We recruited patients with acute stroke of the dorsolateral medulla who met the following criteria and were hospitalized at the Department of Neurology of Tianjin Huanhu Hospital between January 2022 and June 2022. The inclusion criteria were as follows: (i) first-time stroke with confirmed lesions involving the dorsolateral medulla on MRI within 48–72 h of onset; (ii) age between 30 and 75 years; (iii) no history of swallowing disorders before stroke but presenting with dysphagia after stroke, with a score of 3–4 on the Kubota Water Swallowing Test; (iv) clear consciousness, a Mini-Mental State Examination (MMSE) score >24, stable vital signs, and ability to cooperate with the functional examination and swallowing rehabilitation training confirmed by simulated measurements; and (v) voluntarily participated in the study and provided written informed consent.

The exclusion criteria were as follows: (i) dysphagia caused by other neurological diseases or organic lesions; (ii) concomitant cerebral infarction in other brain regions, including the cortex, thalamus, and cerebellum; (iii) patients with metal or other implants who could not undergo 3.0 T MRI; (iv) patients with lesions in the oral cavity, nasal cavity, pharynx, and larynx who could not undergo FEES; and (v) patients with obvious hiccups or severe dysphagia who could not complete the supine swallowing action and those with unstable conditions who were not able to undergo fMRI.

Fifteen patients with an age range of 35–75 years and average age of 54 years were included in the study, and three patients were female. Based on the Edinburgh Handedness Inventory (7), all 15 patients were all right handed before stroke onset (Table 1).

This study was conducted in accordance with the Declaration of Helsinki. The studies involving human participants were reviewed and approved by the Ethics Committee of Tianjin Huanhu Hospital, with approval number (Jinhuan) No. (2020-55). The patients provided their written informed consent to participate in this study.

FEES and BOLD scanning were performed after enrollment, and swallowing rehabilitation training was provided 5 days a week. Swallowing function evaluation, FEES, and BOLD scanning were performed again after 4 weeks of treatment.

A 3.0 T MR scanner (Magnetom Skyra, Siemens, Germany) with a 20-channel head coil was used to obtain functional and structural images. Patients were positioned supine on the scanner bed, with their heads fixed in a birdcage-shaped coil, and they viewed the experimental tasks on a screen through a mirror mounted on the head coil. Foam earplugs and pads were used to reduce scanner noise and head motion, respectively. High-resolution T1-weighted structural imaging, functional imaging, and diffusion-weighted imaging (DWI) were performed sequentially, with the scanning plane parallel to the anterior-posterior commissure line. The structural (T1) image parameters were as follows: TR, 2,000 ms; TE, 2.98 ms; matrix, 256 × 256; field of view, 256 × 256 mm; slice thickness, 1 mm; and 192 slices. DWI was performed using the following parameters: TR, 5,200 ms; TE, 80 ms; field of view, 240 mm × 240 mm; 36 slices; and slice thickness, 4 mm. Functional imaging was performed using the magnetic-sensitive GRE-EPI BOLD contrast T2^*WI imaging sequence, with the following parameters: TR, 2,000 ms; TE, 30 ms; excitation time, 1; flip angle 90°; field of view, 220 mm × 220 mm; matrix, 64 × 64; 36 slices; slice thickness, 4 mm; no slice gap; and imaging time, 370 s; a total of 90 dynamics were acquired to cover the entire frontal lobe cortex to the level of the medulla oblongata.

BOLD paradigm is shown in Figure 1. An initial blank resting module of 18 s was set to eliminate the influence of blood flow signal saturation and the magnetic environment. A final blank module of 10 s was set to avoid the final effect of nuclear magnetism. The task employed a BLOCK experimental design, with saliva swallowing and effortful saliva swallowing presented in a pseudorandom manner through visual text prompts. The participants were instructed to perform as many swallowing motions as possible during the swallowing cue period. During the black screen period, the participants rested quietly without any other movements except for breathing. Figure 1 presents the specific task sequence, with each set of swallowing actions lasting for 18 s, followed by an 18 s rest period for a total of 10 sets. The final 10 s blank module was also included, resulting in a total time of 370 s. During BOLD scanning, the experimenters accompanied the participants to the MRI room and observed their swallowing actions to ensure that they followed the visual cue requirements for swallowing and resting.

