Auricular vagus nerve stimulation for mitigation of inflammation and vasospasm in subarachnoid hemorrhage: a single-institution randomized controlled trial

In Brief

This randomized controlled trial evaluated the effects of noninvasive transauricular vagus nerve stimulation (taVNS) on inflammation and vasospasm following subarachnoid hemorrhage. taVNS significantly reduced key inflammatory markers (TNF-α, IL-6) in plasma and CSF, decreased radiographic vasospasm, and improved clinical outcomes compared to sham treatment. As the first study to demonstrate that taVNS can mitigate the inflammatory response and associated sequelae after subarachnoid hemorrhage, it introduces a novel, nonpharmacological approach that could impact future management of this condition.

Abstract

OBJECTIVE

Inflammation contributes to morbidity following subarachnoid hemorrhage (SAH). The authors of this study evaluate how applying noninvasive transauricular vagus nerve stimulation (taVNS) can target this deleterious inflammatory response following SAH and reduce the rate of radiographic vasospasm.

METHODS

In this prospective, triple-blinded, randomized controlled trial, 27 patients were randomized to taVNS or sham stimulation. Serial blood and CSF samples were collected every 3 days to quantify inflammatory markers. Radiographic cerebral vasospasm severity and functional outcomes (modified Rankin Scale scores) were analyzed.

RESULTS

No adverse events occurred. Radiographic vasospasm was significantly reduced (p = 0.018), with serial vessel caliber measurements demonstrating a more rapid return to normal than in the sham-treated group (p < 0.001). In the taVNS group, tumor necrosis factor–α was significantly reduced in both plasma (days 7 and 10) and CSF (day 13); interleukin-6 was also significantly reduced in plasma (day 4) and CSF (day 13) (p < 0.05). Patients receiving taVNS had higher rates of favorable outcomes at discharge (38.4% vs 21.4%) and first follow-up (76.9% vs 57.1%). Patients treated with taVNS had significant improvement in modified Rankin Scale scores from admission to first follow-up (p = 0.014), unlike patients in the sham-treated group (p = 0.18). The taVNS group had a significantly lower rate of discharge to a skilled nursing facility or hospice (p = 0.04).

CONCLUSIONS

taVNS is a noninvasive method of neuro- and systemic immunomodulation. This trial supports the finding that taVNS following SAH can mitigate the inflammatory response, reduce radiographic vasospasm, and potentially improve functional and neurological outcomes.

Subarachnoid hemorrhage (SAH) resulting from a ruptured aneurysm accounts for 7% of all strokes worldwide, with 40% of patients suffering permanent disability.¹ Secondary injury is a major driver of morbidity following SAH, as mediated by early brain injury, cerebral vasospasm, delayed cortical ischemia, and chronic hydrocephalus.²

Inflammation is thought to be a key factor driving the morbidity associated with SAH. Following SAH, blood within the subarachnoid space triggers local and systemic inflammatory responses, with increases in the inflammatory markers IL-6^3–5 and TNF-α⁶ both systemically and centrally. Notably, these increased inflammatory markers are correlated with adverse sequelae from SAH, including the risk for vasospasm, cerebral edema, hydrocephalus, and poor overall patient outcome.^3,6

Despite increasing evidence for the role of inflammation following SAH, an effective method to modulate the deleterious inflammatory response in patients after SAH is lacking. Pharmacological approaches that target antiinflammatory pathways have been unsuccessful in clinical trials.^7,8 Vagus nerve stimulation (VNS) provides a novel, nonpharmacological approach to systemic immunomodulation. VNS reduces inflammation via a cholinergic antiinflammatory pathway, whereby efferents from the vagus nerve to the spleen and other organs leads to the release of acetylcholine, with downstream inhibition of cytokine release in macrophages.⁹ Studies have demonstrated that VNS reduces systemic inflammatory markers¹⁰ and has had early success treating inflammatory conditions such as arthritis,¹¹ sepsis,¹² and inflammatory bowel disease.¹³ Early evidence has also emerged that noninvasive approaches via transauricular vagus nerve stimulation (taVNS) can accomplish similar effects.¹⁴ Thus, our hypothesis was that implementing taVNS in the acute period following spontaneous SAH would attenuate the expected inflammatory response to hemorrhage and curtail clinical morbidity.

