Subjects were admitted to the rehabilitation medicine department, Liaocheng People’s Hospital, Shandong, China. Inclusion criteria were as follows: (1) computerized tomography (CT) or magnetic resonance imaging (MRI) evidence of a first-ever occurrence of unilateral cerebral infarction or haemorrhage [17]; (2) Age 18–80 years; (3) Stroke onset 3 months [18]; (4) Subjects with swallowing difficulties, as well as choking, coughing, wet voice, or nasal regurgitation after swallowing, that persisted for more than two weeks following a stroke [19]; (5) Stable conscious subjects who were able to complete all examinations and treatments; (6) Videofluoroscopic swallowing study (VFSS) confirmed dysphagia. Exclusion criteria included: (1) Insertion of a cardiac pacemaker or metallic implants in the brain; (2) Seizure history; (3) Swallowing disturbances due to subarachnoid hemorrhage, venous infarctions, and other nervous system diseases. This study was approved by the Ethical Committee (XJS2022250), and each subject signed informed consent.

According to the integrity of CBT, 35 patients with high CBT integrity (relative fractional anisotropy (rFA) $>$0.5) were enrolled and assigned to the high CBT integrity group. Correspondingly, 32 patients with low CBT integrity (rFA 0.5) were enrolled and assigned to the low CBT integrity group. Patients within each of the two main groups were further divided into three subgroups: 5-Hz HF group, 1-Hz LF group, and Sham group, to give a total of six groups. Figs. 1,2 presents the research design and flowcharts.

Fig. 1.

Patients with high CBT integrity flow chart.

Fig. 2.

Patients with low CBT integrity flow chart.

DTI examination was performed at the beginning of the study for all participants. DTI was performed with an 8-channel head coil on a 3.0 T magnetic resonance imager (Ingenia 3.0, Phillips, Amsterdam, Netherlands) before grouping. Imaging parameters were as follows: TR/TE = 4282/100 ms, matrix = 128 $\times$ 128; field of view = 224 $\times$ 224 mm${}^2$; flip angle = 90°; voxel size = 0.9 $\times$ 0.9 $\times$ 0.9 mm${}^3$; 60 contiguous slices applying the 32 direction diffusion sensitizing gradients ($b$ value = 1000 s/mm${}^2$) to obtain the diffusion-weighted images and a single unweighted b0 image ($b$ value = 0 s/mm${}^2$) [15]. MRI images were visually inspected and corrected for image distortions and motion artifacts using eddy current correction tools.

IntelliSpace Portal 9.0 (Phillips, Amsterdam, Netherlands) was employed for fiber tracking and DTT data analysis. The seed region of interest for CBT tracking was placed on the inferior pons. The target region of interest was placed on the lower part of the precentral gyrus, which was located on the top of the lateral ventricles [16]. The CBT’s fractional anisotropy (FA) was evaluated bilaterally. FA is a metric which ranges from 0 to 1 for quantitatively accessing the integrity of CBT, and a decreased FA on the CBT is related to impaired nerve conduction [20]. The FA ratio between the affected and unaffected sides was defined as relative FA (rFA, rFA = FA${}_{\text{ipsilesional}}$/FA${}_{\text{contralesional}}$). An rFA $>$0.5 was defined as high CBT integrity, whereas an rFA $\le$0.5 was defined as low CBT integrity [21] (Figs. 3,4).

Fig. 3.

Patient exemplifying high CBT integrity: A 48-year-old male. (A) Left corona-radiating infarct was detected by MRI flair. (B) Color-coded FA map, whole-brain fiber tracking. (C) DTT diagram shows that the number of CBT fibers in the affected hemisphere decreased slightly (rFA = 0.92).

Fig. 4.

Patient exemplifying low CBT integrity: A 65-year-old male. (A) An infarct in the right cerebral hemisphere was detected by MRI flair. (B) Color-coded FA map, whole-brain fiber tracking. (C) DTT diagram shows the disruption of CBT in the affected cerebral hemisphere (rFA = 0.07).

VFSS is a superior quality method for diagnosis and quantitative investigation of dysphagia because of its ability to visualize the mouth cavity, pharynx cavity, laryngeal cavity, and upper esophagus during deglutition [22]. VFSS examination was performed before and after treatment. The lateral and posteroanterior images were obtained using VFSS after subjects ingested a 5 mL thick liquid bolus with standardized viscosity and composition.

rFA values reflect neural fiber integrity. rFA values are taken in the range of 0 to 1. A value closer to 1 indicates that the neural fiber structure on the affected hemisphere is more intact.

A 7-cm Magnetic Stimulator (CCY-I, Yiruide, Wuhan, China) was used to deliver rTMS. Surface electrodes were used to obtain the electromyogram of the mylohyoid muscle. The coil stimulator was placed tangentially at the apex of the skull, then moved 3 cm anteriorly and 5 cm laterally. While moving around in this region, single pulse stimulations were delivered until the maximum motor-evoked potentials of the contralesional hemisphere were obtained [18]. The pulse intensity was subsequently decreased in 2% steps of stimulator output in this region. The minimal stimulation intensity to elicit motor-evoked potential with peak-to-peak amplitude greater than 50 uV in at least 5 of 10 consecutive trials was defined as the resting motor threshold (RMT) [25].

Each participant received rTMS treatment five days per week for two weeks (Fig. 5). Parameters for patients receiving high-frequency stimulation are as follows: 5 Hz, 100% RMT, 2 s stim-10 s rest, with 950 pulses over the contralesional mylohyoid motor cortex for 19 min. Parameters for patients receiving low-frequency stimulation are as follows: 1 Hz, 100% RMT, 15 s stim-2 s rest, with 1005 pulses over the contralesional mylohyoid motor cortex for 19 min. The parameters for patients receiving sham stimulation were the same as the HF group (frequency, time, and noise), but no current was produced in the brain. All patients received the same conventional swallowing treatment before rTMS treatment.

