Altered functional connectivity in language and non-language brain networks in patients diagnosed with acute post-stroke aphasia

Objective: A resting-state functional magnetic resonance imaging (rs-fMRI) approach was used to explore functional connectivity (FC) in language and non-language brain networks in acute post-stroke aphasia (PSA) patients, with a specific focus on the relationship between these fMRI results and patient clinical presentation. Methods: In total, 20 acute PSA patients and 30 age, sex, and education level-matched healthy control (HC) participants were recruited and subjected to rs-fMRI imaging. In addition, western aphasia battery analyses （ WAB ） were used to compute aphasia quotient (AQ) values for PSA patients. Granger causality was employed to examine connections among cognition-associated resting-state brain networks, and the right middle frontal gyrus (RMFG),the mirror brain regions of Broca ’ s area and the Wernicke ’ s area, the right superior temporal gyrus were selected as regions of interest (ROIs). The REST plus software was then used to perform FC analyses of these regions to analyze changes in FC related to PSA pathogenesis. Results: Relative to HC individuals, PSA patients exhibited significantly higher levels of intra-network FC between the right middle frontal gyrus (RMFG) and the left middle occipital gyrus (LMOG), with such FC being positively correlated with the AQ scores (P = 0.018). Moreover, reduced FC was detected between the Broca ’ s area homolog and the left middle frontal gyrus (LMFG), while FC was enhanced between the Wernicke ’ s area homolog and cerebellar vermis, and this FC was similarly positively correlated with patient AQ scores (P = 0.0297). Conclusion: These results suggest that FC between the bilateral hemispheres of the brain is significantly disrupted in acute PSA patients, interfering with the normal non-specific language network. Aphasia severity was further found to correlate with FC among many of the analyzed regions of the brain.


Introduction
The anatomical damage induced when a stroke occurs can disrupt normally intrinsic task-driver behavioral characteristics and activities, effectively resetting certain aspects of brain connectivity.Indeed, focal lesions can have a profound impact on entire neurological networks involved in intrinsically related activities [1].The etiological basis for post-stroke aphasia (PSA) entails both direct localized damage to specific functional areas of the brain together with consequent dysfunction in connected regions of the brain language networks [2].Efforts to detect changes in language-associated functional networks following stroke incidence can aid in the selection of appropriate therapeutic strategies and targets with the goal of aiding patients recovering from aphasia.
While most studies conducted to date have focused on aphasia incidence during the chronic post-stroke period, this study was developed to specifically focus on patient neurological activity during the acute post-stroke period(within two weeks), given that this period corresponds to higher levels of neuroplasticity as the brain undergoes dynamic remodeling in response to injury.Therapeutic intervention during this acute period may be better able to improve patient outcomes.
According to previous studies，the role of contralesional right hemisphere(RH) homologues playing in recovery of aphasia due to left hemisphere stroke is still undefined.As we assumed that the stroke lesion could seriously affect FC values,all ROIs located in the left hemisphere were excluded from the further analysis.As a result, four regions of the right hemisphere -the right middle frontal gyrus (RMFG),the mirror brain regions of Broca's area and the Wernicke's area, the right superior temporal gyrus were selected as regions of interest (ROIs).And the results of previous studies have confirmed that the FC changes among these ROIs affects chronic aphasia recovery [3].This study was designed to explore the association between changes in language networks during the early post-stroke period and improvements in symptoms of aphasia with the goal of providing a foundation for the therapeutic treatment of acute-phase aphasia patients.

Subjects
Between October 2017 and June 2020, 25 acute stroke patients suffering from aphasia were recruited for this study from our department.To be eligible for inclusion, patients needed to meet the following criteria: (1) 18-75 years of age; (2) first-onset stroke with left hemispheric damage that had been confirmed via diffusion-weighted imaging (DWI) or CT scan; (3) patients presented to the hospital within two weeks following the onset of symptoms; (4) aphasia had been diagnosed using the Western Aphasia Battery (WAB) (aphasia quotient [AQ] < 93.8); (5) patients were right-handed prior to stroke occurrence; (6) patients had not experienced language dysfunction prior to aphasia incidence; and (7) patients were native speakers of Chinese.Patients were excluded if: (1) they exhibited and history of substantial physical, psychiatric, or neurological illness including stuttering or other speech disorders, diabetes, depression, dysarthria, or impaired vision or hearing; (2) they had recently utilized psychotropic or neuromodulatory drugs, or had a history of drug or alcohol abuse; (3) they exhibited contraindications that precluded MRI examination; or (4) they exhibited intensive head movements during fMRI scanning.

