Skip to main content
NCBI home page
NeuroImage : Clinical logoLink to NeuroImage : Clinical
. 2024 Jul 23;43:103648. doi: 10.1016/j.nicl.2024.103648

Functional network reorganization after endovascular thrombectomy in patients with anterior circulation stroke

Tongyue Li a,b,1, Jiaona Xu c,1, Luoyu Wang d, Kang Xu a,b, Weiwei Chen a,b, Liqing Zhang d, Guozhong Niu e, Yu Zhang f, Zhongxiang Ding d,⁎, Yating Lv a,b,⁎
PMCID: PMC11332103  PMID: 39067302

Highlights

  • •

    This study investigated function alternations of brain networks in acute ischemic stroke patients after thrombectomy.

  • •

    FMRI showed altered functional connectivity of advanced brain networks in post-thrombectomy patients.

  • •

    The function conditions of brain networks were different in patients with different reperfusion degree.

Keywords: Acute ischemic stroke, Thrombectomy, Resting-state functional magnetic resonance imaging, Functional brain network, Functional connectivity

Abstract

Background

Endovascular thrombectomy has been confirmed to be an effective therapy for acute ischemic stroke (AIS). However, how functional brain networks reorganize after restoration of blood supply in AIS patients, and whether the degree of reperfusion associates with functional network changes remains unclear.

Methods

Resting-state fMRI data were collected from 43 AIS patients with anterior circulation occlusion after thrombectomy and 37 healthy controls (HCs). Both static and dynamic functional connectivity (FC) within four advanced functional networks including dorsal attention network (DAN), ventral attention network (VAN), executive control network (ECN) and default mode network (DMN), were calculated and compared between post-thrombectomy patients and HCs, and between two subgroups of post-thrombectomy patients with different reperfusion conditions.

Results

As compared to HCs, patients showed significant differences in static FC of four functional networks, and in dynamic FC of DAN, ECN and DMN. Furthermore, patients with better reperfusion conditions exhibited increased static FC with precuneus, and altered dynamic FC within precuneus. Moreover, these alterations were associated with clinical assessments of stroke severity and functional recovery in post-thrombectomy patients.

Conclusions

Collectively, these findings may provide the potential imaging markers for assessment of thrombectomy efficacy and help establish the specific rehabilitation treatments for post-thrombectomy patients.

1. Introduction

Acute ischemic stroke (AIS) is one of the most frequent causes of mortality and chronic disability worldwide, and timely reperfusion of ischemic tissue with recoverable potential is critically important for functional outcome (Mao et al., 2022). As one of the endovascular treatments (EVT), thrombectomy has been confirmed to be an effective treatment for AIS patients with large vessel occlusions (Campbell et al., 2015, Berkhemer et al., 2015), which can extend time window and enhance success of vascular recanalization of patients (Albers et al., 2018, Nogueira et al., 2018). However, the understanding of how brain functions preserved in post-thrombectomy patients is rather limited.

Resting-state functional magnetic resonance imaging (rs-fMRI) measures the variations of blood oxygenation level–dependent (BOLD) signal without tasks, providing quantitative measure of intrinsic neural activity (Biswal et al., 1995). Converging evidence has proved that there exists anatomically distinct but functionally interactive brain regions where neural activity exhibits a high degree of correlation during resting state (Van den Heuvel and Hulshoff Pol, 2010). To date, seven resting-state functional networks has been consistently identified (Yeo et al., 2011, Schaefer et al., 2018), including Visual Network (VN), Sensorimotor Network (SMN), Limbic System, Dorsal Attention Network (DAN), Ventral Attention Network (VAN), Executive Control Network (ECN) and Default Mode Network (DMN). Information exchanged within functional networks, especially within four advanced functional networks (i.e., DAN, VAN, ECN and DMN), is of significant importance for brain to carry out cognitive and action processes. Corresponding functional impairments would subsequently occur when the neural activity within special network is disrupted (Bonkhoff et al., 2020, Bonkhoff et al., 2021). Therefore, investigating the functional changes within different networks is of great significance to explore how the brain responds to the focal damages after stroke.

