Efficacy and Mechanisms of Combined Central-Peripheral Magnetic Stimulation on Prefrontal-Sensorimotor Cortex Connectivity in Post-Stroke Dysphagia after Supratentorial Stroke: A Randomized, Single-Blind, Parallel-Controlled Trial — promising swallowing gains, not yet proof of synergy

Clinical Question (PICO)

In adults aged 30–90 with dysphagia after a first unilateral subcortical ischemic or hemorrhagic stroke, does two weeks of active ipsilesional repetitive transcranial magnetic stimulation (rTMS) followed by active suprahyoid repetitive peripheral magnetic stimulation (rPMS), added to conventional swallowing rehabilitation, improve oral intake and airway protection compared with either single active modality or double sham? The primary measures were the Functional Oral Intake Scale (FOIS) and Penetration-Aspiration Scale (PAS). The report does not present prespecified clinical subgroup treatment estimates.

Context: This is a July 2026 publication reviewed through the look-back pathway, not a new October 22 trial report. Participants were predominantly in the chronic phase, with arm-level mean times since stroke of 10–13 months.

Bottom Line

The authors conclude that combined central-peripheral magnetic stimulation was associated with improved swallowing function, potentially linked to stronger sensorimotor connectivity. Compared with double sham plus rehabilitation, combined treatment produced a 1.09-point higher FOIS immediately after treatment (95% CI 0.39–1.79) and a 1.45-point higher FOIS at the 30-day follow-up (95% CI 1.00–1.91); both adjusted P < 0.001. This is a promising small-study signal, not proof that adding peripheral stimulation is superior to active central stimulation alone or that pneumonia, tube dependence or mortality is reduced.

Design

Trial type: Prospective, single-center, randomized, single-blind, four-arm, sham-controlled trial, with active/sham combinations of two stimulation modalities.
N: 76 screened; 44 randomized, 11 per arm.
- Active rTMS + active rPMS: 11 randomized and analyzed.
- Active rTMS + sham rPMS: 11 randomized; 10 analyzed after one treatment discontinuation.
- Sham rTMS + active rPMS: 11 randomized and analyzed.
- Sham rTMS + sham rPMS: 11 randomized and analyzed.
Randomization: Computer-generated permuted blocks of four; equal allocation. Allocation-concealment procedures are not described in the retrieved full report.
Blinding: Described as single-blind. The VFSS analyst was explicitly blinded to group and time point, with videos randomly reordered; the report does not give a complete role-by-role masking description or a test of sham credibility.
Setting: Rehabilitation Department, Fifth Affiliated Hospital of Guangzhou Medical University, China.
Enrollment: Calendar recruitment dates are not reported. Registry: ChiCTR2500113782; registration date reported as December 3, 2025. Prospective registration relative to enrollment cannot be established from the article alone.
Follow-up: Baseline, one day after the two-week treatment course, and a 30-day follow-up for FOIS. PAS, VFSS physiology and fNIRS were assessed before and immediately after treatment, not at the later FOIS visit.
Analysis: 43 completers; the discontinued participant was excluded, rather than retained in a full intention-to-treat analysis. Generalized estimating equations for FOIS/PAS, Bonferroni-adjusted pairwise comparisons, repeated-measures ANOVA for VFSS parameters, and a nonparametric sensitivity analysis for superior hyoid excursion.
Primary outcomes: FOIS, an ordinal oral-intake measure in which higher scores indicate better function; PAS, an ordinal penetration/aspiration measure in which lower scores indicate better airway protection. These are swallowing outcomes, not clinical pneumonia or survival endpoints.
Planned sample size: 48, or 12 per arm; actual randomized enrollment was 44.

Population

Inclusion Criteria

Age 30–90; first-ever unilateral subcortical ischemic or hemorrhagic stroke confirmed by CT or MRI.
Kubota Water Swallowing Test grade ≥3, followed by VFSS confirmation.
VFSS evidence of penetration (PAS ≥3) or aspiration (PAS ≥6).

Exclusion Criteria

Other relevant neurological diseases, structural oropharyngeal/esophageal abnormalities, prior oropharyngeal/neck surgery, or dysphagia of another cause.
Epilepsy, metal implants, psychiatric disorders or severe claustrophobia.
Cognitive impairment or aphasia, inability to cooperate, or refusal of consent.