Clinical assessment
The National Institutes of Health Stroke Scale (NIHSS) was used to evaluate the neurological function of included patients, while the WAB was employed to compute AQ values based on scores for subscales analyzing naming, repetition, fluency, and listening comprehension.Clinical evaluations of patients were performed by several neurologists (with 5 years minimum experience) of the Center for Speech Pathology and Neurorehabilitation.

fMRI analyses
All MRI scanning was conducted using a 3.0 Tesla Scanner (Philips Medical Systems Nederland B.V., The Netherlands) with a standard head coil in The First Affiliated Hospital of Soochow University Hospital.For diagnostic imaging and to exclude the presence of other organic lesions, T1WI, T2WI, and DWI sequences were routinely acquired.In addition, a resting state functional image EPI sequence was acquired with the following settings: acquisition was parallel to the anterior-posterior joint, TR 2000 ms, TE 30 ms, layer thickness 4 mm, layer spacing 0.4 mm, flip angle 90 • , matrix 64 × 64, field of view 240 × 240 mm 2 , 30 layers, 250 time points.Subjects were directed to lie down and remain still with their eyes closed while not intentionally thinking about anything in particular during scanning.In addition, 3D T1W1 structural images (3D-TFE sequence) were acquired as follows: axial acquisition, TR 7.9 ms, TE 3.5 ms, flip angle 8 • , field of view 250 × 198 mm, matrix 250 × 198, 160 layers (thickness: 1 mm), bandwidth 25 Hz, NEX 1.

Data pre-processing
fMRI images were initially pre-processed using the DPARSF software.Values of head motions in three dimensions were also checked manually and only the data of subjects with the head movements less than 3 mm and 3º were accepted for the further analysis.The first 10 time points were excluded from the analysis to ensure better signal stability.Data were then subjected to slice-trimming, realignment, fusion with 3D T1WI structural images, alignment to the MNI standard space, resampling at 3 × 3 × 3 mm 3 , de-linearization, and covariate regression analyses for covariates including 6 head movement correction-derived parameters and the mean signals for the cerebrospinal fluid, white matter, and whole brain.Following further de-linearization, data were band-pass filtered (0.01-0.08 Hz) and smoothed using a Gaussian kernel (FWHM: 8 mm).

ROI-based functional connectivity analyses
The right middle frontal gyrus (RMFG),the Broca's area homolog,the Wernicke's area homolog and the right superior temporal gyrus were selected as ROIs for analyses focused on aberrant neural connectivity.The REST software was then used to extract the average time series for these ROIs, and correlation analyses were conducted based on the time series for each voxel within the whole brain to establish a whole-brain correlation map.A functional connectivity map was then established through the Fisher Z-transformation of this whole-brain map.

Statistical analyses
SPSS 22.0 (IBM, NY, USA) was used to analyze patient clinical and demographic data using chi-square tests, two-sample t-tests, and paired t-tests.Voxel-based imaging data were subject to multiple statistical comparisons via GRF correction.Signals for areas of the brain that differed within or between groups were then extracted, and correlations between these signal values and AQ scores were analyzed.

Clinical findings
After screening, a total of 30 normal controls and 20 patients with motor aphasia following acute stroke were included in this study for appropriate statistical analyses.The patient group consisted of 16 males and 4 females (average age = 54.2 ± 14.54 years,the average AQ=64.17± 27,the average education level=8.35± 3.08)and the control group consisted of 14 males and 16 females (average age = 52 ± 26.97 years, the average education level=10.47± 4.29).We did not find any significant difference in age, gender, and education between the two groups (P > 0.05).
The demographic and clinical data for these subjects are compiled in Table 1.

