combined sepsis-related keywords (“sepsis,” “septic shock,” “endotoxemia,” “lipopolysaccharide,” “cecal ligation and puncture,” “sepsis-associated organ dysfunction”) with acupuncture-related keywords (“acupuncture,” “electroacupuncture,” “transcutaneous electrical acupoint stimulation,” “ST36,” “Zusanli,” “PC6,” “Neiguan,” “vagus nerve,” “cholinergic anti-inflammatory pathway,” “bioelectronic medicine”) and implementation-related keywords (“intensive care,” “ICU,” “safety,” “adverse events,” “contraindication,” “randomized controlled trial,” “sham acupuncture,” “trial design,” “heart rate variability,” “biosensor,” “artificial intelligence”). Eligible evidence included mechanistic animal studies, human randomized or controlled clinical studies, systematic reviews/meta-analyses, ICU safety literature, reporting guidelines, sepsis guidelines, and clinical trial protocols relevant to EA or acupuncture-based interventions in sepsis. Studies were prioritized when they directly addressed sepsis, endotoxemia, CLP, sepsis-associated organ dysfunction, ICU implementation, or clinical trial design. Non-sepsis acupuncture studies were used only when they clarified general biological or methodological principles, and were explicitly identified as indirect evidence. Retracted publications were not used as efficacy evidence. No quantitative synthesis or formal risk-of-bias scoring was performed because the aim was translational evidence mapping rather than meta-analysis.
The evidence was interpreted according to hierarchy and clinical relevance. Preclinical data were treated as mechanistic plausibility, not proof of ICU efficacy. Pooled effects from small clinical trials were interpreted as hypothesis-generating when GRADE certainty was low or very low. We also distinguished between statistical signals and clinically actionable outcomes. A reduction in inflammatory biomarkers or disease-severity scores may support biological activity, but it does not establish survival benefit, durable organ protection, or routine ICU readiness. Retraction status and methodological quality were also considered.
2.1 Immunological framework and patient-selection logic
A central premise of this review is that EA should not be considered a uniform anti-inflammatory intervention for all septic patients. Human sepsis is immunologically heterogeneous, with overlapping hyperinflammatory, immunosuppressed, endothelial-injury, metabolic, and mixed phenotypes (, , –). Patients with early hyperinflammation, preserved autonomic responsiveness, gastrointestinal barrier injury, and measurable inflammatory activation may theoretically differ from patients with profound lymphopenia, reduced monocyte HLA-DR, secondary infection, or immune exhaustion (, ). Therefore, future EA studies should incorporate immune phenotyping rather than enrolling unselected septic populations. Candidate stratification variables include lymphocyte count, monocyte HLA-DR, neutrophil-to-lymphocyte ratio, cytokine profiles, lactate kinetics, shock status, vasopressor exposure, organ-dysfunction trajectories, and autonomic signal quality when available (–). This immunophenotype-guided approach is essential to distinguish potentially modifiable neuroimmune activation from late-stage immunoparalysis or irreversible organ failure.
To clarify the novelty of the present review, Table 1 compares our scope with recent reviews of acupuncture or EA in sepsis. Previous reviews have primarily summarized ST36-centered animal evidence, anti-inflammatory mechanisms, organ-protective pathways, gastrointestinal dysfunction, or pooled clinical efficacy. By contrast, the present review focuses on evidence hierarchy, sepsis immune phenotypes, ICU-specific confounders, safety governance, and phenotype-specific trial design. Thus, its primary contribution is not the addition of another mechanistic catalog, but the development of a translational framework for determining when, in whom, and how EA-inspired neuroimmune modulation should be tested in septic ICU populations.
Table 1
| Review | Main focus | Evidence base | Key contribution | Limitation addressed by the present review |
|---|---|---|---|---|
| Lai et al. () | ST36 in experimental sepsis | Animal studies | Systematic summary of ST36 preclinical evidence | Limited discussion of ICU implementation, immunophenotyping, and trial design |
| Yang et al. () | Biological mechanisms of acupuncture in sepsis-related inflammation and organ injury | Mechanistic and preclinical studies | Broad catalog of anti-inflammatory and organ-protective mechanisms | Less emphasis on evidence hierarchy, ICU confounders, and clinical readiness |
| Zeng and Yan () | ST36-centered rodent literature and preventive-treatment concepts | Rodent studies | Focused discussion of neuro-endocrine-immune regulation and multi-organ protection | Mainly preclinical; limited focus on human sepsis heterogeneity and ICU trial feasibility |
| Xian et al. (71) | Clinical efficacy and safety of acupuncture as complementary therapy | RCT meta-analysis | Pooled estimates for disease severity, biomarkers, gastrointestinal outcomes, and mortality | Low/very-low certainty evidence; limited mechanistic and phenotype-specific trial-design appraisal |
| Present review | EA-inspired neuroimmune modulation in septic ICU care | Mechanistic, clinical, safety, implementation, and trial-design evidence | Evidence appraisal, sepsis immune phenotypes, ICU barriers, GI-focused trial framework, and safety governance | Addresses how, when, and in whom EA should be tested rather than simply listing mechanisms |
Comparison between the present review and previous reviews on acupuncture/electroacupuncture in sepsis.
To make this hierarchy explicit, Table 2 categorizes the available literature by study type, translational relevance, principal limitations, and overall certainty. This matrix is not intended to replace formal GRADE assessment, but to prevent mechanistic plausibility, preclinical efficacy, pooled low-certainty clinical signals, and speculative engineering concepts from being treated as equivalent levels of evidence.
Table 2
| Evidence category | Representative evidence | Main implication | Key limitations | Overall certainty |
|---|---|---|---|---|
| Local acupoint biology and afferent recruitment | Adenosine/A1 receptor signaling, fascia-associated mechanosensory activation, PROKR2Cre-marked sensory neurons at ST36 | Supports the concept that EA acts through anatomically constrained somatosensory interfaces | Most data are from pain, endotoxemia, or non-ICU models; afferent recruitment in sedated, edematous, vasopressor-dependent patients is unproven | Moderate for biological mechanism; low for septic ICU translation |
| Somato-autonomic and inflammatory reflex mechanisms | Vagal-adrenal signaling, vagal-splenic cholinergic anti-inflammatory pathway, α7nAChR-dependent macrophage modulation | Provides a coherent neuroimmune rationale for adjunctive modulation of inflammation | Circuit behavior may differ with shock, exogenous catecholamines, lymphopenia, autonomic failure, and immune exhaustion | Moderate in preclinical models; low in human sepsis |
| Rodent sepsis and endotoxemia models | LPS and CLP models evaluating cytokines, survival, gut barrier, lung injury, brain injury, kidney injury, liver injury, and cardiac dysfunction | Suggests anti-inflammatory and organ-protective signals across multiple organs | Young healthy animals, pretreatment designs, limited supportive care, and model-specific immune kinetics limit generalizability | Low to moderate |
| Organ-specific clinical signals | Small RCTs and controlled studies assessing gastrointestinal dysfunction, intra-abdominal pressure, inflammatory biomarkers, APACHE II, immune indices, and selected organ outcomes | Suggests possible adjunctive activity, especially for gastrointestinal dysfunction | Small samples, single-center designs, limited blinding, usual-care comparators, variable EA protocols, and incomplete adverse-event reporting | Low |
| Systematic reviews and meta-analyses | Sepsis-focused acupuncture/EA meta-analyses reporting pooled effects on severity scores, biomarkers, intra-abdominal pressure, and 28-day mortality | Helps identify candidate outcomes and recurring signals | Pooled estimates are limited by low/very-low GRADE certainty, heterogeneity, possible publication bias, and frequent absence of sham controls | Low or very low |
| ICU safety and implementation evidence | Acupuncture safety literature, ICU implementation considerations, CONSORT-STRICTA reporting, contraindication guidance | Defines practical boundaries for protocol design and safety governance | Limited sepsis-specific adverse-event data; risks differ across coagulopathy, devices, edema, shock, and staffing environments | Low |
| AI-guided and closed-loop neuromodulation concepts | Bioelectronic medicine, HRV, biosensors, DECIDE-AI, TRIPOD+AI, CONSORT-AI | Offers a future roadmap for phenotype recognition, safety monitoring, and protocol adherence | No direct evidence for AI-guided or closed-loop EA in septic ICU populations; signal quality and regulatory validation remain unresolved | Very low/theoretical |
Evidence matrix and certainty assessment for electroacupuncture in sepsis.
3 Neuroimmune mechanisms under ICU-specific confounding
3.1 Local acupoint biology and afferent recruitment
Local acupoint biology provides an additional layer for understanding how peripheral stimulation may be converted into neuroimmune signals. Experimental work outside sepsis has shown that EA can increase local extracellular adenosine and activate adenosine A1 receptors, contributing to local antinociceptive effects (). Other studies suggest that needling and electrical stimulation can interact with mechanosensitive somatosensory afferents and deep fascia. In sepsis-related neuroimmune research, the most influential finding is that low-intensity stimulation at the hindlimb ST36 region can recruit PROKR2Cre-marked sensory neurons innervating deep hindlimb fascia, thereby driving the vagal-adrenal anti-inflammatory axis in mice (). These observations support a refined concept: the acupoint is not simply a symbolic location but a neuroanatomically and biophysically constrained interface. However, most local acupoint microenvironment studies derive from pain or non-sepsis models. Their relevance to mechanically ventilated, sedated, edematous, or vasopressor-dependent septic ICU patients remains indirect. Future studies should therefore measure afferent recruitment, autonomic output, local tissue impedance, edema, perfusion status, and stimulation tolerability rather than assuming that a stimulation protocol effective in rodents or ambulatory subjects recruits the same sensory circuits in critical illness.