All fMRI data were preprocessed using SPM 12 (Statistical Parametric Mapping, Wellcome Department of Imaging Neuroscience, London, UK) running on MATLAB (MathWorks, Natick, MA). The first 10 time points were removed to eliminate the possibility of unstable AMRI signals. If a patient's motion and rotation parameters exceeded 1.5 mm and 1.5°, respectively, this run of data was excluded from future analysis. The images were subsequently spatially normalized to the Montreal Neurological Institute template brain, resampling voxel size = 3 × 3 × 3 mm³. Functional images were spatially smoothed with a three-dimensional Gaussian kernel of 6 mm full width at half maximum to increase the signal-to-noise ratio and reduce inter-subject differences.

The experimental data were organized into a database using Excel, and the SPSS statistical software package version 20.0 (SPSS Inc., Chicago, IL, USA) was used to analyze the activation volume and intensity of the brain functional areas. The General Linear Model (GLM) was used in the first-level analysis.The activated brain regions and their voxels in each participant were tested using one sample t-test, To observe the activation in each action, we defined saliva swallowing vs. rest period as con1, effortful saliva swallowing vs. rest period as con2, effortful saliva swallowing minus saliva swallowing as con3. In the second-level analysis, the statistical analysis, the beta weights of two contrasts were statistically compared separately using one sample t-test, with the age and sex of participants as covariate. The results were corrected for multiple comparisons using the false discovery rate (FDR) correction at the voxel-level, and the significance threshold was set at FDR-corrected P < 0.01 with a minimum cluster size (the number of voxels) of 10 voxels.

Table 1 provides an overview of the clinical and demographical characteristics of our patients. All lesions were located on the left or right side of the dorsolateral medulla and were confirmed using MRI (Figure 2). After enrollment, DWI scans were conducted along with two BOLD scans before and after rehabilitation, and no new lesions were observed in comparison to the DWI images prior to inclusion in the study. The score of Kubota Water Swallowing Test after rehabilitation training (1.267 ± 0.4577) was significant lower than that at enrollment (3.400 ± 0.5071; P < 0.001, Table 1).

FEES was performed after enrollment (2–3 days after the onset of stroke), revealing impaired pharyngeal swallowing function, indicated by vocal cord paralysis, valleculae and pyriform fossa retention of a saliva-like substance, absence of swallowing reflex upon touching the epiglottis cartilage and posterior pharyngeal wall with a touch stick, a delayed swallowing response, residual vallecular material observed when swallowing viscous food and bread, and leakage and aspiration in some patients. The residual material was removed after repeated swallowing.

After rehabilitation training for 4 weeks, five patients declined follow-up FEES due to their perception of swallowing without any difficulty, and 10 patients underwent a follow-up FEES. Examination of the 10 patients revealed significantly improved pharyngeal swallowing function; there was no vocal cord paralysis and no obvious retention of saliva-like substances in the valleculae or pyriform fossa. In some patients, the swallowing reflex was still delayed upon touching the epiglottis cartilage and posterior pharyngeal wall with a touch stick. Notably, leakage or aspiration was not observed.

In this study, very few brain regions were activated during saliva swallowing in the acute phase of stroke. Specifically, only the bilateral precentral gyri and left lingual gyrus were partially activated. This is similar to the findings of previous studies (9), and the right inferior semilunar lobule and right posterior lobe of the cerebellum were slightly activated. The precentral gyrus is the activated cortex for voluntary movement, whereas the lingual and inferior occipital gyri form a visual association cortex that may have been associated with the visual swallowing cues in this study. The inferior semilunar lobule, a new lobe of the cerebellum located in the posterior lobe of the cerebellum, receives cortical pontine fibers and exhibits very little activation, which may be related to the coordinated movement of swallowing actions. Although the patients only had a bulbar injury, the small number of activated brain areas indicates that the bulbar region is an important component of the swallowing network (10, 11) and that it influences the related brain regions of the swallowing network. Notably, there was a lack of general cortical activation during swallowing.