Baseline Characteristics (combined-active arm)

Age 68 ± 5.5 years; 7 men and 4 women.
Eight ischemic and three hemorrhagic strokes; five left-sided and six right-sided lesions.
Time since stroke 11 ± 5.9 months.
FOIS 1.5 ± 0.5 and PAS 6.2 ± 0.6, indicating substantial baseline impairment.
The baseline table retains 11 participants in the active-rTMS/sham-rPMS arm, whereas the outcome analysis includes only 10; the denominators should not be conflated.

Interventions

All arms received 30 minutes of conventional swallowing rehabilitation after stimulation, six treatment days per week for two weeks.

Active rTMS + active rPMS: rTMS first, then rPMS. Central stimulation targeted ipsilesional pharyngeal M1 using a figure-of-eight coil, at a scalp coordinate 6 cm lateral and 3 cm anterior to Cz. Intensity was 90%–120% of the resting motor threshold measured from the unaffected thenar muscles. Peripheral stimulation used a custom 2.5-cm coil over the suprahyoid region, initially set at 30% of the threshold eliciting a visible twitch in the lesioned-side forearm extensors and adjusted to tolerance.
Active rTMS + sham rPMS: Same central protocol, with the peripheral coil tilted away from the target.
Sham rTMS + active rPMS: Same peripheral protocol, with the central coil tilted away from the target.
Double sham: Both coils tilted 90 degrees away from their respective targets, preserving auditory cues.

Both active modalities used 10 Hz, 60 trains of 20 pulses and a reported 15-second inter-train interval. This is the authors' experimental protocol, not a bedside prescription; the trial did not test reversing the sequence or delivering the modalities simultaneously.

Outcomes

Primary Outcomes — oral intake and airway protection:

FOIS showed a group-by-time interaction (Wald χ² = 39.816, df = 6, P < 0.001); PAS also showed a group-by-time interaction (Wald χ² = 45.433, df = 3, P < 0.001). These are comparisons of trajectories across the four treatment arms, not proof of a biological interaction between the two active modalities.

Reported adjusted between-arm comparison	Mean difference	95% CI	Adjusted P
FOIS, combined active minus double sham, immediately after treatment	+1.09 points	+0.39 to +1.79	<0.001
FOIS, combined active minus double sham, 30-day follow-up	+1.45 points	+1.00 to +1.91	<0.001
PAS, combined active minus peripheral-only active, immediately after treatment	−0.91 points	−1.44 to −0.38	<0.001
PAS, combined active minus double sham, immediately after treatment	−1.55 points	−2.11 to −0.99	<0.001

All four arms improved, including double sham plus rehabilitation. The report does not supply a complete set of adjusted pairwise estimates demonstrating combined-treatment superiority over active rTMS alone. No HR, OR or RR is reported for these scale outcomes, and no verified treatment-effect estimates by stroke type, side, age or chronicity are available. Subgroup-specific benefit must not be inferred from the pooled results.

Secondary Outcomes — swallowing physiology and cortical measurements:

The authors describe increases in laryngeal vestibular closure duration (LCD) and superior hyoid excursion (SHE) in the combined arm. Their reported within-arm contrasts are LCD MD −3.598 (95% CI −4.177 to −3.019) and SHE MD −3.708 (95% CI −4.213 to −3.18), both adjusted P < 0.001. These negative contrasts accompany described improvements; the prose does not clearly specify the contrast direction or units. They are preserved as reported and must not be read as directly interpretable treatment-versus-control effects.
In the combined arm, ipsilesional M1–S1 functional connectivity increased from 0.46 ± 0.31 to 0.71 ± 0.30 (P = 0.04). The reported correlations between connectivity change and LCD/SHE change were Spearman r = 0.78 (P = 0.004) and r = 0.67 (P = 0.023), respectively.
These are exploratory imaging/physiology associations within a small arm, not randomized mediation estimates or proof of a causal neural mechanism. The article does not establish reduced pneumonia, feeding-tube removal, hospitalization or mortality.

Adverse Events / Safety:

One participant assigned to active rTMS plus sham rPMS stopped after six sessions because of intolerable right anterior-neck pain and was excluded from analysis.
Thirteen participants reported transient forehead muscle twitching: six combined-active, two central-only active and five peripheral-only active; none in double sham.
The authors reported no headache, tinnitus, dizziness or seizures. The small, selected cohort and short observation period cannot exclude uncommon or delayed harms.