3.2 Somato-autonomic reflexes and inflammatory reflexes
The best-established mechanistic rationale for EA in sepsis involves somato-autonomic reflexes and inflammatory reflex pathways. Sciatic nerve activation with EA was reported to control systemic inflammation in polymicrobial peritonitis through vagal activation of adrenal dopamine production (). Vagus nerve stimulation attenuates systemic inflammatory responses to endotoxin, and the inflammatory reflex has been described as a neural circuit that senses and modulates peripheral inflammation (, –). In canonical models, vagal output influences splenic sympathetic signaling; norepinephrine from splenic nerve terminals acts on beta2-adrenergic receptors on choline acetyltransferase-positive T cells, which then release acetylcholine to activate alpha7 nicotinic acetylcholine receptors (α7nAChR) on macrophages (). Activation of this pathway suppresses pro-inflammatory cytokine production, including TNF-α, without necessarily eliminating antimicrobial phagocytic capacity (). This circuit provides a plausible basis for neuromodulation, but its translation to advanced septic shock is not guaranteed, particularly when lymphopenia, autonomic dysfunction, or macrophage exhaustion is present.
Across preclinical endotoxemia and CLP models, EA has been linked to several convergent inflammatory modules, including TLR4/MyD88/NF-κB signaling, HMGB1-related late inflammatory signaling, macrophage polarization, and programmed cell death pathways. However, these mechanisms are model-, organ-, acupoint-, and timing-dependent, and should not be interpreted as a uniform mechanism applicable to all septic patients (–). These mechanisms are biologically coherent but not equivalent to clinical efficacy. Many experiments use pretreatment before lipopolysaccharide (LPS) injection or cecal ligation and puncture, whereas ICU clinicians treat established sepsis ().
Modern neuroanatomical studies have substantially strengthened the biological plausibility of EA. Experimental work shows that EA effects are somatotopically organized and intensity dependent (, ). Low-intensity stimulation at hindlimb ST36 can recruit PROKR2Cre-marked sensory neurons and drive the vagal-adrenal anti-inflammatory axis in mice, whereas high-intensity stimulation may recruit different sympathetic pathways (). Another conceptual risk is oversimplifying sepsis into hyperinflammation alone. EA may reduce inflammatory mediators in early or hyperinflammatory phenotypes, but patients with profound immunosuppression could theoretically require immune restoration rather than further anti-inflammatory signaling (). The observation that EA may prevent T-cell lymphopenia in septic mice () is therefore intriguing, because it suggests context-dependent immune regulation rather than uniform suppression. Nevertheless, the mechanism is not a single linear pathway. The anti-inflammatory effects of EA involve multiple systems, levels, and targets and should not be reduced to vagal-adrenal activation alone (). Depending on site, intensity, timing, and disease state, EA may recruit vagal-adrenal, vagal-splenic sympathetic, spinal-sympathetic, hypothalamic-pituitary-adrenal, and local immune-neural pathways. As shown in the left and middle panels of Figure 2, ST36-associated somatosensory afferents may engage vagal-adrenal and vagal-splenic cholinergic outputs; the right panel highlights why stimulation site, intensity, frequency, and disease stage should be treated as dose-defining variables.
Figure 2
3.3 ICU confounders that may modify neuroimmune responsiveness
The major translational risk is that neuroimmune circuits demonstrated in controlled models may not behave similarly in septic ICU patients. The PROKR2-ST36-vagal-adrenal axis is primarily established in mouse models and should not be overstated as a proven human ICU mechanism. Septic shock patients frequently receive exogenous catecholamines, particularly norepinephrine as first-line vasopressor therapy (). This creates a major interpretive problem: exogenous adrenergic support may compete with, mask, or override the endogenous catecholamine signal generated by EA-driven vagal-adrenal activation (, ). Vasopressor dose, shock reversibility, adrenergic receptor desensitization, and baseline autonomic failure may therefore determine whether the same electrical stimulus produces measurable neuroimmune output.
Sedation and neuromuscular blockade introduce a second layer of uncertainty. ICU patients commonly receive opioids, propofol, benzodiazepines, dexmedetomidine, or neuromuscular blocking agents according to clinical need, and these therapies alter arousal, nociceptive processing, autonomic tone, ventilator synchrony, and heart-rate variability (HRV)-derived autonomic indices (, ). Whether deep sedation or paralysis impairs EA recruitment of PROKR2+ somatosensory afferents is unknown. Translational studies should therefore record Richmond Agitation-Sedation Scale scores, analgesic and sedative exposure, neuromuscular blocker use, norepinephrine-equivalent dose, arrhythmia or pacing status, HRV quality metrics, immune phenotype, and organ-function trajectories rather than assuming uniform circuit responsiveness.
Species- and model-specific differences represent another major translational boundary. Most mechanistic pathways discussed above were derived from young, otherwise healthy rodents exposed to LPS or CLP, whereas human sepsis is heterogeneous with respect to age, comorbidity, infection source, pathogen burden, microbiome composition, immune phenotype, timing of treatment, anesthesia exposure, and organ-support intensity (, ). Rodent LPS models capture acute endotoxemia rather than the full biology of polymicrobial infection, while CLP more closely resembles peritonitis but still lacks the diagnostic delay, antimicrobial exposure, source-control variability, fluid resuscitation, vasopressors, ventilation, and renal replacement therapy that shape human ICU sepsis. Therefore, rodent EA mechanisms should be interpreted as hypothesis-generating biological signals rather than directly transferable treatment effects.
4 Organ-specific mechanisms and clinically meaningful endpoints
Earlier reviews have described broad organ-protective effects of EA in sepsis (–). To strengthen clinical relevance, this review reframes organ protection as an endpoint-design problem. The key question is not merely whether EA reduces cytokines in an organ, but which clinically meaningful, organ-specific endpoints should be measured in future trials. Figure 3 organizes the organ-specific discussion by separating shared neuroimmune mechanisms from candidate organ-level endpoints, including gut barrier integrity, alveolar-capillary leakage, neuroinflammation, renal tubular injury, hepatic inflammation, and septic cardiomyopathy.
Figure 3
To avoid implying equivalent readiness across organ systems, the following sections distinguish three levels of evidence: preclinical mechanistic evidence, early clinical signal, and speculative extrapolation. Gastrointestinal dysfunction currently has the most clinically actionable evidence, whereas lung injury/ARDS, sepsis-associated encephalopathy, kidney injury, hepatic injury, septic cardiomyopathy, and ICU-acquired weakness remain predominantly preclinical or exploratory.
4.1 Gastrointestinal dysfunction, barrier injury, and gut-derived crosstalk
The intestine is the most coherent organ target for clinical translation. Septic gastrointestinal dysfunction can amplify systemic inflammation through barrier disruption, ileus, dysbiosis, bacterial translocation, intra-abdominal hypertension, and impaired nutrition (–). Preclinical studies suggest that EA at ST36 may preserve intestinal mucosal immune and mechanical barriers, maintain tight-junction proteins such as occludin and ZO-1, modulate ghrelin and HMGB1, and reduce epithelial injury (, , , 37). More recent work links EA to mitochondrial quality control and HO-1/PINK1-mediated regulation of barrier function (), and to macrophage polarization in intestinal injury ().
Clinically, the gastrointestinal literature provides the most actionable endpoints: intra-abdominal pressure, acute gastrointestinal injury grade, abdominal perimeter, bowel sounds, time to enteral feeding tolerance, gastric residual volume, diarrhea or constipation, intestinal fatty acid-binding protein, diamine oxidase, D-lactate, and microbiome or endotoxemia markers (, , 38, 39). These outcomes are closer to the presumed mechanism than mortality and may require smaller trials to detect signal. Therefore, gastrointestinal dysfunction should be considered the preferred first indication for future EA trials in sepsis. The apparent advantage of gastrointestinal outcomes should therefore be interpreted cautiously. It may reflect a biologically plausible gut-autonomic-immune interface: ST36-centered stimulation is closely linked to gastrointestinal motility, vagal signaling, mucosal immune regulation, epithelial tight-junction integrity, barrier permeability, and microbiota-metabolite crosstalk (, , , , , 37). At the same time, it may also reflect evidence-availability bias, because gastrointestinal endpoints such as intra-abdominal pressure, AGI grade, bowel sound recovery, and enteral feeding tolerance have been assessed more frequently than cardiac, renal, neurological, or long-term functional outcomes in existing clinical studies (40–43). Thus, gastrointestinal dysfunction is best viewed as the most clinically developed indication, not as proof that EA is intrinsically more effective for the gut than for other organ systems.