Currently, only a few studies have examined brain activation patterns in patients with post-stroke dysphagia. Some studies have shown a significant decrease in the cortical activation areas (9, 12, 13); however, Li et al. reported increased activation in swallowing-related regions, including the left precentral gyrus, left postcentral gyrus, left SMA, and insula (14). Further, Zhou et al. observed a significant reduction in the activation of the anterior cingulate gyrus and excessive activation in other non-swallowing-related regions, including the posterior cingulate gyrus, parahippocampal cortex, visual center, and primary auditory center (15). These inconsistent results may be due to differences between the selected participants. For example, mild-to-moderate dysphagia caused by unilateral hemisphere stroke can produce compensation in the unaffected side or other brain regions (14, 15). However, in the present study, all the patients experienced medullary damage, which is a crucial component of the swallowing network. The medulla participants pharyngeal swallowing and requires bilateral coordination, making functional compensation challenging. Consequently, most cerebral regions cannot be activated, and severe dysphagia is clinically observed. In this study, the patients were in the acute phase of stroke, and cortical activation was scarce, similar to patients with brainstem damage 1–12 months after stroke (9). It is speculated that the activation of the cortex during swallowing in patients with dysphagia is more related to the damaged region and degree than to the post-injury period. If the brainstem damage does not recover, most cortices will exhibit decreased activation regardless of the time after injury.

To date, the conclusions of various studies regarding insular activation in patients with dysphagia have been inconsistent. Patients with unilateral cortical and subcortical infarctions and mild dysphagia exhibited bilateral insular activation (14, 15) with increased activation of the undamaged insula compared with patients in the control group (14), whereas other studies reported no insular activation (9, 12, 13). In this study, all patients had pharyngeal dysphagia, and no insular activation was observed in the acute phase; however, bilateral insular activation was observed after recovery. This finding suggests that the impairment of insular activation may be related to pharyngeal dysphagia. The insula contributes to the initiation of swallowing by processing sensory elements and interacting with widespread brain regions in both hemispheres, including the primary sensorimotor cortex (16). The frontal insula is believed to regulate the temporal organization of mouth movements, such as the timing of the onset of swallowing after a series of chewing or other mouth movements (5). The absence of insular activation may affect the patient's ability to plan swallowing movements, leading to uncoordinated or abnormal swallowing completion, which is manifested as a delay in the start of pharyngeal swallowing. Activation of the insula maybe a compensatory mechanism for the recovery of swallowing function.

Only a few studies have compared the activated brain areas during swallowing before and after rehabilitation. Wei et al. found that the volume of activation in the affected hemisphere increased significantly after rehabilitation, whereas there was no change or only a slight increase in the unaffected hemisphere. Therefore, the recovery of cerebral hemisphere function on the affected side is important for the rehabilitation of patients with swallowing disorders (13). In contrast, Li et al. observed an increase in activation in the contralateral hemisphere during the acute phase of stroke and suggested that compensation from the intact side promoted functional recovery (14). These inconsistent results may be attributed to the heterogeneity of the enrolled patients. To increase accuracy, it is advisable to select relatively consistent patients, such as those with unilateral involvement of a certain cortical or subcortical structure.

After language training, patients with dysphagia caused by unilateral hemisphere stroke exhibited significant activation of the bilateral primary motor areas, sensory areas, premotor areas, and the anterior insula during swallowing (6). In patients with brainstem damage, increased activation of brain regions, including the cingulate gyrus, insula, precuneus, frontal lobe, orbitofrontal cortex, and SMAs, was observed after treatment. These findings suggest that other brain regions have some compensatory ability when a swallowing center injury occurs (9). Our study demonstrated the bilateral precentral gyrus, postcentral gyrus, insula, thalamus, and SMA showed increased activation after rehabilitation training. These regions are not completely consistent with the above results (9), which may be attributed to different post-injury periods and different compensatory mechanisms between the early and chronic periods.