Here, we report results from our randomized controlled trial (RCT) of noninvasive auricular vagus nerve stimulation for subarachnoid hemorrhage (NAVSaH).¹⁵ The primary aims of this trial were to determine if taVNS following SAH reduces TNF-α in the plasma and CSF, and reduces the rate of radiographic vasospasm. Exploratory analyses also evaluated the change in clinical outcomes of patients via blinded assessment with modified Rankin Scale (mRS) scores at discharge and follow-up. In aggregate, the findings support taVNS as a potential novel modality to treat SAH-induced inflammation and associated clinical sequelae.

Methods

A full outline of the NAVSaH trial has been previously published.¹⁵ This study was registered with the ClinicalTrials.gov database (http://clinicaltrials.gov), and its registration no. is NCT04557618.

Trial Design

This study was a prospective, triple-blind RCT with two study arms, with assessment based on intention to treat with regard to the assigned treatment arm. Reporting followed the CONSORT guidelines.¹⁶ This study was approved by the Washington University School of Medicine institutional review board. All participants enrolled in the study had given written informed consent. All procedures followed were in accordance with institutional guidelines.

Participants

Adult patients admitted to Barnes Jewish Hospital following an acute, spontaneous (nontraumatic) SAH were screened for enrollment into the trial. Inclusion criteria included the following: 1) patients with CT-identified, nonperimesencephalic SAH; 2) age ≥ 18 years; and 3) the patient or legal representative was able to give consent for enrollment. Patients must have had symptom onset within 24 hours of presentation to the hospital. Patients were excluded if they were 1) < 18 years old; 2) were taking an immunomodulatory or immunosuppressive medication; 3) were receiving ongoing cancer therapy; 4) had an implanted electrical device such as a pacemaker; 5) were bradycardic on admission (heart rate < 40 beats per minute for > 10 minutes); 6) were positive for the COVID-19 virus; or 7) were at risk of imminent death or were not being offered traditional interventions because of poor prognosis. All candidates were screened based on inclusion and exclusion criteria, consent was obtained, and subjects were enrolled within 24 hours of presentation to our hospital.

Randomization

All identified subjects or their legal medical representative gave their informed consent, and randomization occurred after consent and trial enrollment was complete. Subjects were allocated to treatment arm by using a computer-generated randomization sequence, with next assignment concealed prior to time of randomization. Consequent eligible patients were allocated strictly in sequence. Patients were randomized to receive either taVNS or sham treatment at a 1:1 ratio.

Blinding

The patient/family, the care providers and medical team, and the outcomes assessors were blinded to the treatment arm in which the patients were enrolled. Both treatment arms included the application of ear clip electrodes for the same duration and at the same time intervals, and stimulation parameters used in the taVNS arm were subsensory, ensuring that the patient and casual observer remained blinded to the treatment arm.

Interventions

Following enrollment and randomization into a treatment arm, initial samples of blood and CSF (if the patient had an external ventricular drain in place) were collected. The patient began either taVNS or sham stimulation for 20 minutes twice daily, with a morning session between 05:00 and 10:00 and an evening session between 16:00 and 21:00. Treatment sessions occurred twice a day throughout a patient’s ICU stay, and therefore the cumulative number of days of therapy varied by patient, with all patients receiving at least 7 days of therapy per our hospital’s standard ICU stay protocol (Fig. 1A). All patients were fitted with a portable TENS (transcutaneous electrical nerve stimulation) unit connected to two ear clips applied to the left ear during treatment periods. For VNS treatment, these ear clips were placed along the concha of the ear; for sham treatments, the clips were placed along the ear lobe to avoid stimulation of the auricular vagus nerve from tactile pressure alone in the absence of current¹⁷ (Fig. 1B and C). Stimulation parameters were selected based on prior studies that sought to maximize VNS while avoiding the perception of pain.^17,18 Stimulation parameters used in this trial for the taVNS arm had a duration of 20 minutes, frequency of 20 Hz, pulse width of 250 µsec, and an intensity of 0.4 mA. Sham treatments involved no electrical current, and also had a 20-minute duration.

FIG. 1.

Methods and trial design. Following randomization, patients underwent treatment sessions daily and blood/CSF collection every 3 days (A). Anatomy and cutaneous innervation of the outer ear by the auricular branch (B). Position of the electrodes for patients assigned to the VNS (left) and sham (right) treatment groups (C). Patients in both treatment groups had similar distributions of Hunt and Hess scores (D) and modified Fisher grade scores (E) on presentation. HH = Hunt and Hess; mF = modified Fisher. Figure is available in color online only.