Figures

The original figures are embedded unchanged from verified PMC CDN assets. Source: Luo and colleagues, Translational Stroke Research (2026), PMC13400478, under the article's CC BY-NC-ND 4.0 license. Captions below are the source captions. Figures show study findings, not instructions for unsupervised treatment.

Figure 1: Experimental flowchart — *Figure 1.* Experimental flowchart

Figure 2: Illustration of combined central-peripheral magnetic protocol. ( A,C ) rTMS at ipsilesional pharyngeal M1, ( B, D ) rPMS targeting suprahyoid muslces, ( E ) the stimulation parameters for both rTMS and rPMS — *Figure 2.* Illustration of combined central-peripheral magnetic protocol. ( A,C ) rTMS at ipsilesional pharyngeal M1, ( B, D ) rPMS targeting suprahyoid muslces, ( E ) the stimulation parameters for both rTMS and rPMS

Figure 3: Illustration of temporal sequence of swallowing — *Figure 3.* Illustration of temporal sequence of swallowing

Figure 4: Illustration of magnitude of SHE and AHE. ( A ) the images frame from resting state, ( B ) the tilted image from image A. ( C ) the images frame from the peak displacement state, ( D ) the tilted image from image C — *Figure 4.* Illustration of magnitude of SHE and AHE. ( A ) the images frame from resting state, ( B ) the tilted image from image A. ( C ) the images frame from the peak displacement state, ( D ) the tilted image from image C

Figure 5: fNIRS channel arrangements. Abbreviations: DLPFC: dorsolateral prefrontal cortex, FPA: frontopolar area, S1: primary somatosensory cortex, M1: primary motor cortex, PMC&SMA: premotor and supplementary motor area, Red cross illustrates the rTMS stimulation point in the ipsilesional hemisphere — *Figure 5.* fNIRS channel arrangements. Abbreviations: DLPFC: dorsolateral prefrontal cortex, FPA: frontopolar area, S1: primary somatosensory cortex, M1: primary motor cortex, PMC&SMA: premotor and supplementary motor area, Red cross illustrates the rTMS stimulation point in the ipsilesional hemisphere

Figure 6: Analysis of FOIS and PAS scores across different time points. ( A ) all the four groups demonstrate improvements in the FOIS over time at T1 and T2, compared with T0. ( B ) all the four groups showed decreases in the PAS over time at T1, compared with T0 — *Figure 6.* Analysis of FOIS and PAS scores across different time points. ( A ) all the four groups demonstrate improvements in the FOIS over time at T1 and T2, compared with T0. ( B ) all the four groups showed decreases in the PAS over time at T1, compared with T0

Figure 7: Temporal sequences of swallowing and kinematic parameters derived from VFSS. **p < 0.01. UES: upper esophageal sphincter, PC: pharyngeal constriction — *Figure 7.* Temporal sequences of swallowing and kinematic parameters derived from VFSS. **p < 0.01. UES: upper esophageal sphincter, PC: pharyngeal constriction

Figure 8: Comparison of swallowing-related brain activity at T0 and T1 in four groups — *Figure 8.* Comparison of swallowing-related brain activity at T0 and T1 in four groups

Figure 9: Heat map of brain connectivity at T0 and T1 in four groups — *Figure 9.* Heat map of brain connectivity at T0 and T1 in four groups

Figure 10: FC change in combined stimulation group. Blue lines indicate p < 0.05, red lines indicate p < 0.01 — *Figure 10.* FC change in combined stimulation group. Blue lines indicate p < 0.05, red lines indicate p < 0.01

Figure 11: Correlation analyses between ΔFC and ΔLCD ( A ), and between ΔFC and ΔSHE ( B ) — *Figure 11.* Correlation analyses between ΔFC and ΔLCD ( A ), and between ΔFC and ΔSHE ( B )

