In the present study, patients in the early subacute stage after stroke were measured by means of whole-head MEG. Our data revealed distinct patterns of cortical swallowing activation, which depended on the region affected by stroke. While hemispheric stroke in the early subacute phase associated with dysphagia resulted in decreased ipsilesional cortical activation and a nearly extinguished activation of the contralesional hemisphere, in contrast patients with hemispheric strokes without dysphagia were characterized by bilateral activation comparable to healthy controls. Additionally, all groups of patients with hemispheric stroke showed prefrontal activation, which was neither seen in healthy subjects nor in patients after brainstem stroke. In brainstem stroke a strong right hemispheric lateralization was found in both groups (with and without dysphagia).
Localization of cortical activation
Cortical beta ERD in patients as well as control subjects were observed in the bilateral superior somatosensory areas. These activated cortical areas are located more superior and medial with respect to what has to be expected based on the homunculus. Additionally the swallowing related cortical activation is spread extensively in primary and secondary motor and sensory areas in both hemispheres, found in healthy controls and non-dysphagic patients with hemispheric stroke. This is a well-known phenomenon in functional brain imaging of human swallowing processing, which has been observed in several former MEG studies by our group [10, 15–17, 30, 46]. But also other studies using TMS, PET and MEG demonstrated similar patterns of activation [8, 12, 47]. As an explanation for this widespread cortical activation in swallowing processing projections to and from the swallowing tract which is located in rostral non-primary motor areas, are suggested by Hamdy and co-workers . The swallowing network therefore extends beyond the primary sensorimotor cortex and involves secondary areas.
Hemispheric stroke with dysphagia
Irrespective of the affected hemisphere, a strongly reduced cortical activation was observed in dysphagic stroke patients. While little pericentral activation remained in the affected hemisphere, no significant activation of the primary sensory and motor areas was seen on the contralesional side.
The appearance of dysphagia after unilateral stroke irrespective of the affected hemisphere is a well-known phenomenon [2, 14, 48, 49]. It is often seen after infarction of the middle cerebral artery. In the current experiment, the reduced activation of the affected hemisphere is, at least partly, explained by the reduction of grey matter due to the ischemic stroke, which makes less neurons and synapses available for the initiation and coordination of deglutition. More remarkable, though is the nearly abolished activation in the contralesional hemisphere found in patients after acute right and left hemispheric stroke. This could either be a sign of a general initial cortical shock after stroke , or point to a more network or function-specific reaction. To our knowledge to date no imaging studies on swallowing in patients in this acute stage after stroke have been performed. However, in the language domain, Saur and co-workers  examined aphasic patients and found a strongly reduced activation of the contralesional hemisphere when studying patients within the first two days after stroke. This pattern changed to an increased brain activation of the contralesional hemisphere in the subacute phase (mean 12.1 days after stroke), with subsequent normalization in the chronic phase . Saur and co-workers concluded that the initial decrease of contralesional brain activation in the acute stage might be explained by disruption of the language network in terms of diaschisis. The concept of diaschisis was coined by von Monakow in 1914. In brief, it explains loss of function in undamaged brain areas connected to an acutely damaged area, as a consequence of a disruption of function in the remaining intact system [52, 53]. We propose a comparable effect in the present study. We hypothesize that the distinct reduction of contralesional activation is caused by a disturbance of the swallowing network of the ipsilesional hemisphere, leading to a strongly reduced or even missing activation in these areas.
In contrast to the former study by Saur where patients were examined within two days, patients in the present study were measured about eight days after stroke. Unfortunately, no longitudinal data are available in the present study. Based on previous results [20, 51, 54], we hypothesize that in dysphagic stroke patients, following the acute underactivation of the contralesional hemisphere and a subsequent over-activation, a normalization of activation is predicted in the chronic stage in those patients who recover well.
Hemispheric stroke without dysphagia
In patients without any signs of dysphagia after hemispheric stroke, extensive bilateral pericentral activation comparable to the group of healthy controls was observed in the early subacute phase after stroke. The observed lateralization to the contralesional hemisphere was only little pronounced and did not reach significance in the small group of examined patients. Still, these results are in line with previous TMS studies by Hamdy and co-workers. Patients that recovered from dysphagia presented with an increased motor representation of the pharynx in the unaffected hemisphere, while patients with persisting dysphagia showed no relevant changes of cortical motor representation . Taken together, the Hamdy 1998 study and our own results suggest that activation of, and reorganization within the unaffected hemisphere is essential for enabling a normal swallowing function in patients after stroke.
In all groups of hemispheric stroke patients, beta synchronization of the dorsolateral prefrontal cortex (DLPFC; BA 44, 45 and 46) and/or insular cortex (BA 13) was found. Bilateral ERS was seen in both groups of non-dysphagic patients, while in dysphagic patients with hemispheric stroke only the right hemispheric premotor synchronization reached significance. Remarkably, this synchronization was neither seen in healthy control subjects nor in dysphagic and non-dysphagic brainstem stroke patients.
So far, the relevancy of beta ERS is not completely understood. On the one hand, it is known to be involved in the post-movement stage and higher cognitive functions [55–57]. On the other hand, it seems to play an important role in resting .
