Transauricular Vagus Nerve Stimulation in Acute Ischaemic Stroke Requiring Mechanical Thrombectomy: Sham-Controlled, Randomised Device Trial — a neutral primary endpoint

Clinical Question (PICO)

Among adults with acute ischemic stroke referred for emergency mechanical thrombectomy who have hypertension or elevated admission blood pressure, does bilateral transcutaneous auricular vagus nerve stimulation (tVNS), compared with an indistinguishable sham device, reduce the coefficient of variation of systolic blood pressure over the first 24 hours? Whole-blood transcriptomics and heart-rate variability were exploratory/secondary mechanistic measures; no prespecified treatment-effect subgroups were reported.

Bottom Line

Early bilateral tVNS was feasible and produced no reported serious adverse events in this 36-person phase 2 pilot, but did not improve its prespecified primary endpoint: 24-hour systolic-BP coefficient of variation was 0.106 ± 0.029 with active stimulation versus 0.107 ± 0.027 with sham (P = 0.93). The transcriptomic signal is hypothesis-generating, not evidence of improved stroke recovery.

Design

Trial type: Single-center, prospective, phase 2, randomized, sham-controlled device trial; participants, treating clinicians, investigators, laboratory staff and primary-outcome assessors masked as described by the authors.
N: 36 randomized (44 screened).
- Active bilateral tVNS: 18.
- Sham bilateral tVNS: 18.
Randomization: 1:1, online allocation in blocks of four; no minimization.
Setting: Royal London Hospital, Barts Health NHS Trust, London, UK.
Enrollment: April 26–July 14, 2023.
Follow-up: Primary physiological endpoint through 24 hours after thrombectomy; clinical infection and adverse-event surveillance through day 7. No 90-day disability endpoint was tested.
Analysis: Prespecified intention-to-treat primary analysis, assuming missingness at random; sensitivity analysis for early repatriation. Two-sided alpha 0.05.
Primary outcome: Coefficient of variation (standard deviation divided by mean) of serial systolic BP over 24 hours after mechanical thrombectomy.

Population

Inclusion Criteria

Age >18 years and acute ischemic stroke referred for emergent mechanical thrombectomy.
Established hypertension and/or elevated admission BP (systolic >140 mmHg or diastolic >90 mmHg).

Exclusion Criteria

Enrollment in a trial addressing a similar biological mechanism or previous enrollment in this study.
Ear anatomy precluding device placement; pregnancy.

Baseline Characteristics

Mean age 68.7 years in active and 68.6 in sham; women 13/18 (72%) versus 9/18 (50%).
Admission NIHSS mean 16 (SD 7) active versus 13 (SD 5) sham.
General anesthesia 14/18 (78%) active versus 10/18 (56%) sham; thrombectomy median 80 versus 51 minutes, respectively. These imbalances matter in a small physiological study.

Interventions

Active tVNS: Bilateral electrodes on the ear tragi, starting before sedation/anesthesia and continuing throughout thrombectomy; 3 mA, 200-µs pulses at 25 Hz. A second, 60-minute session was scheduled the following morning.
Sham tVNS: Identical setup and schedule, but a manufactured break in the ear leads prevented delivery of current. Standard thrombectomy and BP treatment continued in both groups; antihypertensive management was not protocolized.
Nine patients were repatriated the same day; four active and five sham patients missed the second session.

Outcomes

Primary Outcome (24-hour systolic-BP coefficient of variation; prespecified ITT):

Active 0.106 ± 0.029 versus sham 0.107 ± 0.027; P = 0.93. Observed absolute difference = −0.001, approximately a 0.9% relative decrease against sham. The paper does not report a primary-outcome confidence interval or subgroup effect estimate; do not invent one.
Prespecified subgroups: None reported. Do not mistake time-epoch comparisons or secondary biomarkers for randomized subgroup effects.

Secondary Outcomes / Mechanistic Measures:

The authors describe differences in alternate measures of BP dispersion and in low/very-low-frequency heart-rate-variability power, but these did not rescue the neutral primary endpoint. Only 22 participants had complete Holter recordings for HRV analysis; 10 with atrial fibrillation were ineligible.
In next-morning whole-blood RNA sequencing, active tVNS was associated with differential expression of inflammatory-pathway genes (including TNF signaling); the figure depicts samples from six participants. Gene-expression changes are not proof of restored immune function, fewer infections or better disability outcomes. The Results text reports 52 upregulated/4 downregulated genes, while the Figure 5 caption says 52/5—an internal discrepancy.
At 24 hours, median NIHSS was 10 (IQR 8–18) active versus 10 (5–19) sham; this pilot was not powered for functional recovery.