ROI-based functional connectivity analyses
Increased FC was detected between the right middle frontal gyrus and the left middle occipital gyrus in aphasia patients relative to healthy controls, and this increase was positively correlated with the AQ scores in these patients (P = 0.018) (Fig. 1a).Moreover, reduced FC was detected between the mirrored Broca's area and the left middle frontal gyrus (LMFG) (Fig. 1b) while FC was enhanced between the mirrored Wernicke's area and the cerebellar vermis (Fig. 1c), and this FC was similarly positively correlated with patient AQ scores (P = 0.0297).Reduced FC was also detected between the right superior temporal gyrus and the left anterior-posterior central gyrus, insula, and bilateral occipital lobes in these patients, whereas FC with the left cerebellar hemisphere was increased (Fig. 1d).

Analyses of the correlations between functional connectivity findings and patient symptoms
The higher level of intra-network FC between the right middle frontal gyrus (RMFG) and the left middle occipital gyrus (LMOG), was positively correlated with the AQ scores (P = 0.018).The enhanced FC between the mirror brain regions of Wernicke's area and cerebellar vermis was similarly positively correlated with patient AQ scores (P = 0.0297)..

Discussion
Aphasia is a condition that results from the disruption of languagerelated cognition resulting in impaired auditory comprehension, reading, writing, naming, repetition, and/or spontaneous conversational abilities.Stroke incidence is the most common driver of aphasia incidence, and 20-40 % of stroke patients ultimately experience acute aphasia that can contribute to higher rates of morbidity, mortality, and negative social outcomes [4][5][6].Aphasia incidence during the chronic phase of stroke recovery remains similarly high, affecting up to 25 % of patients [7,8].There thus remains a pressing need for new clinical interventions capable of reliably and efficiently alleviating the symptoms of aphasia in affected individuals.
The use of rs-fMRI strategies can offer insight into intrinsic neurological activity and shifts in FC among different regions of the brain [9].During these resting-state scans, participants do not engage in any specific tasks.A resting network can then be established based on brain regions with BOLD activation patterns that are reliably synchronized over time.As rs-fMRI scanning capabilities are widely available, they are frequently used in the evaluation of aphasia patients [10].Such rs-fMRI scanning is of particular value in acute aphasia patients given that it can be readily conducted even in cases where the degree of aphasia is severe.The resultant functional networks captured through a single round of rs-fMRI sequence acquisition can offer insight into a range of cognitive domains while simultaneously evaluating various potential forms of impairment [11].
Cortical reorganization is a key post-stroke neuroplasticity-related process that entails the migration of neurological functions normally associated with damaged regions of the brain to other undamaged regions.In mouse studies, contralateral limb stimulation has been shown to induce ipsilateral cerebral cortical activity for 1-3 days after stroke incidence, consistent with the ongoing reorganization of sensory inputs within the intact period during this period.However, by 1-2 weeks poststroke this activity shifts back to the damaged hemisphere of the brain, with intact peripheral cortical regions assuming the functional roles of the damaged areas of the brain [12][13][14].Many fMRI and positron emission tomography (PET) analyses have confirmed similar phenotypic shifts in the motor and language domains of the brain in humans recovering from strokes [10][11][12].Here, significantly reduced FC was observed between the injured area of the brain and the mirrored area during the acute post-stroke period likely owing to structural damage, whereas the mirrored area exhibited enhanced connectivity with peri-lesional areas that may be related to improved language function, given that this language mirror area is an important mediator of appropriate acute recovery through the functional reorganization of the residual peri-lesional area.
Here, the mirror of Wernicke's area and the right superior temporal gyrus exhibited enhanced connectivity with the cerebellum that was positively correlated with patient AQ.This indicates that the cerebellum plays a key role in recovery from aphasia.Enhanced connections with the cerebellum and were positively correlated with AQ.In addition to serving as a key regulator of motor functions, the cerebellum also governs key processes such as behavioral regulation, executive function, and visuospatial processing.The cerebral cortex projects to the cerebellum in both mice and humans [15], and changes in the cerebellum may result from structural damage to the brain that disrupts cortico-cerebellar loops and induces functional reorganization [16].Turkeltaub et al. determined that the tDCS-based stimulation of right cerebellar nodes in healthy volunteer study participants resulted in significant improvements in phonological fluency that were related to improved FC between the cerebellum and speech motor regions of the brain together with resting-state remote connectivity between language motor areas and speech areas in the left hemisphere.This indicates that cerebellar stimulation has the potential to support rehabilitative outcomes in aphasia patients [17].Here, enhanced connectivity was detected between language mirror areas and the cerebellum following stroke-induced brain damage consistent with language networks undergoing functional reorganization, thereby contributing to the symptoms of aphasia.
We additionally detected reduced FC between the right superior temporal gyrus and the left anterior-posterior central gyrus, insula, and bilateral occipital lobes in these aphasia patients during the acute poststroke period, consistent with a dual-flow model of aphasia recovery [18,19].This dual-flow model posits that language functions arise from functional connections between parietal, frontal, and distal temporal regions of left-lateralized brain networks, organizing language abilities using two key processing systems.One of these systems is the bilateral ventral stream, which facilitates information processing, the semantic interpretation of auditory inputs, and associated comprehension.This system is primarily localized to the lateral temporal lobe, extending to the posterior frontal gyrus via the hook fasciculus or orbitally to the inferior frontal gyrus orbitally.The second of these systems is the left dorsal stream, which corresponds to the connections between the temporoparietal region and the left frontal language area, functioning to process auditory to articulatory information.Key brain areas involved in this system include the frontoparietal region including the envelope, triangle, postcentral and precentral regions, and portions of the parietal lobe.In aphasia patients included in the present study, reduced connectivity was detected between the right superior temporal gyrus and the insula/posterior central gyrus, respectively corresponding to the ventral and dorsal streams consistent with damage to these two systems.The insula cortex is an important facet of the language center necessary for the appropriate formation and expression of language, and for both implicit and extrapolative grammatical learning.This region is also closely associated with language-related brain regions including those involved in comprehension (Wernicke's area), repetition (supramarginal gyrus), and production (Broca's area) [20].Impaired syntactic accuracy has been linked to impaired activity in the left inferior frontal gyrus, insula, postcentral gyrus, precentral gyrus, and superior parietal inferior border gyrus [21].These results may better explain the observed impairment of motor language and comprehension in aphasia patients.