4.2 Acute lung injury and ARDS
The lung evidence is promising but mainly preclinical. Early studies showed that EA stimulation at ST36 attenuated LPS-induced acute lung injury in rats (44). Subsequent studies implicated Nrf2/ARE signaling (45), JAK1/STAT3 signaling (46), SIRT1-dependent pathways (47), macrophage polarization (48), and alpha7 nicotinic acetylcholine receptor-mediated anti-inflammatory signaling (49). A translational study reported that EA promoted resolution of inflammation by modulating specialized pro-resolving mediators, particularly lipoxin A4, via vagus nerve and alpha7 nicotinic acetylcholine receptor-dependent mechanisms in LPS-induced acute lung injury, with preliminary human observations in sepsis-related ARDS (50).
Despite these mechanistic advances, ARDS is a high-risk clinical domain. Ventilation strategy, driving pressure, PEEP, prone positioning, fluid balance, neuromuscular blockade, corticosteroid use, and infection source strongly influence outcomes (51, 52). EA trials in lung injury models should therefore avoid relying only on cytokines and should include clinically meaningful ARDS endpoints, including ventilator-free days, oxygenation indices, imaging or lung ultrasound findings when feasible, epithelial/endothelial injury biomarkers, barotrauma, sedative or neuromuscular-blocker exposure, and mortality (53). Registered trials examining EA in sepsis-related ARDS and specialized pro-resolving mediator metabolism should be followed (54), but their results should not be presumed. Accordingly, lung injury/ARDS should currently be considered a preclinical and early-translational domain rather than a clinically ready indication for EA in sepsis.
4.3 Sepsis-associated encephalopathy (SAE)
SAE encompasses delirium, impaired consciousness, long-term cognitive impairment, blood-brain barrier dysfunction, microglial activation, oxidative stress, synaptic injury, and neuronal death (55). EA pretreatment or treatment has been reported to attenuate microglial activation, oxidative stress, apoptosis, HMGB1/TLR4 and HMGB1/RAGE signaling, and neuronal pyroptosis in experimental sepsis models (56–60). These studies support a neuroimmune rationale, especially because peripheral neuromodulation may influence central inflammation without direct intracranial intervention.
However, the clinical evidence for SAE remains insufficient. Cognitive outcomes are difficult to measure in sedated, mechanically ventilated patients. Future trials should move beyond inflammatory readouts alone and incorporate clinically meaningful neurologic and cerebral perfusion-related endpoints. Near-infrared spectroscopy may provide dynamic information on regional cerebral oxygenation, whereas transcranial Doppler may estimate cerebral blood-flow velocity and cerebrovascular autoregulation (61, 62). These tools should be combined with delirium assessment, sedation exposure, electroencephalography (EEG) when available, and long-term cognitive outcomes (63), and should be interpreted as adjunctive monitoring modalities rather than definitive evidence of neurologic efficacy. Thus, SAE remains a biologically plausible but clinically unproven target, and future studies should treat neurological outcomes as exploratory until validated in adequately controlled ICU trials (64).
4.4 Acute kidney injury and hepatic injury
Kidney and liver evidence is largely derived from endotoxemia models. In LPS-stimulated rabbits, EA ameliorated acute renal injury through PI3K/Akt/Nrf2 signaling and HO-1 induction (65). Earlier rat studies also suggested attenuation of inflammatory responses and acute kidney injury after EA pretreatment (66), and a Wistar sepsis model reported effects on urea and creatinine (67). Hepatic studies indicate that EA at PC6 may attenuate liver injury in endotoxemic rats, and other studies reported effects on hepatic blood flow and lipid peroxidation in septic rats (68).
These data provide mechanistic plausibility but remain far from clinical implementation. Sepsis-associated acute kidney injury is influenced by shock duration, nephrotoxic drugs, venous congestion, rhabdomyolysis, contrast exposure, and renal replacement therapy (69, 70). Hepatic dysfunction is similarly multifactorial. Trials should use KDIGO criteria, renal replacement therapy-free days, creatinine kinetics, urine output, tubular injury biomarkers, bilirubin, coagulation parameters, lactate clearance, and hepatocellular injury markers rather than isolated inflammatory biomarkers (71, 72). Therefore, kidney and liver outcomes should currently be regarded as mechanistic or exploratory domains rather than established clinical indications.
4.5 Septic cardiac dysfunction and sepsis-induced myopathy
Septic cardiac dysfunction and ICU-acquired weakness are clinically important because they influence shock persistence, vasopressor exposure, ventilator weaning, rehabilitation, and long-term disability (73, 74). Preclinical studies suggest that EA pretreatment may attenuate sepsis-induced cardiac inflammation and dysfunction through calpain-2/STAT3 signaling (75), and vagus nerve-dependent mechanisms (76). In animal models, this intervention was associated with preserved left ventricular contractility and reduced local cardiac inflammation (76). However, these observations should not be interpreted as proof of clinical cardioprotection. Septic cardiomyopathy is a heterogeneous syndrome involving systolic dysfunction, diastolic dysfunction, right ventricular impairment, altered ventriculo-arterial coupling, myocardial strain abnormalities, and catecholamine-dependent changes in contractility (77, 78).
A key translational issue is the dissociation between macro-hemodynamics and microcirculatory perfusion, often described as hemodynamic incoherence. Septic shock resuscitation can normalize mean arterial pressure and cardiac output while tissue perfusion, endothelial glycocalyx integrity, capillary flow heterogeneity, and mitochondrial oxygen utilization remains impaired. This phenomenon provides a rationale for monitoring microcirculation and tissue perfusion in addition to conventional systemic hemodynamic variables (79, 80). EA-related autonomic modulation could theoretically influence vascular tone and inflammatory endothelial injury, but this signal may be masked or overridden by norepinephrine, fluid loading, inotropes, and vasoplegia. Therefore, future studies should not evaluate cardiac effects solely through inflammatory biomarkers or left ventricular ejection fraction (81, 82). They should incorporate integrated cardiovascular phenotyping, including vasopressor-inotrope dose, lactate clearance, capillary refill time or peripheral perfusion indices, microcirculatory imaging where feasible, bedside echocardiography, and markers of endothelial or mitochondrial injury.
At the cellular level, septic cardiomyopathy involves mitochondrial dysfunction, impaired fatty-acid oxidation, altered glucose utilization, oxidative stress, calcium-handling abnormalities, and inflammatory myocardial depression (74, 83). Future EA studies should therefore test whether neuromodulation modifies myocardial metabolic stress or mitochondrial injury rather than merely reducing circulating cytokines (83). Echocardiographic endpoints should include left ventricular ejection fraction or fractional area change, tissue Doppler indices, E/e′ or other diastolic parameters, right ventricular function, strain when feasible, cardiac output, and vasopressor-inotrope exposure (84). For sepsis-induced myopathy, trials should include Medical Research Council sum score, handgrip strength when feasible, electrophysiology, diaphragm ultrasound, ventilator liberation, mobilization milestones, and long-term functional status (85, 86). Overall, cardiac dysfunction and ICU-acquired weakness remain exploratory targets, and EA-related benefits in these domains require prospective clinical validation before any claim of clinical readiness can be made.
Together, these organ-specific sections show why the evidence matrix in Table 2 and the endpoint hierarchy in Table 3 are needed to separate mechanistic plausibility from clinically actionable trial outcomes.
Table 3
| Domain | Preferred primary clinical endpoints | Secondary clinical endpoints | Mechanistic/biomarker endpoints | Key ICU confounders |
|---|---|---|---|---|
| Gastrointestinal dysfunction | AGI-grade improvement, GI failure-free days, enteral feeding tolerance | IAP, bowel sound recovery, abdominal perimeter, gastric residual volume, diarrhea/constipation, ICU length of stay | I-FABP, DAO, D-lactate, citrulline, tight-junction proteins, microbiome/endotoxemia markers | Sedation, opioids, vasopressors, feeding protocol, abdominal surgery, ileus, fluid balance |
| ARDS/acute lung injury | Ventilator-free days, oxygenation improvement, mortality in adequately powered trials | PEEP/driving pressure trajectories, lung ultrasound or imaging scores, barotrauma, prone-positioning exposure | Inflammatory mediators, epithelial/endothelial injury markers, SPMs, ferroptosis-related markers | Ventilation strategy, PEEP, driving pressure, prone positioning, fluid balance, NMBA, corticosteroids |
| Sepsis-associated encephalopathy | Delirium-free/coma-free days, long-term cognitive outcome | CAM-ICU, RASS-adjusted assessments, EEG, NIRS, TCD, sedation-free neurologic assessment | HMGB1/RAGE/TLR4 markers, BBB injury markers, neurofilament, inflammatory cytokines | Sedation, analgesia, hypoxemia, shock, renal/liver failure, baseline cognition |
| AKI/hepatic injury | KDIGO progression, RRT-free days, AKI-free days; clinically relevant hepatic dysfunction endpoints | Creatinine kinetics, urine output, bilirubin, coagulation parameters, lactate clearance | NGAL, KIM-1, cystatin C, tubular injury markers, hepatocellular injury markers | Shock duration, nephrotoxins, venous congestion, contrast, RRT timing, liver disease |
| Septic cardiac dysfunction | Shock reversal, vasopressor/inotrope-free days, cardiac dysfunction-free days | LVEF/FAC, diastolic indices, RV function, strain, cardiac output, lactate clearance, peripheral perfusion | Troponin, BNP/NT-proBNP, endothelial injury markers, mitochondrial/metabolic stress markers | Norepinephrine, fluids, inotropes, pre-existing heart disease, arrhythmia, mechanical ventilation |
| ICU-acquired weakness/myopathy | Ventilator-free days, mobilization milestones, long-term physical function | MRC sum score, handgrip strength, diaphragm ultrasound, electrophysiology | CK, inflammatory markers, neuromuscular injury markers | Sedation, corticosteroids, NMBA, glycemic control, immobilization, rehabilitation intensity |
| Safety and feasibility | Serious EA-related adverse events, treatment discontinuation, protocol adherence | Bleeding, infection, skin injury, monitoring artifacts, interference with lines/devices | HRV quality, stimulation tolerance, signal quality, adverse-event phenotype | Coagulopathy, thrombocytopenia, ECMO/CRRT, pacemakers/ICDs, edema, staffing and training |
Endpoint hierarchy and trial-design priorities for EA studies in sepsis-associated organ dysfunction.