Adverse Events / Safety:

No serious adverse events attributable to active tVNS were reported. One episode of skin irritation was away from the intervention site.
Clinically detected arrhythmias: 5/18 active versus 4/18 sham. Two deaths within seven days occurred in the sham arm and none in the active arm; this small, unpowered difference cannot establish a mortality benefit. Intracranial hemorrhage was not reported as a comparative safety endpoint.

Figures

*Figure 3.* Blood pressure and blood pressure variability. A Systolic blood pressure over first 24 h after admission for mechanical thrombectomy, calculated as mean values over 6 h epochs (median,25–75th centile box/whisker plots). P value refers to post-hoc Tukey-Kramer comparisons between measurements made during the first 6 h period and each subsequent 6 h periods. B Standard deviation of systolic blood pressure over first 24 h after admission for mechanical thrombectomy (median, 25–75th centile box/whisker plots). C Median absolute deviation of systolic blood pressure over first 24 h after admission for mechanical thrombectomy (median, 25–75th centile box/whisker plots). Each individual’s summary measure for the 6 h period is shown in the overlaying dot plots. P value refers to post-hoc Tukey-Kramer comparisons between first 6 h period and each subsequent 6 h period. (Supplementary table 1 for detailed data).

*Figure 4.* Heart rate and heart rate variability. A Heart rate over first 24 h after admission for mechanical thrombectomy (median,25–75th centiles). P values refer to post-hoc Tukey-Kramer comparison after repeat-measures ANOVA (within group comparison) B RR-interval. C Standard deviation of heart rate (SD HR). D Standard Deviation of Normal-to-Normal intervals (SDNN). E Root Mean Square of Successive Differences (RMSSD). F Absolute very-low power (log scale). G Absolute low power (log scale). H Absolute high-frequency power (log scale). For panels 4B-H, box/whisker show median, 25–75th centiles, with overlaying dot plots showing each individual’s summary measure for each stimulation period (both MT and morning after MT intervention periods). P values refer to within group comparison for parameter before and after each stimulation period (both MT and morning after MT intervention periods). Italicised p value with asterisk refers to comparison between sham versus active tVNS over the first 24 h of admission for MT. Additional HRV analyses are provided in Supplementary Figures 2, 3

*Figure 5.* Systemic inflammation quantified by transcriptomic changes in whole blood. A Neutrophil/lymphocyte ratio on admission for mechanical thrombectomy. Box and whiskers plot show median, 25–75th centiles. B Heatmap demonstrating transcriptomic changes on the morning after mechanical thrombectomy for 6 participants randomised to sham or active therapy. C Volcano plot depicting differentially expressed genes, of which 52 were upregulated and 5 downregulated by active tVNS (false-discovery rate <0.01; minimum fold-change ≥1.5). There were no differences in gene transcription in admission blood samples. D Signaling pathways significantly altered, adjusted for multiple testing with a false-discovery p-value <0.01. The number of genes differentially expressed are shown in the bars for each signaling pathway.

Criticisms

Neutral primary result: Feasibility and transcriptomic associations must not be presented as clinical efficacy.
Small and imbalanced: N = 36; active patients had longer thrombectomy and stimulation exposure and more general anesthesia, any of which could affect hemodynamics.
Missing and selected measures: Nine early repatriations and nine missed second sessions constrain the 24-hour physiology; HRV excluded atrial fibrillation and had only 22 complete recordings. The transcriptomic figure shows six samples, limiting inference.
Endpoint/power mismatch: The sample-size justification targeted systolic-BP standard deviation falling from 15 to 10 mmHg, whereas the primary analyzed endpoint was the coefficient of variation; a clinically important smaller effect may not have been excluded.
External validity: Single UK thrombectomy center, selected hypertensive patients and short follow-up; no tested 90-day modified Rankin Scale effect. Device settings and ideal dose remain uncertain.

Funding

The study reports support from the INNOVATE-UK/UKRI research program; individual investigators received British Heart Foundation program grants, an NIHR Advanced Fellowship, a Royal College of Physicians Wolfson award and a John Snow award. The article does not specify a funder role in analysis or publication. Authors declare no competing interests. Trial registration: NCT05417009 (registered June 8, 2022).