Conclusions
In summary, these results offer additional insight into resting-state changes in the brain networks of aphasia patients during the acute post-stroke period and the associations between these changes and language recovery, offering a new foundation for the treatment of affected patients.However, the sample size included in this study was limited and the degree of aphasia varied substantially among included patients.As such, further large-scale studies will be essential to explore the consistency of changes in brain networks associated with varying levels of infarct size and aphasia in order to better direct patient care.

Funding
This work was supported by National Natural Science Foundation Program of China (No. 81701669).The funders had no role in the study's design, interpretation of data, or in writing the manuscript.They had contributed to the data collection.

CRediT authorship contribution statement
Liying Han,JunKe and Dawei Zhang designed the study and interpretation of data and drafting of the manuscript.Boye Ni, Yuanyuan Tao and Qingqing Zhou made acquisition of data.
Liying Han did an analysis of data.All of the authors have approved the final version of the manuscript to be published and agreed to be accountable.

Declaration of Competing Interest
There are no competing interests for all the other authors of this manuscript.

Fig. 1 .
Fig. 1. Results of ROI-based functional connectivity analysis between groups (a) functional connectivity of the right middle frontal gyrus; (b) functional connectivity of mirror brain regions in Broca's area; (c) functional connectivity of mirror brain regions in Wernicke's area; (d) functional connectivity of the right superior temporal gyrus.

Fig. 2 .
Fig. 2. Correlation analysis between functional connectivity and clinical symptoms in the mirror brain regions of the right middle frontal gyrus and Wernicke's area MFG, middle frontal gyrus; MOG, middle occipital gyrus; FC, functional connectivity;RH homologue,right hemispheric homologue.

Table 1
Demographic data and clinical indicators of the subjects.
Categorical variables are presented as number (%).Continuous variables are presented as mean ± standard deviation.L. Han et al.