Criticisms

Small and below the planned enrollment: 44 randomized versus 48 planned, with only 10–11 analyzed per arm. Large-looking effects require replication rather than extrapolation to an unrestricted stroke population.
Not full intention-to-treat: Excluding the participant who discontinued because of pain can bias the efficacy estimate and detach tolerability from the population actually randomized.
Combination is not automatically synergy: A significant four-arm group-by-time test does not establish a central-by-peripheral treatment interaction. The strongest practical comparator is central stimulation alone; a clear adjusted incremental benefit over that arm is not demonstrated in the reported pairwise estimates.
Masking uncertainty: Coil tilt reproduces sound, not necessarily the sensations of active treatment. Transient twitching occurred only in active-treatment arms, which could reveal allocation. The study reports no masking-success assessment.
Short and selective endpoints: Later follow-up was FOIS only. Better PAS immediately after treatment is not evidence of fewer aspiration pneumonias, and an average FOIS shift does not specify how many individuals became fully orally fed.
Internal reporting problems: The within-arm FOIS paragraph contains inconsistent signs, including an interval for combined treatment at follow-up that crosses zero despite a stated P < 0.001. The VFSS paragraph repeats the same LCD estimate and interval for two arms. These should be checked against source data, not silently corrected or treated as independent confirmation.
Mechanistic overreach: Small within-arm connectivity correlations, many brain-region comparisons and incompletely described multiplicity control cannot establish mediation. Right-lesion fNIRS channels were flipped to the left for pooling, potentially masking laterality differences.
External validity: First-ever subcortical strokes in cooperative patients without aphasia or cognitive impairment are not a proxy for all dysphagic stroke patients, particularly acute or brainstem presentations. A custom peripheral coil and nonindividualized scalp targeting also complicate replication.

Funding

Supported by the Guangzhou Municipal Science and Technology Bureau Basic Research Project (High-Level University) Joint Funding, grant 2025A03J3290; the 2023 Guangzhou Teaching Achievement Cultivation Project, achievement 2023128383; and the 2025 Guangzhou Education Science Planning Project, grant 2024111264, to Zirui Luo. The authors state that the funder had no further role in design, collection, analysis, interpretation, writing or the submission decision. No competing interests were declared. Registration: ChiCTR2500113782, reported registration date December 3, 2025.

The paper

Authors. Zirui Luo et al.
Title. Efficacy and Mechanisms of Combined Central-Peripheral Magnetic Stimulation on Prefrontal-Sensorimotor Cortex Connectivity in Post-Stroke Dysphagia after Supratentorial Stroke: A Randomized, Single-Blind, Parallel-Controlled Trial.
Journal. Translational Stroke Research.
Year. 2026; volume 17, issue 4.
DOI. 10.1007/s12975-026-01482-1
PMCID. PMC13400478
PMID. 42496873

Deep Dive — click to expand

What this is

[F] This four-arm 2026 experiment asks whether stimulating both the cortical swallowing network and the peripheral suprahyoid region improves recovery beyond a shared rehabilitation program. The combined arm outperformed double sham on FOIS by 1.09 points immediately and 1.45 points at follow-up, while only 43 participants contributed to analysis. [I] The useful interpretation is an encouraging efficacy signal that generates a sharper comparative trial—not a completed demonstration of synergy or clinical complication prevention. Throughout this commentary, [F] marks a reported fact, [I] an inference, [H] a testable hypothesis and [M] missing evidence.

1. Shadow Audit

[F] Every arm improved, and every arm received conventional swallowing therapy; the clearest reported FOIS contrasts compare combined stimulation with double sham, not with active central stimulation. [I] The headline therefore obscures the purchasing decision: whether a second modality adds enough benefit to justify its coil, staff time and tolerability burden. [M] The decisive adjusted combined-versus-central-only estimate, its confidence interval and the proportion reaching meaningful oral-intake milestones are not supplied as a complete decision-ready comparison.

[I] The cheapest discriminating test is a prespecified reanalysis of the randomized central-active arms, retaining the discontinued participant through a justified missing-data strategy and reporting the incremental FOIS/PAS contrasts. A near-zero incremental effect would favor the simpler protocol; a credible, clinically important advantage would justify a larger add-on trial.

2. Inversion Engine

[F] The lower confidence bound for combined-versus-double-sham FOIS is +0.39 immediately after treatment and +1.00 at follow-up. [I] As a simple sensitivity benchmark—not a formal bias correction—an unaccounted favorable shift of 0.39 or 1.00 FOIS point, respectively, would bring those lower bounds to zero. This quantifies how much interpretation can change without claiming that such bias actually occurred.