Outcomes

All outcomes were assessed based on the intention to treat with regard to the assigned treatment arm.

Impact of taVNS Following SAH on the Inflammatory Markers TNF-α and IL-6

The primary outcome assessed was the change in the key inflammatory marker TNF-α in the plasma and CSF in patients with SAH undergoing the taVNS versus sham treatment. IL-6 was also evaluated as a secondary outcome in an exploratory manner. Blood samples and CSF samples, when a ventriculostomy was in place, were collected every 3 days throughout patient hospitalization following acute SAH. Baseline blood and CSF samples were collected prior to the first treatment session. Blood and CSF samples were collected and processed immediately. See Supplemental Materials for details of specimen processing.

Impact of taVNS on SAH-Induced Angiographic Vasospasm

The primary outcome assessed was the difference in the presence of moderate or severe radiographic vasospasm between the taVNS and sham stimulation groups, as evaluated by blinded clinician assessment and quantitatively via measurement of vessel caliber on serial examinations. In addition to initial diagnostic testing, all patients underwent a repeat CT or catheter angiogram approximately 7 days after admission, per the institution’s protocol. Further vascular imaging was also performed if there was a concern for clinical vasospasm or stroke. For both planned and indicated imaging sessions, each vascular imaging study was reviewed by a neuroradiologist or endovascular neurointerventionalist blinded to treatment arm, who described the overall imaging study as it related to vasospasm as none, mild, moderate, or severe, to account for proximal vessel caliber and overall/distal perfusion.

To assess radiographic vasospasm more quantitatively, all radiological images were reviewed by a single physician reviewer (D.J.R.) blinded to patient identity, cohort assignment, and indication for the imaging study. The reviewer was fellowship trained in neurointerventional radiology. Images were in DICOM format and reviewed and measured with the RadAnt DICOM viewer (Medixant) software. Cerebral vessel diameters were measured bilaterally at specific locations on maximally magnified images, with consistent measurement locations maintained across serial imaging studies (Fig. 2B). Vessel diameter on each serial image was normalized to the initial angiographic test, with comparisons made only between matching modalities and accounting for anatomical variations or imaging limitations. Vasospasm was defined for each vessel compared with baseline as none/mild if < 25%, moderate if 25%–50%, and severe if > 50%, as previously described.¹⁹ Vessel caliber was also analyzed over time by treatment group. See Supplemental Materials for a detailed description of image analysis techniques.

FIG. 2.

Impact of VNS on radiographic vasospasm. Rate of radiographic vasospasm, as assessed by a blinded interpreter (left) and serial vessel caliber measurements (right) (A). Vessel measurements from a predetermined list of vessel locations (B) were used to assess vessel caliber on serial vascular imaging studies (C). **p < 0.05. ICA = internal carotid artery. Figure is available in color online only.

Secondary outcomes also included the number of vascular imaging studies obtained and interventions undertaken to treat vasospasm, including the following: 1) blood pressure augmentation while in the ICU; 2) treatments performed during catheter angiogram, such as administration of intraarterial vasodilators; and 3) the use of intrathecal vasodilators. Presence of infarct on subsequent CT or MRI was also recorded, as a surrogate marker for delayed cerebral ischemia.

Secondary Outcomes

Additional secondary outcomes included clinical outcome metrics, including discharge destination (home, inpatient rehabilitation facility, skilled nursing facility, hospice, or death), and mRS scores at discharge and first follow-up.

Sample Size

The estimated required sample size for the pilot NAVSaH clinical trial, as previously published,¹⁵ was 50 patients. However, enrollment to study this endpoint was terminated early following an interim analysis, because the effect size for the reported primary aim (presence of moderate/severe radiographic vasospasm) was larger than predicted. In the initial trial design based on early preliminary data, there was anticipated to be a 30% reduction in moderate/severe vasospasm in the taVNS-treated arm, but interim analysis demonstrated a > 40% reduction in the treatment arm. Therefore, results are reported based on 27 patients who completed their initial hospital stay and at least their first outpatient follow-up.