Fig. 5.

Subject undergoing rTMS therapy over the contralateral hemisphere.

Data collected in this study were analyzed using the SPSS 26.0 software (IBM Corp., Chicago, IL, USA) package. The Kolmogorov-Smirnov test was used to ensure the parameters were normally distributed. Quantitative variables are expressed as means $\pm$ standard deviation (SD). Before and after therapy, intra-group comparisons were analyzed using paired t-tests. An Independent t-test was used for comparisons between two groups, and a one-way ANOVA was employed for multiple-group comparisons. A p-value 0.05 was considered statistically significant.

There were 67 participants recruited in this study. An aggregate of 61 patients completed the study and was included in the final statistical analyses. No significant differences in sex distribution, age, duration of onset of stroke, stroke location, type of stroke, National Institute of Health Stroke Scale (NIHSS), Pneumonia, rFA, and clinical characteristics distribution at baseline among the three subgroups of the same CBT integrity (p $>$ 0.05) (Tables 1,2). Changes in SSA, PAS, and DOSS scores at baseline and post-treatment in each subgroup of patients with high and low integrity of CBT are presented separately in Figs. 6,7.

Fig. 6.

Changes in baseline and post-treatment swallowing function scores in subjects with high CBT integrity after rTMS stimulation. HF group, 5 Hz-high frequency rTMS group; LF group, 1 Hz-low frequency rTMS group; Sham group, Sham stimulation group; SSA, standardized swallowing assessment; PAS, penetration aspiration scale; DOSS, dysphagia outcome and severity scale. *p 0.05.

Fig. 7.

Changes in baseline and post-treatment swallowing function scores in subjects with low CBT integrity after rTMS stimulation. HF group, 5 Hz-high frequency rTMS group; LF group, 1 Hz-low frequency rTMS group; Sham group, Sham stimulation group; SSA, standardized swallowing assessment; PAS, penetration aspiration scale; DOSS, dysphagia outcome and severity scale. *p 0.05.

Patients with high CBT integrity improved significantly in three subgroups in the SSA scores after therapy (p 0.05). The further inter-group comparison found that the HF and LF groups had better SSA scores than the Sham group (p 0.05), but the HF and LF groups did not differ significantly (p $>$ 0.05). Meanwhile, significant post-treatment improvements in SSA scores were seen in three subgroups of participants with low CBT integrity (p 0.05), with the HF group performing significantly better than the LF and Sham groups (p 0.05), but there were no appreciable changes between the LF and Sham groups statistically.

Regarding patients with high CBT integrity, PAS scores decreased significantly (p 0.05) from pre-treatment to post-treatment for all three subgroups. Multiple comparisons revealed that the decline was more pronounced in the HF and LF groups than in the Sham group (p 0.05), but there were no significant differences between the HF and LF groups (p $>$ 0.05). For patients with low CBT integrity, the HF group exhibited a significant decrease in PAS scores from baseline to post-treatment (p 0.05) while also showing significant differences compared to the LF and Sham groups (p 0.05), but the changes were not statistically significant between the LF and Sham groups.

Similar results were observed in patients with high CBT integrity DOSS scores. After two weeks of therapy, the DOSS scores of all three subgroups with high CBT integrity significantly improved compared to pre-treatment (p 0.05). When compared to the Sham group, the HF and LF groups’ post-treatment DOSS scores improved considerably (p 0.05), but there was no difference between the HF group and the LF group (p $\mathrm{>}$ 0.05). For patients with low CBT integrity, DOSS scores significantly increased in all three subgroups after therapy (p 0.05). Further comparisons revealed that the HF group outperformed the LF group and the Sham group (p 0.05), whereas there were no appreciable differences between the LF group and the Sham group (p $>$ 0.05).

No serious adverse effects, such as seizures or recurrent strokes, were seen throughout all treatments. During the first few days, two subjects who received high-frequency stimulation experienced dizziness, and two who received low-frequency stimulation experienced transitory headaches. The symptoms readily disappeared without any particular therapy.

In the case of high CBT integrity patients, the rFA values of HF, LF, and control groups were reduced after two weeks of treatment. Differences were found in the rFA value at post-treatment between the HF and LF groups. Similarly, in the case of low CBT integrity patients, rFA values were lower than pre-treatment in all three groups after two weeks of treatment, there was a statistical difference in rFA values between the HF group and LF group after treatment (Fig. 8). This finding indicates a relative reduction in the integrity of CBT fibers in the affected hemisphere during stroke recovery and that changes in CBT fibers varied at different frequencies of rTMS.

Fig. 8.

Baseline and post-treatment rFA values in subjects with high CBT integrity (a) and low CBT integrity (b) after rTMS stimulation. *p 0.05.

There are some limitations in the present study. Firstly, the sample size of each subgroup was small. Secondly, the swallowing function at post-treatment was evaluated, but a long-term follow-up was not performed. Thirdly, for patients with high CBT integrity, the baseline NIHSS score of the HF group and LF group were lower than 11.5, while the baseline NIHSS score of the Sham group was equal to 11.5. This could introduce a bias since NIHSS score lower than 11.5 was associated with transient dysphagia, while NIHSS score equal to or superior to 11.5 was related to persistent dysphagia [35]. Future studies should combine DTI with functional Magnetic Resonance Imaging (fMRI) to further explore the mechanism of swallowing function recovery after stroke.