5 Clinical evidence: signals, uncertainty, and evidence hierarchy
The most comprehensive sepsis-focused meta-analysis included 17 randomized controlled trials and 1,099 patients and reported favorable pooled effects on APACHE II score, inflammatory biomarkers, lactate, intra-abdominal pressure, selected immune indices, and 28-day mortality (40). However, all outcomes were graded as low or very low certainty, and the clinical trial base remains limited by small sample size, usual-care rather than sham comparators, inconsistent blinding, heterogeneous acupoint prescriptions, variable stimulation parameters, and incomplete adverse-event reporting. A key prospective randomized controlled trial in 60 patients with sepsis reported that EA at bilateral ST36 and RN4/CV4 improved APACHE II score and immune indices, including T-cell subsets and monocyte HLA-DR, but did not significantly reduce 28-day mortality (87). In many trials, comparator groups were usual care rather than sham acupuncture, making performance and placebo-related bias difficult to exclude. Mortality reduction should therefore be interpreted as hypothesis-generating rather than clinically established.
Importantly, no current clinical evidence establishes EA as a mortality-reducing intervention in sepsis. Reported mortality signals from meta-analyses should be considered hypothesis-generating because they are derived from small trials with heterogeneous interventions, limited sham control, unclear allocation concealment, inconsistent blinding, and possible publication bias (79, 82). At present, the most defensible clinical interpretation is that EA may produce measurable physiological or biomarker signals, particularly in gastrointestinal dysfunction, but remains unproven for survival, durable organ protection, long-term recovery, or routine ICU implementation (88).
A more systematic appraisal of these clinical studies further limits the strength of inference. Many trials do not clearly report allocation concealment, prespecified primary outcomes, blinded outcome assessment, or complete intention-to-treat analysis. Blinding is particularly challenging because penetrating acupuncture, electrical sensation, practitioner-patient interaction, and acupoint-specific expectations can introduce performance and placebo-related effects. In addition, selective reporting and publication bias cannot be excluded because negative or neutral acupuncture trials may be less likely to be published, and funnel-plot-based assessments are underpowered when the number and size of trials are small. These methodological constraints justify a cautious interpretation of positive clinical signals and support the need for sham-controlled, adequately concealed, prospectively registered, multicenter RCTs with standardized adverse-event reporting (40).
The strongest clinical signal currently concerns sepsis-associated gastrointestinal dysfunction. RCTs of EA at ST36-ST37 reported reductions in inflammatory responses, intra-abdominal pressure, intestinal dysfunction scores, and intestinal barrier-related biomarkers in septic patients (41). A systematic review and meta-analysis focused on septic gastrointestinal dysfunction suggested improvements in intra-abdominal pressure, AGI grade, APACHE II score, abdominal perimeter, and bowel sounds, whereas mortality effects were not robust (42). The EAGISM trial protocol is particularly relevant because it focuses on prevention of acute gastrointestinal injury in mechanically ventilated patients with sepsis and prespecifies ICU-relevant outcomes (43).
This distinction between statistical signal and clinical certainty is central. Existing trials differ in manual acupuncture, EA, transcutaneous electrical acupoint stimulation, acupoint combinations, stimulation frequency, current intensity, session duration, treatment frequency, and timing relative to sepsis onset. These limitations are amplified in sepsis, where source control, antimicrobial timing, fluid balance, vasopressor exposure, nutrition, sedation, mechanical ventilation, renal replacement therapy, and organ-support strategies strongly confound outcomes (). Figure 4 visually separates biological plausibility from translational readiness, showing that gastrointestinal dysfunction and biomarker modulation are more developed than mortality benefit, cardiac endpoints, renal hard endpoints, and routine ICU implementation.
Figure 4
Because organ protection should be tested with organ-specific clinical endpoints rather than generic inflammatory biomarkers alone, Table 3 translates the current evidence gaps into preferred endpoints, major ICU confounders, and trial-design implications for future EA studies in sepsis. Future trials should prespecify a hierarchy of outcomes. Primary outcomes should be clinically meaningful and organ-specific, whereas inflammatory and barrier biomarkers should be treated as mechanistic or secondary endpoints (88). Without this hierarchy, positive biomarker studies are unlikely to change sepsis guidelines ().
6 ICU implementation: clinical boundaries, contraindications, and safety governance
EA-based interventions in sepsis should be framed strictly as adjunctive neuroimmune modulation, not as an alternative to standard sepsis care. Current sepsis guidelines prioritize early antimicrobials, source control, fluid resuscitation, vasopressors, mechanical ventilation, renal replacement therapy, and organ support when indicated (). Therefore, EA should never delay or replace these time-sensitive interventions (); this boundary should be explicitly stated in both clinical protocols and review articles to avoid therapeutic misclassification in a life-threatening infection.
Implementation in the ICU requires predefined safety governance. Although ICU-focused evidence has not shown a clear increase in serious acupuncture-related adverse events, the certainty of safety evidence remains limited because adverse-event definitions and reporting are inconsistent (89). Needle-based interventions should be avoided or deferred in patients with severe thrombocytopenia, disseminated intravascular coagulation, uncontrolled anticoagulation, local skin infection, burns, severe edema at intended acupoints, refractory shock, uncontrolled agitation, or anatomical conflict with extracorporeal membrane oxygenation cannulas, central venous catheters, arterial lines, drains, or surgical wounds (90). WHO safety guidance advises avoiding needling in bleeding and clotting disorders or anticoagulated patients, and a conservative trial threshold of platelet count <50 × 10∧9/L is reasonable for invasive needle EA, although no universally accepted acupuncture-specific platelet cutoff exists (91, 92). In patients with platelet counts between 50 and 100 × 10∧9/L or unstable coagulation trends, superficial needling, transcutaneous electrical acupoint stimulation, or deferral should be considered.
EA may introduce electrical artifacts into electrophysiological monitoring, including ECG-derived and processed EEG-based systems such as bispectral index; therefore, signal quality should be checked during stimulation and stimulation should be stopped if clinically relevant artifacts occur. Skin should be inspected before and after each session for bleeding, erythema, blistering, pressure injury, edema-related maceration, or local infection. Stimulation should be stopped immediately if new arrhythmia, hemodynamic deterioration, clinically relevant monitoring artifact, unexpected pain, bleeding, skin injury, or interference with life-support devices occurs. In addition, because transcutaneous electrical stimulation has been reported to cause electromagnetic interference with pacemakers and implantable cardioverter-defibrillators, EA should be used cautiously in patients with pacemakers, implantable cardioverter-defibrillators, neurostimulators, or other electrically sensitive devices, particularly when current pathways are close to the thorax or implanted leads (93–95). Device compatibility should be assessed before enrollment, including pacemakers, implantable cardioverter-defibrillators, neurostimulators, and other electrically sensitive devices (96). These risks do not preclude research, but they mandate sterile technique, operator credentialing, standardized documentation of needling depth and electrical parameters, and predefined stopping rules (90).
Practical implementation also depends on feasibility, training, and cost-effectiveness across healthcare systems. Although EA and transcutaneous acupoint stimulation may appear less expensive than many ICU technologies, their true cost-effectiveness cannot be assumed without accounting for practitioner time, credentialing, sterile supplies, device maintenance, monitoring burden, infection-control procedures, workflow disruption, and adverse-event prevention. Multicenter trials should therefore define minimum operator training, competency assessment, standardized stimulation documentation, staff responsibilities, and escalation pathways for safety concerns. In resource-limited settings, simplified non-invasive protocols may be more feasible than needle-based EA, whereas high-resource ICUs may prioritize device compatibility, electronic documentation, and integration with continuous monitoring systems.
Future RCTs should be indication-specific and protocolized. For sepsis-associated gastrointestinal dysfunction, the most clinically developed domain, trials should standardize acupoint prescriptions such as ST36/ST37, define acute gastrointestinal injury grade, and prespecify endpoints including intra-abdominal pressure, feeding tolerance, bowel sound recovery, intestinal barrier biomarkers, SOFA/APACHE II trajectories, ICU length of stay, mechanical ventilation duration, 28-day mortality, and adverse events (39). For ARDS, sepsis-associated encephalopathy, AKI, or ICU-acquired weakness, endpoints should align with organ-specific standards rather than relying only on inflammatory biomarkers, including ventilator-free days and oxygenation indices for ARDS, consensus renal-function criteria for AKI, delirium and cognitive outcomes for SAE, and mobility or muscle-strength outcomes for ICU-acquired weakness (, 97, 98). CONSORT-STRICTA reporting should be followed to ensure transparent reporting of acupoint rationale, needling details, stimulation frequency, intensity, pulse width, session duration, practitioner background, co-interventions, and sham or usual-care comparators (89). A biomarker-enriched strategy incorporating SOFA/APACHE II, shock status, vasopressor dose, lactate, monocyte HLA-DR, lymphocyte count, cytokine profile, and autonomic tone may help identify patients with preserved neuroimmune responsiveness (, ).