Statistical Methods

Patient demographics and clinical characteristics were summarized using counts and frequencies for categorical variables or means and standard deviations for continuous variables. The distributions of baseline patient characteristics across taVNS and sham-stimulation groups were compared using the Student t-test or chi-square test as appropriate. Between-group differences in the cross-sectional outcomes (such as the presence of moderate or severe radiographic vasospasm, discharge destination, etc.) were compared using a t-test or chi-square test as appropriate. Between-group differences in these repeatedly measured outcomes (such as mRS score [coded as mRS < 3 or not], biomarkers of cytokines, etc.) were compared using linear mixed models (for normality data) or generalized linear mixed models (for nonnormality data) to account for potential correlation among multiple measures taken from the same patients. The fixed effects of the model included intervention groups (taVNS vs sham), the measurement times (treated as a categorical variable), and their interaction. The random effect included the subject-specific intercept, which described the average deviation in an individual from the overall level at baseline. Subsequent, one-sided post hoc tests were performed to assess whether taVNS improved outcomes at specific time points. Normalized vessel caliber was also assessed using a similar mixed model to estimate its average change over time and to compare the effect of taVNS on vasospasm, where the measurement time was treated as a continuous variable instead. The assumption of normality was assessed graphically based on residuals from models. All the analyses were performed using SAS 9.4 (SAS Institute). Unless stated otherwise, all the statistical tests were one-sided for outcome variables and two-tailed for baseline characteristics, with a p value of < 0.05 for significance.

Results

Subject Enrollment, Allocation, and Baseline Demographics

Participant Flow

A flow diagram detailing enrollment, allocation, and follow-up can be found in Supplemental Fig. 1. Sixty-four patients were assessed for eligibility for enrollment in the trial between January 2021 and October 2023. Of those, 27 met the inclusion criteria and were randomized to a treatment arm (13 patients with taVNS and 14 with sham stimulation). One patient in each study arm ceased treatment sessions and laboratory draws before the completion of the intervention. Twenty-seven patients were analyzed for clinical outcomes based on intention to treat, and 26 patients were analyzed for laboratory inflammatory cytokine endpoints, with 1 patient in the taVNS group excluded from analysis because only a single baseline laboratory draw was performed.

Baseline Data

Patients in both treatment arms were not significantly different with regard to sex, race, or method of aneurysm treatment (Table 1). They were also similar with regard to Glasgow Coma Scale score, Hunt and Hess grade, and modified Fisher score on admission (Table 1, Fig. 1D and E).

TABLE 1.

Patient demographics and presenting characteristics on admission

Characteristic	taVNS, n = 13	Sham, n = 14	p Value
Sex			0.30
Male	4 (30.8%)	2 (14.3%)
Female	9 (69.2%)	12 (85.7%)
Race			0.90
White	9 (69.2%)	10 (71.4%)
Black/African American	4 (30.8%)	4 (28.6%)
Tx modality for ruptured aneurysm			0.69
Endovascular	11 (84.6%)	11 (78.6%)
Surgical	2 (15.4%)	3 (21.4%)
Age in yrs	62.9 ± 16.1	56.9 ± 12.6	0.29
Hunt & Hess grade	2.4 ± 1.0	2.6 ± 0.8	0.48
Modified Fisher score	3.2 ± 0.9	3.3 ± 0.9	0.71
Glasgow Coma Scale score	12.8 ± 3.5	12.6 ± 4.1	0.89
mRS score	3.3 ± 1.3	3.4 ± 1.2	0.96

Tx = treatment.

Values are expressed as the number of patients (%) or as the mean ± SD. The parametric two-tailed p value was calculated by ANOVA for numerical covariates and by chi-square test for categorical covariates.

Safety of Intervention

There were no reportable adverse events, including no reports of pain or irritation related to the stimulation site.

Impact of taVNS on Radiographic Vasospasm

When interpreted by blinded evaluators, patients treated with taVNS had a significant reduction in the presence of any radiographic vasospasm (p = 0.035), or radiographic vasospasm described as moderate or severe (p = 0.018) (Table 2, Fig. 2A). When cerebral vessel diameters on serial vascular studies were measured, there was also a significant reduction in the number of vessels with moderate or severe vasospasm when normalized to baseline (p = 0.015) in the taVNS-treated goup (Table 2, Fig. 2A). The normalized vessel calibers significantly increased over time in the taVNS group (with average daily change of 0.030, 95% CI 0.012–0.047), whereas the average daily change in the sham-treated group was −0.002 (95% CI −0.013 to 0.008) (Fig. 2C). The serial vascular studies with normalized vessel caliber showed a significant interaction effect between day and treatment (p = 0.0026) (Fig. 2C). The mean number of follow-up vascular studies obtained was 2.8 (SD 0.8) for the taVNS group and 3.4 (SD 1.2) for the sham group. There were numerically fewer interventions performed for vasospasm, including blood pressure augmentation, intraarterial vasodilator infusion or angioplasty during angiography, and intrathecal vasodilator administration, with no significant differences between groups in this initial cohort (Table 2).