The insular cortex has connections to several brain regions linked to swallowing, including the premotor cortex, the frontal operculum, and secondary somatosensory and retroinsular area of the parietal lobe . Regarding lateralization of insular activation, several studies observed either a bilateral activation, or a right hemispheric dominance [8, 9, 11, 60, 61]. In humans, damage to the frontal operculum as well as damage to the insula have been reported to cause dysphagia [62, 63].
The DLPFC is known to play a crucial role in working memory . An involvement of the prefrontal cortex in central swallowing processing has been seen in previous imaging studies on human swallowing . Moreover, in primates stimulation of the frontal operculum is known to evoke swallowing .
It remains unresolved why ERS of the DLPFC and the insular was observed in hemispheric stroke patients in the present study. A former MEG study of our group utilizing an identical self-paced swallowing paradigm on a 151 MEG channel system found, significant ERS in the frontal operculum and insular cortex in healthy subjects during volitional swallowing . In none of our previous studies on cortical swallowing processing using the 275 channel system, significant ERS in these areas were found [10, 15, 16, 18]. The artifact caused by oropharyngeal muscle activation during deglutition makes it difficult to study activation in subcortical and bulbar structures [12, 34]. We suggest that the higher number of channels of our MEG system might even aggravate the problem of the muscular artefact and by this reduce the observable subcortical activation. The DLPFC and insular activation in hemispheric stroke patients is therefore, supposedly, much stronger than that of healthy controls and patients with brainstem stroke. In hemispheric stroke patients, the swallowing-related pericentral areas of the affected hemisphere are destroyed by ischemic stroke. This leads to consecutive disruption of the cortical swallowing network, including a distinct reduction of sensorimotor activation in the contralesional hemisphere, as discussed above. We therefore conclude that the increase of DLPFC and insular activation observed in patients with hemispheric stroke is caused by compensational mechanisms in the acute post-stroke phase.
Both groups of patients with brainstem stroke were very small. Both groups demonstrated with a right hemispheric lateralization of sensorimotor cortical activation. The lateralization effect was stronger for the group of dysphagic patients. Overall, activation in non dysphagic brainstem stroke patients was stronger compared to dysphagic brainstem stroke patients, but reduced compared to the control group.
We propose that the observed effects in cortical swallowing processing after brainstem stroke can be explained by two different effects. First, the ischemic stroke of central pattern generators in bulbar areas has a strong impact on the cerebral network of swallowing processing. It results in an overall reduction of cortical activation in both patient groups, which reflects the concept of diaschisis, as already discussed above. Second, the right hemispheric lateralization found in both groups of brainstem stroke patients supports the hypothesis of hemispheric task sharing in swallowing processing [15, 66]. The left hemisphere more selectively mediates the oral phase and therefore volitional components, whereas the right hemisphere contributes more to the pharyngeal phase and automatic reflexive aspects of swallowing. This was hypothesized by Daniels and co-workers after the utilization of a dual-task paradigm to examine deglutition , and also underlined by a previous MEG study of our group . The influence of right hemispheric stroke on the pharyngeal transit duration and a prolonged oral transit time after left hemispheric stroke was already demonstrated about 13 years earlier by Robbins and co-workers .
The bulbar areas mainly coordinate the pharyngeal (reflexive) phase of deglutition . Therefore, patients with brainstem stroke involving the central pattern generators show predominant impairment of the pharyngeal phase [69–72]. Thus, the relative preponderance of right-sided cortical activation in patients with brainstem-stroke may indeed be interpreted as cortical compensation of subcortically caused dysphagia. This effect is found in dysphagic as well as in non dysphagic patients, suggesting a fast and early onset of compensatory mechanisms. In case of non dysphagic patients, these adjustment mechanisms might even be sufficient to avoid clinically manifest dysphagia.
Apart from the technical restrictions, the small group sizes are a major limitation of the present study. Especially the group size in the brainstem stroke group was very small. 37 patients with ischemic stroke were recruited over 18 month. Due to the necessity to transport patients to the MEG laboratory, only stable patients that did not need continuously monitoring and were able to give informed consent could participate. Therefore, the patients in the present study are about 10 years younger compared to the mean age of stroke patients in general. Aging affects in cortical swallowing processing have been demonstrated for healthy subjects with a mean age of 24 compared to healthy subjects of about 72 years of age . An increase of somatosensory cortical activation during swallowing execution in elderly subjects compared to the young control group was found. This effect was present in both hemispheres, pointing to adaptive cerebral changes in response to aging effects on the complex process of swallowing. The results also underline the relevance of age matched control groups in neuroimaging studies related to deglutition. The control group in the present study was therefore carefully age matched to the groups of stroke patients.
Also patients with severe aphasia had to be excluded because they were not able to give informed consent. This might have caused a bias, since patients with stroke in the language dominant hemisphere are under-represented. Due to the small group sizes a further division into subgroups (subcortical versus cortical hemispheric stroke and pontine versus non pontine brainstem stroke) was not possible. Further studies have to examine the distinct patterns of processing in these subgroups. Additionally, it would be interesting to examine further influence factors on the cortical swallowing processing, including the cerebral blood flow, white matter lesions and brain atrophy.