6.1 Proposed trial framework for sepsis-associated gastrointestinal dysfunction
Because sepsis-associated gastrointestinal dysfunction currently provides the strongest clinical signal and the clearest mechanistic link to autonomic, barrier, mucosal immune, and microbiota-metabolite regulation, it represents the most suitable first target for a phenotype-specific EA trial in sepsis. Table 4 outlines a pragmatic ICU-compatible trial framework that could be adapted to local resources while preserving methodological rigor. EA should be withheld or discontinued if any of these events occur or if urgent ICU procedures or escalating organ support make stimulation unsafe.
Table 4
| Trial component | Proposed design |
|---|---|
| Target population | Adult ICU patients with Sepsis-3-defined sepsis or septic shock and early sepsis-associated gastrointestinal dysfunction, preferably AGI grade I-II or mechanically ventilated patients at high risk of AGI. |
| Key inclusion criteria | ICU admission; expected ICU stay >48–72 h; completed or planned standard source control without delay; hemodynamic status stable enough for non-emergent adjunctive intervention; measurable GI dysfunction or high-risk phenotype. |
| Key exclusion criteria | Uncontrolled source requiring immediate intervention; refractory shock; severe thrombocytopenia or platelet count <50 × 10∧9/L for needle EA; disseminated intravascular coagulation or uncontrolled anticoagulation; local infection, burn, severe edema, or wound at acupoints; uncontrolled arrhythmia; implanted electrical devices at risk; anatomical conflict with lines, drains, or cannulas. |
| Intervention | Standard sepsis care plus protocolized EA, preferably bilateral ST36/ST37, with predefined frequency, intensity, pulse width, session duration, treatment frequency, operator training, sterile procedures, and treatment window. |
| Comparator | Standard care plus sham EA or validated sham transcutaneous acupoint stimulation with matched attention, device appearance, patient contact, and session duration when feasible. |
| Primary endpoint | Change in AGI grade, GI failure-free days, or enteral feeding tolerance within the first 7 days, selected a priori according to trial phase and feasibility. |
| Secondary endpoints | Intra-abdominal pressure, gastric residual volume, bowel sounds, abdominal distension, diarrhea/constipation, SOFA/APACHE II trajectories, ventilator-free days, ICU length of stay, renal replacement therapy-free days, and 28-day mortality as exploratory outcome unless adequately powered. |
| Mechanistic endpoints | I-FABP, diamine oxidase, D-lactate, citrulline, tight-junction markers, inflammatory cytokines, monocyte HLA-DR, lymphocyte count, microbiome/endotoxemia markers, and autonomic indices when signal quality is adequate. |
| Biomarker stratification | Baseline inflammatory phenotype, lymphocyte count, monocyte HLA-DR, cytokine profile, lactate kinetics, shock status, vasopressor dose, AGI grade, and HRV quality if available. |
| Safety monitoring and stopping rules | Predefined monitoring for bleeding, skin injury, local infection, arrhythmia, hemodynamic deterioration, monitoring artifacts, device interference, severe agitation, urgent procedures, line/drain conflict, or escalating vasopressor requirement. |
| Reporting standards | CONSORT, STRICTA, prespecified adverse-event definitions, protocol adherence reporting, and intention-to-treat analysis. |
Proposed trial framework for EA in sepsis-associated gastrointestinal dysfunction.
7 Closed-loop neuromodulation and artificial intelligence: a future perspective
This section is intentionally framed as a future perspective. At present, to our knowledge, there is no direct clinical evidence that AI-guided, adaptive, or closed-loop EA improves outcomes in septic ICU patients. These concepts are discussed only to identify engineering requirements, signal-validation needs, safety constraints, and physician-supervised implementation pathways for future research, not to imply current clinical readiness (99).
A realistic near-term role for AI is decision support rather than autonomous stimulation. AI models could potentially integrate hemodynamic trajectories, vasopressor exposure, ventilator data, lactate kinetics, laboratory trends, infection source, microbiology, bedside ultrasound, and electronic health-record variables to support phenotype recognition, eligibility screening, safety monitoring, and protocol adherence. However, these models would require prospective validation and transparent reporting before being incorporated into EA trials (100–102). Key technical barriers remain unresolved. HRV is not a reliable standalone control signal in many ICU patients because arrhythmias, cardiac pacing, sedation, vasoactive drugs, mechanical ventilation, and signal-processing heterogeneity can distort interpretation (). Continuous cytokine or sepsis-biomarker biosensing also remains developmental and faces challenges related to sensitivity, calibration drift, biofouling, multiplexing, and regulatory validation (103). Therefore, adaptive dose titration based on real-time immune signals should be treated as speculative.
Any future AI-guided neuromodulation platform should remain physician-in-the-loop. Clinicians must be able to review model inputs, understand the rationale for suggested stimulation changes, override the system, and activate predefined stopping rules. Figure 5 should therefore be read as a roadmap for future research linking wearable interfaces, biosensors, autonomic monitoring, and explainable algorithms, not as evidence that automated EA titration is ready for routine sepsis management.
Figure 5
8 Conclusions
EA provides a biologically plausible framework for adjunctive neuroimmune modulation in sepsis by engaging local sensory afferents, somato-autonomic reflexes, vagal-adrenal and cholinergic anti-inflammatory pathways, and downstream immune-cell regulation. The most consistent clinical signal is observed in sepsis-associated gastrointestinal dysfunction, whereas evidence for lung, brain, kidney, liver, cardiac, and neuromuscular complications remains predominantly preclinical or exploratory.
The next phase of research should move beyond broad claims of organ protection toward phenotype-specific, sham-controlled, multicenter trials with standardized stimulation protocols, organ-specific endpoints, biomarker-informed patient selection, and rigorous ICU safety monitoring. EA should be positioned as a biologically plausible but clinically unproven adjunct to guideline-based sepsis care, not as an established treatment for sepsis-induced multiorgan dysfunction. The innovation of future work will depend less on adding another anti-inflammatory pathway and more on making neuroimmune modulation measurable, safe, reproducible, and clinically interpretable in real ICU environments.
Statements
Author contributions
YQ: Writing – review & editing, Conceptualization, Validation, Formal analysis, Visualization, Writing – original draft, Data curation. CL: Writing – original draft, Data curation, Formal analysis, Writing – review & editing, Resources. HL: Investigation, Writing – review & editing, Visualization, Software. Investigation, Visualization, Formal analysis, Writing – original draft. YL: Writing – original draft, Resources, Formal analysis, Investigation. KX: Writing – original draft, Resources, Formal analysis, Investigation. YS: Data curation, Writing – review & editing, Supervision, Formal analysis, Writing – original draft, Investigation, Resources, Funding acquisition. LX: Funding acquisition, Visualization, Formal analysis, Writing – review & editing, Investigation, Writing – original draft, Software.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This work was supported by the Zhejiang Provincial Natural Science Foundation of China (Grants No. LTGY24H150002), Medical Health Science and Technology Project of Zhejiang Provincial (2024KY479, 2025KY1673 and 2025HY1301), and Basic Public Welfare Project of Shaoxing City (2025A14026).
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was used in the creation of this manuscript. ChatGPT by OpenAI, including its image-generation functionality, was used to assist in creating and refining conceptual schematic figures for Figures 1–5. The AI-generated outputs were used only as illustrative visual aids. No original research data, patient images, clinical images, or quantitative results were generated or altered using AI. All figure labels, scientific content, and final layouts were checked, edited, and approved by the author(s), who take full responsibility for the final manuscript and figures.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed.2026.1891245/full#supplementary-material
Supplementary Table S1Provides additional details on the targeted search strategy and evidence-selection principles used for this critical translational review.
Abbreviations
AI, artificial intelligence; AKI, acute kidney injury; APACHE II, Acute Physiology and Chronic Health Evaluation II; ARDS, acute respiratory distress syndrome; CLP, cecal ligation and puncture; CT, computed tomography; EA, electroacupuncture; ECG, electrocardiography; EEG, electroencephalography; HLA-DR, human leukocyte antigen-DR; HMGB1, high-mobility group box 1; HO-1, heme oxygenase-1; HRV, heart-rate variability; ICD, implantable cardioverter-defibrillator; ICU, intensive care unit; I-FABP, intestinal fatty acid-binding protein; IAP, intra-abdominal pressure; LPS, lipopolysaccharide; NF-κB, nuclear factor-κB; Nrf2, nuclear factor erythroid 2-related factor 2; PC6, Neiguan; PROKR2, prokineticin receptor 2; RCT, randomized controlled trial; RN4/CV4, Guanyuan; SAE, sepsis-associated encephalopathy; SOFA, Sequential Organ Failure Assessment; SPMs, specialized pro-resolving mediators; ST36, Zusanli; STAT3, signal transducer and activator of transcription 3; TEAS, transcutaneous electrical acupoint stimulation; TLR4, Toll-like receptor 4; TNF-α, tumor necrosis factor-α; α7nAChR, α7 nicotinic acetylcholine receptor.