TABLE 2.

Vasospasm following SAH

	taVNS, n = 13	Sham, n = 14	p Value
Clinician assessment of radiographic vasospasm*
Any radiographic vasospasm	7 (53.9%)	12 (85.7%)	0.035
Moderate or severe vasospasm on any vascular imaging study	4 (30.8%)	10 (71.4%)	0.018
Quantitative vasospasm assessment*†
None/mild, <25%	135 (82.3%)	136 (72.3%)	0.028
Moderate, 25–50%	27 (16.5%)	45 (23.9%)
Severe, >50%	2 (1.2%)	7 (3.7%)
Moderate or severe vasospasm	29 (17.7%)	52 (27.7%)	0.015
Vasospasm intervention‡
Blood pressure augmentation	3 (23.1%)	6 (42.9%)	0.41
Intraarterial infusion during angiography	4 (30.8%)	6 (42.9%)	0.35
Angioplasty during angiography	0 (0%)	1 (7.1%)	0.50
Intrathecal vasodilator	2 (15.4%)	2 (14.3%)	0.47
No. of imaging studies	2.8 ± 0.8	3.4 ± 1.2	0.28
No. of days of blood pressure augmentation	1.8 ± 3.5	3.6 ± 5.1	0.28
Delayed cerebral ischemia‡
Presence of infarct on CT or MRI	3 (23.1%)	6 (42.9%)	0.41

Unless otherwise indicated, values are expressed as the number of patients (%) or as the mean ± SD. The p value was calculated by ANOVA for numerical covariates and by chi-square test for categorical covariates. Boldface type indicates statistical significance.

^*One-tailed statistical tests.

^†Values are given as the number of vessels (%). There were 164 vessels in the taVNS group and 188 in the sham-stimulation group.

^‡Two-tailed statistical tests.

Impact of taVNS on Inflammatory Markers Following SAH

A total of 26 patients were included in the analysis of plasma TNF-α and IL-6, with 1 patient excluded due to only having a single baseline level collected. The proinflammatory cytokine TNF-α was significantly reduced in the plasma on treatment days 7 and 10 in patients treated with taVNS (p = 0.015 and p = 0.030, respectively), and IL-6 was significantly lower in the plasma on day 4 in patients treated with taVNS (p = 0.024) (Fig. 3A and B). A total of 13 patients had an extraventricular drain in place (6 with taVNS, 7 with sham treatment) and were included in the analysis of CSF TNF-α and IL-6. The proinflammatory cytokines TNF-α and IL-6 were both significantly reduced in the CSF on treatment day 13 in patients treated with taVNS (p = 0.031 and p = 0.025, respectively) (Fig. 3C and D).

FIG. 3.

Impact of VNS on inflammatory cytokines. Serum (A and B) and CSF (C and D) measurements of IL-6 and TNF-α over the course of admission following SAH in pg/mL. Day 1 represents baseline prior to initiation of first treatment. **p < 0.05. Figure is available in color online only.

Impact of taVNS on Patient Outcomes

Blinded assessments of the mRS scores on admission, discharge, and at first outpatient follow-up are detailed in Tables 1 and 3 and in Fig. 4. One patient in each of the taVNS and sham groups was lost to follow-up, with no mRS score assessed after discharge. mRS scores on admission were not significantly different between groups. A good outcome, as assessed via the mRS, refers to scores 0–2, whereas a poor outcome refers to scores 3–6. The change in mRS scores over the course of admission and at first follow-up for all patients is illustrated in Fig. 4C. Patients receiving taVNS demonstrated higher rates of favorable outcomes compared to those in the sham group both at discharge (38.4% vs 21.4%) and at first follow-up (76.9% vs 57.1%). There was no significant interaction between mRS scores over time and treatment group (p = 0.20). However, on post hoc analysis there was a significant difference in improvement observed in the taVNS group from admission to first follow-up (p = 0.014), but no significant difference was noted in the sham group (p = 0.18) (Fig. 4D). Following hospitalization for SAH, patients had one of 5 discharge outcomes: home, inpatient rehabilitation facility, skilled nursing facility, hospice, or death. No patient in either treatment arm died before discharge from the hospital. With discharge destinations divided into good discharge (home and inpatient rehabilitation) versus poor discharge (skilled nursing facility, hospice, or death), there was a significantly lower rate of poor discharge in patients treated with taVNS (p = 0.04) (Table 3 and Fig. 4E).