The paper

Authors. Ackland GL et al.
Title. Transauricular Vagus Nerve Stimulation in Acute Ischaemic Stroke Requiring Mechanical Thrombectomy: Sham-Controlled, Randomised Device Trial.
Journal. Translational Stroke Research.
Year. 2026 volume/publication year (published online December 27, 2025).
DOI. 10.1007/s12975-025-01404-7
PMCID. PMC12743696
License. CC BY 4.0; attributed article figures reproduced for teaching.

Deep Dive — click to expand

What this is

This 2023-enrolled, sham-controlled thrombectomy pilot asked whether stimulating the auricular vagus nerve could stabilize the autonomic response after ischemic stroke. The central measured outcome was not disability but 24-hour systolic-BP variability; it was essentially unchanged (0.106 versus 0.107, P = 0.93). The paper’s useful contribution is acute-procedure feasibility and a provisional physiological/transcriptomic research signal, not a reason to add tVNS to standard stroke treatment.

1. Shadow Audit

The positive-sounding biological narrative sits beside a decisively neutral primary comparison: an absolute coefficient-of-variation difference of only −0.001. The RNA-seq heatmap represents six patients, while nine participants were transferred early and missed the next-morning stimulation (four active, five sham). The 52-versus-4/5 downregulated-gene inconsistency between Results and Figure 5 caption should be clarified before treating a particular gene count as firm. There is no demonstrated reduction in infection, hemorrhage, disability or mortality.

2. Inversion Engine

For tVNS to become a BP-stabilizing intervention despite these data, a better-powered study would need to show a reproducible treatment difference much larger than the observed −0.001 (~0.9% relative), with a prespecified clinically meaningful threshold and confidence interval excluding trivial benefit. The investigators powered this pilot for an SD reduction from 15 to 10 mmHg (one-third), not for the reported CV effect; those are different scales, so the SD target cannot be silently translated into a validated CV cutoff. Conversely, a larger trial reproducing a CV near 0.106 versus 0.107 and failing to improve patient-centered outcomes would close the efficacy case for this regimen.

3. Second-Order Catalyst

The immediate mover should be a thrombectomy research service, not a routine stroke-call protocol: feasibility now supports a multicenter sham-controlled trial with standardized BP management, consistent stimulation exposure, predefined HRV handling for atrial fibrillation, and 90-day functional outcomes. Device setup must not delay reperfusion. A unit could also prospectively standardize BP sampling and track transfers before enrolling, rather than mistake a pilot signal for a practice guideline.

4. Asymmetric Leverage

A noninvasive, short-duration device could scale across many thrombectomies if it eventually prevented a meaningful outcome, so a modest validated patient-level effect might matter. But in this sample the clinical numerator is zero demonstrated functional benefit, the BP signal is ~0.9% relative, and the attractive transcriptomic signal comes from a tiny subset. The asymmetry today favors rigorous testing of a low-burden intervention, not purchasing devices or exposing every patient on the strength of biomarkers.

5. Paradigm Destroyer

This study should end the reflex that a plausible vagal/autonomic mechanism plus an altered gene-expression panel equals stroke benefit. Protocol update tomorrow: “During thrombectomy, prioritize timely reperfusion and established hemodynamic care; do not add auricular tVNS routinely for BP stabilization on current evidence. Offer stimulation only through an approved, sham-controlled protocol that does not delay treatment and records patient-centered outcomes.”

MVP — Minimum Viable Proof

A multicenter sham-controlled study must show a prespecified, nontrivial improvement in 24-hour BP variability and a credible improvement in a patient-relevant outcome such as 90-day function or a rigorously adjudicated complication, with an acceptable safety profile. This 36-person pilot meets the feasibility prerequisite, not that proof.

Best Combination

Interpret this as an adjunct to the established thrombectomy pathway and the broader literature linking post-thrombectomy BP variability with outcomes—not evidence that changing variability by tVNS changes those outcomes. Combine physiology, standardized BP treatment, and clinical endpoints in the next trial; avoid substituting mechanistic plausibility for randomized functional benefit.

Overvalue Warning

First, do not call the lower seven-day death count in the active arm a survival advantage (0/18 versus 2/18; profoundly underpowered). Second, do not describe a six-patient RNA-seq signal as prevention of post-stroke immunosuppression or infection; neither clinical claim was established.

Educational use notice. The Time Is Brain Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes, with attribution to the original publication. We do not claim copyright over any embedded figure; all rights remain with the original publisher and authors under the article's published license. If you are a rights holder and would like a figure removed, please contact the editorial team. See our editorial notice for the full policy.