References
1.
SingerMDeutschmanCSSeymourCWShankar-HariMAnnaneDBauerMet al. The third international consensus definitions for sepsis and septic shock (Sepsis-3). JAMA. (2016) 315:801–10. doi: 10.1001/jama.2016.0287
2.
RuddKEJohnsonSCAgesaKMShackelfordKATsoiDKievlanDRet al. Global, regional, and national sepsis incidence and mortality, 1990–2017: analysis for the global burden of disease study. Lancet. (2020) 395:200–11. doi: 10.1016/S0140-6736(19)32989-7
3.
EvansLRhodesAAlhazzaniWAntonelliMCoopersmithCMFrenchCet al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med. (2021) 47:1181–247. doi: 10.1007/s00134-021-06506-y
4.
HotchkissRSMonneretGPayenD. Sepsis-induced immunosuppression: from cellular dysfunctions to immunotherapy. Nat Rev Immunol. (2013) 13:862–74. doi: 10.1038/nri3552
5.
van der PollTvan de VeerdonkFLSciclunaBPNeteaMG. The immunopathology of sepsis and potential therapeutic targets. Nat Rev Immunol. (2017) 17:407–20. doi: 10.1038/nri.2017.36
6.
PandaAKBasuB. Regenerative bioelectronics: a strategic roadmap for precision medicine. Biomaterials. (2023) 301:122271. doi: 10.1016/j.biomaterials.2023.122271
7.
TraceyKJ. The inflammatory reflex. Nature. (2002) 420:853–9. doi: 10.1038/nature01321
8.
Rosas-BallinaMOlofssonPSOchaniMValdés-FerrerSILevineYAReardonCet al. Acetylcholine-synthesizing T cells relay neural signals in a vagus nerve circuit. Science. (2011) 334:98–101. doi: 10.1126/science.1209985
9.
LiuSWangZSuYQiLYangWFuMet al. A neuroanatomical basis for electroacupuncture to drive the vagal-adrenal axis. Nature. (2021) 598:641–5. doi: 10.1038/s41586-021-04001-4
10.
Torres-RosasRYehiaGPeñaGMishraPdel Rocio Thompson-BonillaMMoreno-EutimioMAet al. Dopamine mediates vagal modulation of the immune system by electroacupuncture. Nat Med. (2014) 20:291–5. doi: 10.1038/nm.3479
11.
YangLZhouDCaoJShiFZengJZhangSet al. Revealing the biological mechanism of acupuncture in alleviating excessive inflammatory responses and organ damage in sepsis: a systematic review. Front Immunol. (2023) 14:1242640. doi: 10.3389/fimmu.2023.1242640
12.
ZengLYanJ. Acupuncture at Zusanli (ST36) for the benefits of sepsis–laboratory evidence for “preventive treatment of disease” and “holistic concept” in the protective effects of acupuncture: a literature review of rodent studies. Front Neurol. (2025) 16:1645906. doi: 10.3389/fneur.2025.1645906
13.
LaiFRenYLaiCChenRYinXTanCet al. Acupuncture at Zusanli (ST36) for experimental sepsis: a systematic review. Evid Based Complement Alternat Med. (2020) 2020:3620741. doi: 10.1155/2020/3620741
14.
XiaoYXuLJiangYWangQDengJLuoZet al. Immunomodulatory therapy for sepsis: from pathophysiological mechanisms to precision treatment. Front Cell Infect Microbiol. (2025) 15:1746254. doi: 10.3389/fcimb.2025.1746254
15.
Ruiz-RodriguezJCPlata-MenchacaEPChiscano-CamónLRuiz-SanmartinAPérez-CarrascoMPalmadaCet al. Precision medicine in sepsis and septic shock: from omics to clinical tools. World J Crit Care Med. (2022) 11:1–21. doi: 10.5492/wjccm.v11.i1.1
16.
TorresLKPickkersPvan der PollT. Sepsis-induced immunosuppression. Annu Rev Physiol. (2022) 84:157–81. doi: 10.1146/annurev-physiol-061121-040214
17.
GoldmanNChenMFujitaTXuQPengWLiuWet al. Adenosine A1 receptors mediate local anti-nociceptive effects of acupuncture. Nat Neurosci. (2010) 13:883–8. doi: 10.1038/nn.2562
18.
BorovikovaLVIvanovaSZhangMYangHBotchkinaGIWatkinsLRet al. Vagus nerve stimulation attenuates the systemic inflammatory response to endotoxin. Nature. (2000) 405:458–62. doi: 10.1038/35013070
19.
ChavanSSPavlovVATraceyKJ. Mechanisms and therapeutic relevance of neuro-immune communication. Immunity. (2017) 46:927–42. doi: 10.1016/j.immuni.2017.06.008
20.
LimHDKimMHLeeCYNamgungU. Anti-inflammatory effects of acupuncture stimulation via the vagus nerve. PLoS ONE. (2016) 11:e0151882. doi: 10.1371/journal.pone.0151882
21.
Báez-PagánCADelgado-VélezMLasalde-DominicciJA. Activation of the macrophage α7 nicotinic acetylcholine receptor and control of inflammation. J Neuroimmune Pharmacol. (2015) 10:468–76. doi: 10.1007/s11481-015-9601-5
22.
WallEAZavzavadjianJRChangMSRandhawaBZhuXHsuehRCet al. Suppression of LPS-induced TNF-alpha production in macrophages by cAMP is mediated by PKA-AKAP95-p105. Sci Signal. (2009) 2:ra28. doi: 10.1126/scisignal.2000202
23.
ZhangZShiYCaiDJinSZhuCShenYet al. Effect of electroacupuncture at ST36 on the intestinal mucosal mechanical barrier and expression of occludin in a rat model of sepsis. Acupunct Med. (2018) 36:333–8. doi: 10.1136/acupmed-2016-011187
24.
WuJLyuBGanTWangLZhuM. Electroacupuncture improves acute bowel injury recovery in rat models. Exp Ther Med. (2017) 14:4655–62. doi: 10.3892/etm.2017.5159
25.
LvZYShiYLBassiGSChenYJYinLMWangYet al. Electroacupuncture at ST36 (Zusanli) prevents T-Cell lymphopenia and improves survival in septic mice. J Inflamm Res. (2022) 15:2819–33. doi: 10.2147/JIR.S361466
26.
XuXHuangXXiaoLWangJYangXWuY. Mechanism of electro-acupuncture in alleviating intestinal injury in septic mice via polyamine-related M2-macrophage polarization. Front Immunol. (2024) 15:1373876. doi: 10.3389/fimmu.2024.1373876
27.
ZhangYMengZWuLLiuXGuoCYuJet al. Protective effect of electroacupuncture on the barrier function of intestinal injury in endotoxemia through HO-1/PINK1 pathway-mediated mitochondrial dynamics regulation. Oxid Med Cell Longev. (2023) 2023:1464853. doi: 10.1155/2023/1464853
28.
AydinPBekmezH. Experimental sepsis models: advantages and limitations Eurasian. J Med. (2023) 55:120–4. doi: 10.5152/eurasianjmed.2023.23364
29.
LiuSWangZFSuYSRayRSJingXHWangYQet al. Somatotopic organization and intensity dependence in driving distinct NPY-expressing sympathetic pathways by electroacupuncture. Neuron. (2020) 108:436–50.e437. doi: 10.1016/j.neuron.2020.07.015
30.
FanAY. Anti-inflammatory mechanism of electroacupuncture involves the modulation of multiple systems, levels and targets and is not limited to “driving the vagus-adrenal axis”. J Integr Med. (2023) 21:320–3. doi: 10.1016/j.joim.2023.06.001
31.
DevlinJWSkrobikYGélinasCNeedhamDMSlooterAJCPandharipandePPet al. Clinical practice guidelines for the prevention and management of pain, agitation/sedation, delirium, immobility, and sleep disruption in adult patients in the ICU. Crit Care Med. (2018) 46:e825–73.
32.
KarmaliSNSciuscoAMaySMAcklandGL. Heart rate variability in critical care medicine: a systematic review. Intensive Care Med Exp. (2017) 5:33. doi: 10.1186/s40635-017-0146-1
33.
BachmannKFCotoiaAReintam BlaserA. Gastrointestinal function and nutritional interventions in septic shock. Curr Opin Crit Care. (2025) 31:599–607. doi: 10.1097/MCC.0000000000001302
34.
LiuLDuanJHeWWaydhasC. Gastrointestinal complications in intensive care. Med Rev. (2021) 5:452–76. doi: 10.1515/mr-2025-0039
35.
ZhangDLiYDingLFuYDongXLiH. Prevalence and outcome of acute gastrointestinal injury in critically ill patients: a systematic review and meta-analysis. Medicine. (2018) 97:e12970. doi: 10.1097/MD.0000000000012970
36.
ZhuMFXingXLeiSWuJNWangLCHuangLQet al. Electroacupuncture at bilateral zusanli points (ST36) protects intestinal mucosal immune barrier in sepsis. Evid Based Complement Alternat Med. (2015) 2015:639412. doi: 10.1155/2015/639412
37.
XieDPZhouGBChenRLQinXLDuJDZhangYet al. Effect of electroacupuncture at zusanli (ST36) on sepsis induced by cecal ligation puncture and its relevance to spleen. Evid Based Complement Alternat Med. (2020) 2020:1914031. doi: 10.1155/2020/1914031
38.