TABLE 3.

Clinical outcomes following SAH

	taVNS, n = 13	Sham, n = 14	p Value
Discharge destination
Home	8 (61.5%)	6 (42.9%)
Inpatient rehabilitation	4 (30.8%)	3 (21.4%)
Skilled nursing facility	0 (0%)	5 (35.7%)
Hospice	1 (7.7%)	0 (0%)
Death	0 (0%)	0 (0%)
Good vs poor discharge destination*
Good discharge	12 (92.3%)	9 (64.3%)
Poor discharge	1 (7.7%)	5 (35.7%)	0.04
mRS score
On discharge	2.7 ± 1.5	3.3 ± 1.2	0.18
At first follow-up	1.8 ± 1.5	2.1 ± 1.8	0.40
Good vs poor outcome per mRS score†
Good outcome at discharge	5 (38.5%)	3 (21.4%)	0.21
Good outcome at first follow-up	10 (76.9%)	8 (57.1%)	0.19

Values are expressed as the number of patients (%) or as the mean ± SD. The p value was calculated by ANOVA for numerical covariates and by chi-square test for categorical covariates, all one-tailed tests. Boldface type indicates statistical significance.

^*Good discharge destination includes home or inpatient rehabilitation, and poor discharge destination includes skilled nursing facility, hospice, or death.

^†Good outcome is mRS score < 3.

FIG. 4.

Impact of VNS on clinical outcomes. mRS scores from blinded assessors at admission, discharge, and at first follow-up for patients treated with VNS (A), and sham (B) stimulation. Patients with a good mRS score of < 3 shown over time (C) and as a change for presentation (D) for both treatment groups. Discharge destination by treatment group (E). **p < 0.05. LTF = lost to follow-up; SNF = skilled nursing facility. Figure is available in color online only.

Discussion

This study is the first to use noninvasive neuromodulation to mitigate inflammation in patients with SAH. Despite improvements in managing these patients, mortality and morbidity rates remain high.¹ Targeting posthemorrhage inflammation is important for improving outcomes.^3,4,6 We show that taVNS can significantly reduce key inflammatory cytokines (TNF-α and IL-6), reduce cerebral vasospasm, and improve clinical outcomes. These findings provide a new therapeutic approach for treating SAH sequelae.

Following SAH, blood within the subarachnoid space triggers central and systemic inflammatory responses. Key drivers of SAH-induced inflammation are the cytokines TNF-α and IL-6. In animal models and humans, these cytokines are associated with vasospasm, hydrocephalus, delayed cerebral ischemia, and poor outcomes.^4,5,20–25 The relationship between increased inflammation and worsened clinical outcomes is believed to arise from several mechanisms that include inducing vasoconstriction, degrading the blood-brain barrier, and prompting neuronal cell death within the parenchyma itself.²⁶

Numerous antiinflammatory interventions have been assessed in trials in humans following SAH to target inflammatory pathways. In smaller enrollment studies, there has been some early evidence of clinical benefit with cyclosporine A²⁷ and steroids.^28,29 Other medications, such as clazosentan,³⁰ cilostazol,³¹ and IL-1 antagonists³² demonstrated no impact on overall outcomes. In larger meta-analysis studies, simvastatin,³³ aspirin, nonsteroidal antiinflammatory medications, and thienopyridines³⁴ all demonstrated no improvement. Thus, while some pathway-targeted pharmacological approaches have led to changes in secondary outcomes of vasospasm and delayed cerebral ischemia in clinical trials,^31,35 these approaches have ultimately failed to produce an effective intervention that reliably improves functional or neurological outcomes in patients with SAH.^35,36 This dissociation between early trial endpoints and a lack of improved neurological clinical outcomes suggests that a narrow, molecular pathway-oriented approach may be insufficient to address the broad cascade of physiological drivers that result in poor clinical outcomes following SAH.