[F] One randomized patient was excluded after pain, and active-arm twitching was absent from double sham. [H] Differential expectations or outcome missingness could inflate the apparent advantage. A masked rescore of the existing VFSS videos, combined with an explicit sensitivity analysis for the excluded participant, would help distinguish physiological benefit from assessment or attrition effects. [I] Even if combined treatment beats double sham, its opposite practical conclusion—no need to add peripheral stimulation—can still hold if the incremental effect over central-only treatment is negligible.

3. Second-Order Catalyst

[F] This protocol requires cortical stimulation, a custom peripheral coil and 30 minutes of conventional therapy, repeated across a two-week course. [I] The first institutions able to test it are specialist rehabilitation centers already equipped for rTMS and instrumental swallowing assessment, not an acute stroke unit simply adding an order to its admission set.

[H] If the effect replicates, workflow capacity and patient attendance may become more limiting than device availability. A short, reversible implementation study should record screening eligibility, completed visits, staff minutes and discomfort before expanding referrals. [M] Cost, attendance burden and pneumonia or feeding-tube outcomes are not supplied. [I] Those missing downstream measures determine whether better scale scores translate into a service worth scaling.

4. Asymmetric Leverage

[F] Participants had very low baseline FOIS scores, but the study reports mean score differences rather than the number crossing a clinically actionable feeding threshold. [I] The highest-return next step is not a larger imaging panel: it is a patient-level transition table showing who moved from restricted or tube-supported intake to meaningful oral intake, alongside blinded safety assessment and the central-only comparison.

[H] A modest average improvement could be valuable if concentrated in patients who attain a major feeding milestone; the same mean improvement could have little practical effect if nobody changes management. Reporting those existing-data transitions is cheap relative to buying another stimulation system. [I] The clinical denominator is eligible patients achieving a durable milestone, not the number of significant cortical connections.

5. Paradigm Destroyer

[F] The randomized comparison supports a swallowing-score signal, while the proposed sensorimotor mechanism rests on within-arm connectivity changes and correlations in 11 combined-treatment patients. [I] The reflex this paper should challenge is “chronic dysphagia means recovery is finished,” not “a plausible circuit diagram proves a treatment works.” Conversely, the strongest criticism is that the trial has not yet earned its preferred story of synergistic central-peripheral repair.

[I] A proportionate two-sentence protocol update would be: “Offer patients with persistent post-stroke dysphagia a structured reassessment rather than assuming no further improvement is possible. Consider combined magnetic stimulation only within a specialist, evidence-generating pathway, without replacing established swallowing rehabilitation or advancing diet without clinical reassessment.” [H] A multicenter, masked add-on trial against active central stimulation, with sustained oral-intake and pneumonia outcomes, would determine whether the additional modality deserves routine adoption.

MVP — Minimum Viable Proof

[F] The minimum defensible numerical claim today is +1.45 FOIS points at 30-day follow-up versus double sham plus rehabilitation, 95% CI +1.00 to +1.91, in this selected, small completer cohort. [I] The missing practice-changing proof is a reproducible, clinically meaningful incremental advantage over central stimulation alone, accompanied by durable feeding benefit and acceptable harms. It would be inappropriate to convert this ordinal mean difference into a number needed to treat.

Best Combination

[F] In its discussion, the paper describes prior combined-stimulation studies and notes that the relative ranking of central-only and peripheral-only treatment differed from an earlier trial. [I] Treat that discussion as context reported by these authors, not as an independently verified synthesis or a guideline recommendation. The appropriate sequence is Shadow Audit → Inversion Engine: identify the missing add-on comparison, then test whether it survives attrition and masking sensitivity analyses. [M] A verified synthesis of directly comparable protocols and a larger multicenter replication remain necessary before drawing a general treatment hierarchy.

Overvalue Warning

[I] Do not read “best-performing arm” as proof of synergy or superiority over every active comparator. The reported group-by-time interaction answers a different question.
[I] Do not turn short-term PAS/FOIS improvements or fNIRS correlations into claims of fewer pneumonias, feeding-tube removal, durable independence or demonstrated causal circuit repair. Those endpoints and mechanisms were not established here.

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.