TatsumiH. Enteral tolerance in critically ill patients. J Intensive Care. (2019) 7:30. doi: 10.1186/s40560-019-0378-0
39.
Reintam BlaserAMalbrainMLStarkopfJFruhwaldSJakobSMDe WaeleJet al. Gastrointestinal function in intensive care patients: terminology, definitions and management. Recommendations of the ESICM working group on abdominal problems. Intensive Care Med. (2012) 38:384–94. doi: 10.1007/s00134-011-2459-y
40.
XianJWangLZhangCWangJZhuYYuHet al. Efficacy and safety of acupuncture as a complementary therapy for sepsis: a systematic review and meta-analysis. Acupunct Med. (2023) 41:3–15. doi: 10.1177/09645284221086288
41.
MengJBJiaoYNXuXJLaiZZZhangGJiCLet al. Electro-acupuncture attenuates inflammatory responses and intraabdominal pressure in septic patients: a randomized controlled trial. Medicine. (2018) 97:e0555. doi: 10.1097/MD.0000000000010555
42.
XuJLiJDengDWuJZhengBLiJ. Efficacy and safety of acupuncture for septic gastrointestinal dysfunction: a systematic review and meta-analysis of randomized controlled trials. Front Med. (2026) 13:1680999. doi: 10.3389/fmed.2026.1680999
43.
XuJXieTTWuJXDengDWHuangJYuYet al. Electroacupuncture for the prevention of acute gastrointestinal injury in patients with sepsis undergoing mechanical ventilation (EAGISM): a protocol for an open-label randomized controlled trial. Front Med. (2026) 13:1695929. doi: 10.3389/fmed.2026.1695929
44.
HuangCLHuangCJTsaiPSYanLPXuHZ. Acupuncture stimulation of ST-36 (Zusanli) significantly mitigates acute lung injury in lipopolysaccharide-stimulated rats. Acta Anaesthesiol Scand. (2006) 50:722–30. doi: 10.1111/j.1399-6576.2006.01029.x
45.
YuJBShiJGongLRDongSAXuYZhangYet al. Role of Nrf2/ARE pathway in protective effect of electroacupuncture against endotoxic shock-induced acute lung injury in rabbits. PLoS ONE. (2014) 9:e104924. doi: 10.1371/journal.pone.0104924
46.
XieCWuSLiZHuangBZengW. [Electroacupuncture protects septic rats from acute lung injury through the JAK1/STAT3 pathway]. Nan Fang Yi Ke Da Xue Xue Bao. (2020) 40:1662–7. doi: 10.12122/j.issn.1673-4254.2020.11.20
47.
LuoDLiuLZhangHMZhouYDZhouMFLiJXet al. Electroacupuncture pretreatment exhibits lung protective and anti-inflammation effects in lipopolysaccharide-induced acute lung injury via SIRT1-dependent pathways. Evid Based Complement Alternat Med. (2022) 2022:2252218. doi: 10.1155/2022/2252218
48.
ZhouJLiLQuMTanJSunGLuoFet al. Electroacupuncture pretreatment protects septic rats from acute lung injury by relieving inflammation and regulating macrophage polarization. Acupunct Med. (2023) 41:175–82. doi: 10.1177/09645284221118588
49.
ZhangYZhengLDengHFengDHuSZhuLet al. Electroacupuncture alleviates LPS-induced ARDS through α7 nicotinic acetylcholine receptor-mediated inhibition of ferroptosis. Front Immunol. (2022) 13:832432. doi: 10.3389/fimmu.2022.832432
50.
HuangYDongSZhangYZhangYGuoYShiJet al. Electroacupuncture promotes resolution of inflammation by modulating SPMs via vagus nerve activation in LPS-induced ALI. Int Immunopharmacol. (2025) 147:113941. doi: 10.1016/j.intimp.2024.113941
51.
AmatoMBMeadeMOSlutskyASBrochardLCostaELSchoenfeldDAet al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med. (2015) 372:747–55. doi: 10.1056/NEJMsa1410639
52.
QadirNSahetyaSMunshiLSummersCAbramsDBeitlerJet al. An update on management of adult patients with acute respiratory distress syndrome: an official american thoracic society clinical practice guideline. Am J Respir Crit Care Med. (2024) 209:24–36. doi: 10.1164/rccm.202311-2011ST
53.
BosLDJArtigasAConstantinJMHagensLAHeijnenNLaffeyJGet al. Precision medicine in acute respiratory distress syndrome: workshop report and recommendations for future research. Eur Respir Rev. (2021) 30:200317. doi: 10.1183/16000617.0317-2020
54.
ClinicalTrials.gov. Electroacupuncture Modulates SPMs Metabolism and Respiratory Symptoms in Patients With Sepsis Complicating ARDS NCT06006325. Available online at: https://www.trialx.com/clinical-trials/listings/271720/electroacupuncture-modulates-spms-metabolism (Accessed July 4, 2026).
55.
ChaudhryNDuggalAK. Sepsis associated encephalopathy. Adv Med. (2014) 2014:762320. doi: 10.1155/2014/762320
56.
ChenYLeiYMoLQLiJWangMHWeiJCet al. Electroacupuncture pretreatment with different waveforms prevents brain injury in rats subjected to cecal ligation and puncture via inhibiting microglial activation, and attenuating inflammation, oxidative stress and apoptosis. Brain Res Bull. (2016) 127:248–59. doi: 10.1016/j.brainresbull.2016.10.009
57.
HanYGQinXZhangTLeiMSunFYSunJJet al. Electroacupuncture prevents cognitive impairment induced by lipopolysaccharide via inhibition of oxidative stress and neuroinflammation. Neurosci Lett. (2018) 683:190–5. doi: 10.1016/j.neulet.2018.06.003
58.
MoYWangLRenMXieWYeXZhouBet al. Electroacupuncture prevents LPS- induced neuroinflammation via upregulation of PICK-TLR4 complexes in the microglia of hippocampus. Brain Res Bull. (2021) 177:295–304. doi: 10.1016/j.brainresbull.2021.10.010
59.
XinYWangJChuTZhouYLiuCXuA. Electroacupuncture alleviates neuroinflammation by inhibiting the HMGB1 signaling pathway in rats with sepsis-associated encephalopathy. Brain Sci. (2022) 12:1732. doi: 10.3390/brainsci12121732
60.
LiYLiZHeFQinCFanRZhangFet al. Electroacupuncture alleviates cognitive dysfunction and neuronal pyroptosis in septic mice. Acupunct Med. (2023) 41:246–56. doi: 10.1177/09645284221117847
61.
WoodMDBoydJGWoodNFrankJGirardTDRoss-WhiteAet al. The use of near-infrared spectroscopy and/or transcranial doppler as non-invasive markers of cerebral perfusion in adult sepsis patients with delirium: a systematic review. J Intensive Care Med. (2022) 37:408–22. doi: 10.1177/0885066621997090
62.
PierrakosCAttouRDecorteLKolyvirasAMalinverniSGottigniesPet al. Transcranial doppler to assess sepsis-associated encephalopathy in critically ill patients. BMC Anesthesiol. (2014) 14:45. doi: 10.1186/1471-2253-14-45
63.
PandharipandePPGirardTDJacksonJCMorandiAThompsonJLPunBTet al. Long-term cognitive impairment after critical illness. N Engl J Med. (2013) 369:1306–16. doi: 10.1056/NEJMoa1301372
64.
SonnevilleRBenghanemSJeantinLde MontmollinEDomanMGaudemerAet al. The spectrum of sepsis-associated encephalopathy: a clinical perspective. Crit Care. (2023) 27:386. doi: 10.1186/s13054-023-04655-8
65.
YuJBShiJZhangYGongLRDongSACaoXSet al. Electroacupuncture ameliorates acute renal injury in lipopolysaccharide-stimulated rabbits via induction of HO-1 through the PI3K/Akt/Nrf2 Pathways. PLoS ONE. (2015) 10:e0141622. doi: 10.1371/journal.pone.0141622
66.
GuGZhangZWangGHanFHanLWangKet al. Effects of electroacupuncture pretreatment on inflammatory response and acute kidney injury in endotoxaemic rats. J Int Med Res. (2011) 39:1783–97. doi: 10.1177/147323001103900521
67.
HarpinDSimadibrataCLMihardjaHBarasilaAC. Effect of electroacupuncture on urea and creatinine levels in the wistar sepsis model. Med Acupunct. (2020) 32:29–37. doi: 10.1089/acu.2019.1369
68.
LiuHWLiuMCTsaoCMLiaoMHWuCC. Electro-acupuncture at ‘Neiguan' (PC6) attenuates liver injury in endotoxaemic rats. Acupunct Med. (2011) 29:284–8. doi: 10.1136/aim.2010.003525
69.
PeerapornratanaSManrique-CaballeroCLGómezHKellumJA. Acute kidney injury from sepsis: current concepts, epidemiology, pathophysiology, prevention and treatment. Kidney Int. (2019) 96:1083–99. doi: 10.1016/j.kint.2019.05.026
70.