VNS may provide a broader immunomodulatory approach. VNS reduces inflammation through the cholinergic antiinflammatory pathway. Sensory neurons activated by infection or injury travel to the brainstem via the vagus nerve.³⁷ Subsequent neural efferent input to the spleen and other organs leads to the release of acetylcholine, which interacts with α7 nicotinic acetylcholine receptors on immunocompetent cells, inhibiting cytokine release in macrophages⁹ (Fig. 5). VNS has been successfully used in models of cerebral ischemia/reperfusion,³⁸ rheumatoid arthritis,¹¹ sepsis,¹² inflammatory bowel diseases,¹³ and cerebral aneurysms and SAH.³⁹ Clinically, VNS has historically been performed by surgical cervical neck dissection and placement of a cuff electrode directly around the nerve. Alternatively, VNS can be accomplished noninvasively by stimulating the auricular branch of the vagus nerve as it courses through the external ear¹⁴ (Fig. 1). This noninvasive and low-risk form of stimulation lends itself to use in patients in the ICU who are medically fragile and would not tolerate surgical implantation of a device. Prior to this study, however, there has been a dearth of understanding of VNS effects on patients with SAH.

FIG. 5.

Mechanism of action for VNS to attenuate the inflammatory response following SAH. Given that SAH induces a deleterious inflammatory response resulting in increased central and peripheral proinflammatory markers IL-6 and TNF-α, and can be associated with intracranial vasospasm. Stimulating the transcutaneous auricular branch of the vagus nerve reduces inflammatory cytokines as well as cerebral vasospasm, as mediated by downstream action on macrophages as a result of vagus nerve splenic innervation. © Eric Leuthardt, published with permission.

This prospective RCT is the first to report the impact of noninvasive VNS applied following spontaneous SAH. We demonstrate a significant reduction in radiographic vasospasm as assessed by two key metrics: 1) assessment of overall clinical vasospasm by blinded radiology interpreters; and 2) quantitative serial vessel caliber measurements. The blinded assessor’s interpretation of vasospasm captures the qualitative radiological assessment that accounts for large proximal vessel caliber changes, as well as harder to quantify distal perfusion alterations from medium vessel caliber changes. The serial quantitative analysis more analytically compares changes in the large proximal vessels over time, limiting potential differences in qualitative assessments by different neuroradiologists. Our finding of significant changes in radiographic vasospasm is a substantial one, but alteration in radiographic vasospasm alone has not historically been directly or clearly related to changes in functional outcome.⁴⁰ Thus, the significant reduction in poor discharge destinations (skilled nursing facilities and hospice) and early evidence of superior improvement in mRS scores between admission and first follow-up suggests that the impact of taVNS is not limited to improvement in radiographic vasospasm alone. Despite the promising nature of these preliminary findings, further validation and study of clinical outcomes in a larger cohort will be needed. For example, we saw numerically fewer patients assigned an mRS score of 0 in the taVNS group at first follow-up compared to the sham group (1 vs 3, respectively), despite an improvement in the average mRS score in the taVNS group.

In this study, we hypothesized that the underlying mechanism for this vascular and clinical effect was due to taVNS’s ability to blunt the deleterious inflammatory response following SAH (Fig. 5). We demonstrate that taVNS can achieve significant reductions in TNF-α and IL-6 both centrally and systemically. Notably, these effects seem to have different time scales. Whereas serum markers generally showed a more linear trend of reduced TNF-α and IL-6 from the onset of treatment, the difference in CSF TNF-α and IL-6 was more delayed and significantly different at day 14. These findings suggest that the serial interventions (i.e., 20 minutes, twice daily, for 14 days) may have a cumulative effect over time that is necessary to achieve the full clinical benefit. While these findings validated the immunomodulatory impact of taVNS, a more comprehensive evaluation of inflammatory cytokines will be required in the future to define further how taVNS fully impacts the inflammatory cascade and the interplay of central and systemic inflammation.

Although inflammation is being altered with taVNS, there may be other mechanisms that play a role. Following SAH, there is a widespread sympathetic response, which can manifest with increased risk for cerebral vasospasm, delayed cortical ischemia, and extracerebral organ damage.⁴¹ Although no acute bedside perceptible alterations in cardiovascular metrics were seen in any of our study’s patients, our rigorous analyses of continuous heart rate, blood pressure, and respiratory data throughout patients’ ICU stays indicate that treatment with VNS may restore the SAH-induced autonomic imbalance by enhancing parasympathetic input.^42,43 Another phenomenon observed in patients with SAH is cortical spreading depolarizations (CSDs), with evidence of a causal relationship between CSDs and worse neurological outcomes.⁴⁴ VNS has been demonstrated to decrease CSDs in a study of migraine,⁴⁵ which supports the possibility that taVNS following SAH may have a similar effect. These multipathway mechanisms of taVNS may explain the robust clinical benefits seen in this study, whereas narrowly focused pharmacological approaches have thus far been unsuccessful.