Mihai et al. found that even after swallowing function recovery, the overall activation level of swallowing-related brain regions in patients was still lower than that of the control group (17). Meanwhile, there was increased activation in the contralateral S1 area; the more obvious the initial damage, the more obvious the activation in the contralateral S1 area after functional recovery. This indicates that activation of the contralateral S1 area is the reason for the recovery of swallowing function. Domin et al. revealed that the integrity of callosal fibers interconnecting the S1 swallowing representation were significantly associated with swallowing compliance (18). An increase in activation was observed in the postcentral gyrus in this study, which is similar to previous results (17, 18). This suggests that after the motor function is impaired, increased somatosensory input is required to complete swallowing tasks; that is, sensory reorganization contributes to compensation.

Oral sensory stimulation can activate the sensory and motor areas associated with swallowing, which can enhance sensory input and participate in the control center of the brainstem and cortex, potentially affecting the autonomous components of swallowing (19). Studies have revealed that pharyngeal stimulation first activates the sensory-motor cortex, then excites the swallowing motor cortex and changes the recruitment pattern of cortical activation related to swallowing (20, 21). In this study, all patients had pharyngeal dysphagia, and FEES revealed decreased sensation in the pharynx. When the probe touched the epiglottis and posterior pharyngeal wall, it did not elicit a swallowing reflex, causing a delay in the swallowing response; this is an important cause of choking. The patients in this study underwent ice-cotton swab stimulation training, which improved pharyngeal sensation and increased sensory input to the postcentral gyrus, possibly by activating the thalamus. Sensory compensation contributed to clinical recovery, as confirmed using FEES, which showed improved pharyngeal sensation after recovery. This finding suggests that the tactile input during swallowing may be a crucial factor in recovery. In the treatment of dysphagia, especially pharyngeal dysphagia, it is necessary to focus not only on motor training but also on sensory stimulation training to better promote swallowing function.

After treatment, there was only a slight increase in activation in the left SMA, right supramarginal gyrus, and left postcentral gyrus when effortful saliva swallowing was compared with saliva swallowing. Peck et al. found that healthy individuals exhibited increased activation in the angular gyrus, cingulate gyrus, inferior parietal lobule, middle frontal gyrus, superior frontal gyrus, and supramarginal gyrus when comparing effortful saliva swallowing to saliva swallowing (22). The results of the present study were similar, although the range of differences in activation was smaller. However, before treatment, a significant increase was observed in brain activation between the two swallowing actions, including the bilateral SMA, bilateral postcentral gyri, and right insula. This finding suggests that patients with acute medullary stroke need to exert more effort to complete effortful saliva swallowing and activate more brain regions, which are the compensatory areas activated when normal saliva swallowing is performed after recovery.

Gandhi et al. (23) found that effortful swallowing can improve swallowing function in patients with PD more effectively than normal swallowing, whereas Kim et al. (24) found that tongue-to-palate training can increase the oral-pharyngeal swallowing function of subacute stroke patients. The results of these two studies are consistent with our findings. Additionally, Bahia et al. reviewed 23 studies and found that the biomechanical effects of effortful swallowing include increased pressure in the oral, pharyngeal, and esophageal regions (25). The BOLD imaging results confirmed that the recovery of swallowing function after stroke can be attributed to the compensation of more brain regions; however, the results cannot confirm that rehabilitation training itself stimulates the compensatory brain regions. Here, we observed that effortful swallowing after MI can activate these compensatory brain regions and promote faster recovery of swallowing function. This indicates that in pharyngeal dysphagia caused by MI, effortful saliva swallowing can activate more brain regions and better promote the recovery of swallowing function, making it a more suitable rehabilitation training method than saliva swallowing.