Functional connectivity (FC) calculates synchronization patterns of spontaneous neural activities to illustrate the relationship between the regions within specific neural circuits (Biswal et al., 1995). Previous studies have revealed that there are aberrant changes in FC within the advanced networks following stroke (Zhao et al., 2018, Chen et al., 2019). However, patients in those studies were not treated with any EVT, and thus cannot reflect how functional networks reorganize after restoration of blood supply. Recently, Puig and colleagues (Puig et al., 2018) found that AIS patients with better outcome after thrombolytic therapy exhibited increased FC than patients with poor outcome. However, whether the alterations occur in different functional networks of post-thrombectomy patients, and whether the degree of reperfusion would affect these alterations remain unclarified.

Therefore, the present study employed rs-fMRI to investigate the connectivity alterations of brain networks in AIS patients after thrombectomy and test whether the FC within functional networks could differentiate AIS patients from HCs and distinguish patients with different reperfusion conditions.

2. Materials and methods

2.1. Participants

Sixty-eight AIS patients with anterior circulation occlusion were collected at the Hangzhou First People’s Hospital from July 2020 to May 2022. Each patient was received thrombectomy treatment within 24 h (7 h ± 3h55min) after stroke onset and was prescribed antiplatelet and statins agents within three months after thrombectomy treatment. The inclusion criteria for patients were as follows: (a) anterior circulation occlusion which included internal carotid artery (ICA), anterior cerebral artery (ACA) and middle cerebral artery (MCA) confirmed by CTA; (b) onset within 6 h, recommended for vascular intervention; onset between 6 and 24 h, recommended for vascular intervention after rigorous imaging selection (CTP) with a core infarct volume less than 70 ml, ratio of ischemic penumbra to core volume ≥1.8, and absolute value of ischemic penumbra volume ≥15 ml; and (c) no hemorrhage. Patients with hemorrhage, severe systemic diseases, psychiatric abnormalities or contraindications for MRI were excluded in this study. Thirty-nine age- and sex- matched HCs were recruited and none of the HCs had a history of neurological disease or psychiatric disorder.

This study was approved by the Ethics Committee for Medical Technology, Clinical Applications, and Research at Hangzhou First People's Hospital. All participants or their legal guardians signed written informed consent before participating in the present study.

2.2. Data acquisition

2.2.1. Clinical assessment

All patients underwent a series of assessments from admission to 90-day after stroke onset. These assessments included the National Institutes of Health Stroke Scale (NIHSS) at three time points: at admission, 24 h after thrombectomy and 90-day after stroke onset. Additionally, the expanded Thrombolysis in Cerebral Infarction (eTICI) scale was estimated during thrombectomy, and the modified Rankin Scale (mRS) was also applied which accessed the functional outcome at 90-day after stroke. Patients were further categorized into two subgroups based on eTICI: poor reperfusion (PR) with scores 0-2b; good reperfusion (GR) with scores 2c-3.

2.2.2. MRI data acquisition

MRI data were collected with a 3T MR scanner (Siemens, Germany) at the Hangzhou First People’s Hospital. All AIS patients underwent scans within one week (3 ± 1 days) after thrombectomy. The standard MRI protocol included rs-fMRI, structural MRI (sMRI), and diffusion weighted imaging (DWI). Scanning parameters were listed in Supplementary Materials.

2.3. Data analysis

2.3.1. Creation of the lesion map

Lesion masks of AIS patients were delineated based on sMRI and DWI images by one radiologist (L.Z.) with experience over ten years using ITK-SNAP software (https://www.itksnap.org/pmwiki/pmwiki.php). Delesion processing was then conducted on sMRI by replacing lesions of abnormal intensities with contralesional homologous areas using clinicaltbx (https://www.nitrc.org/plugins/mwiki/index.php/clinicaltbx:MainPage) (Nachev et al., 2008). The sMRI, DWI images and lesion masks were flipped from right to left for patients with right hemisphere occlusions. The delesioned sMRI images and lesion masks were further normalized to the Montreal Neurological Institute (MNI) space. Finally, we took the union of lesion masks with all patients and obtained lesion map (Fig. 1).

Fig. 1.