The results of this study are encouraging, but the current report has limitations. The difference in electrode placement between treatment arms could allow detection of a group difference, although to which group a patient was assigned would not be clear. Clinician presumption of treatment arm could influence their decisions regarding other medical care or interventions. Although this study was randomized, it was done at a single institution, and thus we cannot infer yet whether these findings are more generalizable. This randomized cohort, while showing a positive and robust effect, is still small. Importantly, the SAH population can be heterogeneous and a patient’s hospital course can be influenced by many factors. Therefore, in a small cohort there may be confounding factors including a patient’s medical history, intraprocedural events during aneurysm treatment, or variability in management between clinicians that may influence patient outcomes. Ultimately, a larger multicenter clinical trial will be essential to fully validate the clinical effect of taVNS and demonstrate that the treatment can be widely and effectively applied.

Conclusions

This single-center RCT represents the first reported effect of taVNS for the treatment of SAH. taVNS modulates the endogenous neuro-immunological system through a nonpharmacological bioelectric approach and attenuates the inflammatory response and concomitant vasculopathic sequelae following SAH. Given the low risk, ease of application, and evidence of clinical benefit, this is a novel technique that can potentially impact SAH management in the future.

Acknowledgments

No sponsor or funding source had any role in the trial design, data acquisition, data analysis, or writing of this report. We thank the Bursky Center for Human Immunology & Immunotherapy Programs (CHiiPs) Core. Our study was supported by grants from the following sources: the Aneurysm and AVM Foundation (A.L.H.); the AANS, Robert J. Dempsey, MD, Cerebrovascular Research Award (A.L.H.); Washington University, Just in Time Funding (A.L.H., E.C.L.); NINDS, StrokeNet Fellowship Program (A.L.H.); NINDS grant R21-NS128307 (A.L.H., E.C.L., G.J.Z., P.B.); National Institute of Biomedical Imaging and Bioengineering, grant P41-EB018783 (E.C.L., P.B.); and the McDonnell Center for System Neuroscience (E.C.L., P.B.) during the conduct of the study.

Disclosures

Dr. Huguenard reported stock ownership and board membership from Aurenar outside the submitted work; in addition, she had a patent 18548755 pending. Dr. Osbun reported personal fees from Stryker, Inc., Penumbra, Inc., Terumo, Inc., Medtronic, Inc., and Microvention, Inc., outside the submitted work. Dr. Limbrick reported being the Chief Medical Officer at Rhaeos, Inc., and being an advisor at Deepsight, Inc., outside the submitted work. Dr. Leuthardt reported equity from Aurenar outside the submitted work; in addition, he had a patent for PCT/US2022/018864 pending and a patent for PCT/US2022/04584 l pending.

Author Contributions

Conception and design: Leuthardt, Huguenard, Osbun, Zipfel, Brunner. Acquisition of data: Leuthardt, Huguenard, Rivet, Johnson, Adamek, Coxon, Kummer, Osbun, Vellimana. Analysis and interpretation of data: Leuthardt, Huguenard, Tan, Rivet, Gao, Johnson, Vellimana, Limbrick, Zipfel. Drafting the article: Leuthardt, Huguenard. Critically revising the article: Leuthardt, Huguenard, Tan, Rivet, Johnson, Kummer, Vellimana, Limbrick, Zipfel, Brunner. Reviewed submitted version of manuscript: Leuthardt, Huguenard, Rivet, Johnson, Adamek, Coxon, Kummer, Osbun, Vellimana, Limbrick, Zipfel, Brunner. Approved the final version of the manuscript on behalf of all authors: Leuthardt. Statistical analysis: Tan, Gao. Administrative/technical/material support: Adamek, Coxon, Kummer, Limbrick, Brunner. Study supervision: Leuthardt, Huguenard, Brunner.

Previous Presentations

Preliminary findings presented at: AANS meeting (Philadelphia, PA, 4/29/2022–5/2/2022), Academy of Neurological Surgeons meetings (Colorado Springs, CO, 9/28/2022–10/1/2022), and Society of NeuroInterventional Surgery Annual Meeting (San Diego, CA, 7/31/2023–8/4/2023), and the Southern Neurosurgical Society Meeting (Orlando, FL, 3/6/2024–3/9/2024).

Preprint Server

An earlier version of this article can be found on a preprint server.

Preprint server name: medRxiv.

Preprint DOI: 10.1101/2024.04.29.24306598.