ZarbockANadimMKPickkersPGomezHBellSJoannidisMet al. Sepsis-associated acute kidney injury: consensus report of the 28th acute disease quality initiative workgroup. Nat Rev Nephrol. (2023) 19:401–17. doi: 10.1038/s41581-023-00683-3
71.
LeafDEWaikarSS. End points for clinical trials in acute kidney injury. Am J Kidney Dis. (2017) 69:108–16. doi: 10.1053/j.ajkd.2016.05.033
72.
WoznicaEAInglotMWoznicaRKŁysenkoL. Liver dysfunction in sepsis. Adv Clin Exp Med. (2018) 27:547–51. doi: 10.17219/acem/68363
73.
Nordon-CraftAMossMQuanDSchenkmanM. Intensive care unit-acquired weakness: implications for physical therapist management. Phys Ther. (2012) 92:1494–506. doi: 10.2522/ptj.20110117
74.
AissaouiNBoissierFChewMSingerMVignonP. Sepsis-induced cardiomyopathy. Eur Heart J. (2025) 46:3339–53. doi: 10.1093/eurheartj/ehaf340
75.
LiXWangLYingXZhengYTanQYuXet al. Electroacupuncture pre-treatment alleviates sepsis-induced cardiac inflammation and dysfunction by inhibiting the calpain-2/STAT3 pathway. Front Physiol. (2022) 13:961909. doi: 10.3389/fphys.2022.961909
76.
WuZXiaYWangCLuWZuoHWuDet al. Electroacupuncture at neiguan (PC6) attenuates cardiac dysfunction caused by cecal ligation and puncture via the vagus nerve. Biomed Pharmacother. (2023) 162:114600. doi: 10.1016/j.biopha.2023.114600
77.
JentzerJCBergDDChondeMDDahiyaGElliottARampersadPet al. Mixed cardiogenic-vasodilatory shock: current insights and future directions. JACC Adv. (2025) 4:101432. doi: 10.1016/j.jacadv.2024.101432
78.
BeesleySJWeberGSargeTNikravanSGrissomCKLanspaMJet al. Septic cardiomyopathy. Crit Care Med. (2018) 46:625–34. doi: 10.1097/CCM.0000000000002851
79.
InceC. Hemodynamic coherence and the rationale for monitoring the microcirculation. Crit Care. (2015) 19:S8. doi: 10.1186/cc14726
80.
WangHDingHWangZYZhangK. Research progress on microcirculatory disorders in septic shock: a narrative review. Medicine. (2024) 103:e37273. doi: 10.1097/MD.0000000000037273
81.
HernándezGOspina-TascónGADamianiLPEstenssoroEDubinAHurtadoJet al. Effect of a resuscitation strategy targeting peripheral perfusion status vs serum lactate levels on 28-day mortality among patients with septic shock: the ANDROMEDA-SHOCK randomized clinical trial. JAMA. (2019) 321:654–64. doi: 10.1001/jama.2019.0071
82.
MasseyMJHouPCFilbinMWangHNgoLHuangDTet al. Microcirculatory perfusion disturbances in septic shock: results from the ProCESS trial. Crit Care. (2018) 22:308. doi: 10.1186/s13054-018-2240-5
83.
StanzaniGDuchenMRSingerM. The role of mitochondria in sepsis-induced cardiomyopathy. Biochim Biophys Acta Mol Basis Dis. (2019) 1865:759–73. doi: 10.1016/j.bbadis.2018.10.011
84.
SanfilippoFCorredorCArcadipaneALandesbergGVieillard-BaronACecconiMet al. Tissue doppler assessment of diastolic function and relationship with mortality in critically ill septic patients: a systematic review and meta-analysis. Br J Anaesth. (2017) 119:583–94. doi: 10.1093/bja/aex254
85.
HermansGClerckxBVanhullebuschTSegersJVanpeeGRobbeetsCet al. Interobserver agreement of medical research council sum-score and handgrip strength in the intensive care unit. Muscle Nerve. (2012) 45:18–25. doi: 10.1002/mus.22219
86.
DresMGoligherECHeunksLMABrochardLJ. Critical illness-associated diaphragm weakness. Intensive Care Med. (2017) 43:1441–52. doi: 10.1007/s00134-017-4928-4
87.
YangGHuRYDengAJHuangYLiJ. Effects of electro-acupuncture at zusanli, guanyuan for sepsis patients and its mechanism through immune regulation. Chin J Integr Med. (2016) 22:219–24. doi: 10.1007/s11655-016-2462-9
88.
ChenWTSunCZhouYBLiuDHPengZLChenJet al. Evaluation on the effect of acupuncture on patients with sepsis-induced myopathy (ACU-SIM pilot study): a single center, propensity-score stratified, assessor-blinded, prospective pragmatic controlled trial. Medicine. (2020) 99:e20233. doi: 10.1097/MD.0000000000020233
89.
MacPhersonHAltmanDGHammerschlagRYoupingLTaixiangWWhiteAet al. Revised standards for reporting interventions in clinical trials of acupuncture (STRICTA): extending the CONSORT statement. PLoS Med. (2010) 7:e1000261. doi: 10.1371/journal.pmed.1000261
90.
ZhangJShangHGaoXErnstE. Acupuncture-related adverse events: a systematic review of the Chinese literature. Bull World Health Organ. (2010) 88:915–921c. doi: 10.2471/BLT.10.076737
91.
CybularzPABrothersKSinghGMFeingoldJLLewisMENiesleyML. The safety of acupuncture in patients with cancer therapy-related thrombocytopenia. Med Acupunct. (2015) 27:224–9. doi: 10.1089/acu.2015.1099
92.
BiolatoMVitaleFGalassoTGasbarriniAGriecoA. Minimum platelet count threshold before invasive procedures in cirrhosis: evolution of the guidelines. World J Gastrointest Surg. (2023) 15:127–41. doi: 10.4240/wjgs.v15.i2.127
93.
VasilakosDGFyntanidouBP. Electroacupuncture on a patient with pacemaker: a case report. Acupunct Med. (2011) 29:152–3. doi: 10.1136/aim.2010.003863
94.
BadgerJTaylorPSwainI. The safety of electrical stimulation in patients with pacemakers and implantable cardioverter defibrillators: a systematic review. J Rehabil Assist Technol Eng. (2017) 4:2055668317745498. doi: 10.1177/2055668317745498
95.
DoyalASDoyalHEFlynnDNSchoenherrJKwonJJKumarP. Electrocardiographic interference on bispectral index monitor: a case of crossed wires. Cureus. (2022) 14:e29511. doi: 10.7759/cureus.29511
96.
LiPWuYWuYGeDZhangQLiYet al. Acupuncture for ICU patients: evidence, mechanisms, and implementation challenges. Front Neurol. (2025) 16:1711600. doi: 10.3389/fneur.2025.1711600
97.
GrasselliGCalfeeCSCamporotaLPooleDAmatoMBPAntonelliMet al. ESICM guidelines on acute respiratory distress syndrome: definition, phenotyping and respiratory support strategies. Intensive Care Med. (2023) 49:727–59. doi: 10.1007/s00134-023-07050-7
98.
VanhorebeekILatronicoNVan den BergheG. ICU-acquired weakness. Intensive Care Med. (2020) 46:637–53. doi: 10.1007/s00134-020-05944-4
99.
LermanIBuYSinghRSilvermanHABhardwajAMannAJet al. Next generation bioelectronic medicine: making the case for non-invasive closed-loop autonomic neuromodulation. Bioelectron Med. (2025) 11:1. doi: 10.1186/s42234-024-00163-4
100.
VaseyBNagendranMCampbellBCliftonDACollinsGSDenaxasSet al. Reporting guideline for the early stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI. BMJ. (2022) 377:e070904. doi: 10.1136/bmj-2022-070904
101.
CollinsGSMoonsKGMDhimanPRileyRDBeamALVan CalsterBet al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. (2024) 385:e078378. doi: 10.1136/bmj-2023-078378
102.
LiuXCruz RiveraSMoherDCalvertMJDennistonAKChanA-Wet al. Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. Nat Med. (2020) 26:1364–74. doi: 10.1136/bmj.m3164
103.
LiuLHanZAnFGongXZhaoCZhengWet al. Aptamer-based biosensors for the diagnosis of sepsis. J Nanobiotechnology. (2021) 19:216. doi: 10.1186/s12951-021-00959-5
Summary
Keywords
electroacupuncture, immunophenotyping, inflammatory reflex, neuroimmune modulation, sepsis
Citation
Qian Y, Lin C, Lin H, Leng Y, Xie K, Shan Y and Xu L (2026) Electroacupuncture-inspired neuroimmune modulation in sepsis: evidence appraisal and ICU trial-design priorities. Front. Med. 13:1891245. doi: 10.3389/fmed.2026.1891245
Received
26 May 2026
Revised
04 July 2026
Accepted
09 July 2026
Published
30 July 2026
Volume
13 - 2026
Edited by
Monirah Abdulrahman Albabtain, Prince Sultan Cardiac Center Hospital, Saudi Arabia
Reviewed by
Guangdong Wang, First Affiliated Hospital of Xi'an Jiaotong University, China
Tomasz Reysner, Poznan University of Medical Sciences, Poland
Updates
Copyright
© 2026 Qian, Lin, Lin, Leng, Xie, Shan and Xu.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Lili Xu, 40455573@qq.com; Yue Shan, 3002024141@usx.edu.cn
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.