Fig. 1

Open in a new tab

Infarct lesion map of AIS patients. Lesions were overlapped in patients and shown in axial slices. The color indicates the number of patients with focal brain lesion at any given location. Abbreviations: IL, ipsilesional hemisphere; CL, contralesional hemisphere; AIS, acute ischemic stroke.

2.3.2. Preprocessing of resting-state fMRI data

The rs-fMRI images were initially flipped to ensure that all lesions were located in left hemisphere. Regular preprocessing operations were then conducted with the RESTplus toolbox (version 1.25) (Jia et al., 2019) as shown in Supplementary Materials. Besides, individual brain mask was extracted from the normalized images and gray matter mask was obtained by calculating intersection between overlap of all individual brain masks and Anatomical Automatic Labeling (AAL) atlas with lesion and subcortical regions removed. Specially, the effects of hemodynamic lags were removed by shifting the time series of each voxel, in accordance with calculated hemodynamic lags using the time-shift analysis in stroke patients (Lv et al., 2013). In this study, twenty-five patients and two HCs were excluded due to maximal head displacement >3 mm or rotation >3°, leaving forty-three AIS patients and thirty-seven HCs.

2.4. Regions of interest (ROIs) selection

The present study specifically concentrated on four advanced functional networks, including DAN, VAN, ECN and DMN. To define ROIs in each network, we first partitioned whole brain into 200 ROIs based on Schaefer’s 200-area parcellation (Schaefer et al., 2018). ROIs were further obtained after intersecting with gray matter mask, leaving 86 ROIs larger than 10 voxels for four networks: DAN with 19 ROIs, VAN with 15 ROIs, ECN with 21 ROIs and DMN with 31 ROIs, as shown in Table S1.

2.5. Static functional connectivity (sFC) within each network

To access sFC within each network, Pearson correlation coefficients were calculated between averaged time series of each pair of ROIs in the network. These correlation coefficients were then transformed into z-values via Fisher’s r-to-z transformation.

2.6. Dynamic functional connectivity (dFC) within each network

To detect temporal variability in functional connectivity within each network, dFC analysis was performed for each pair of ROIs within network using sliding window method. The sliding window had a length of 50 TRs and shifted by 1 TR, resulting in a total of 121 windows for each ROI. Within each time window, the Pearson correlation coefficient was calculated and transformed to z-values via Fisher’s r-to-z transformation. Subsequently, the coefficient of variation (CV = SD/mean) of 121 correlation coefficients for each ROI pair was calculated for next analyses.

2.7. Statistical analysis

We compared demographic and clinical characteristics between HCs and AIS patients, and between two subgroups of patients using the Statistical Packages for the Social Sciences (SPSS; Chicago, IL). Chi-square tests were used for categorical data (i.e., gender) and analysis of variance and two-sample t-test were applied for continuous data.

Between-group difference in intra-regional FCs between post-thrombectomy patients and HCs was inferred using nonparametric permutation test. For each connectivity, we calculated between-group difference of two groups and then recalculated the differences between randomized groups by randomly dividing FC values into two groups with 10,000 permutations. The 95th percentile point of empirical distribution was utilized as critical value to estimate between-group difference and a false discovery rate (FDR) procedure was employed to correct for multiple comparisons (p < 0.05). Similarly, the differences in intra-regional FCs between subgroups of patients were also compared using permutation approach (p < 0.05). The effects of age and gender were removed from FCs using general linear model.

For any FC value exhibiting between-group difference, a Pearson correlation analysis was used to explore its relationship with clinical characteristics (i.e., NIHSS at admission, 24 h NIHSS after thrombectomy, and 90-day NIHSS and mRS). The FDR procedure was also used to correct for multiple comparisons (p < 0.05).

Additionally, the FC values showing between-group differences were correlated with both lesion sizes and the time intervals from thrombectomy to MRI scan in AIS patients to test their possible effects on the results.

2.8. Machine learning analysis for classification

To examine whether the FC alterations could serve as markers to differentiate AIS patients from HCs, and to distinguish patients with different reperfusion conditions, we employed multiple machine learning models for classification analyses, which could enhance prediction accuracy and robustness (Bakasa and Viriri, 2023, Solihin et al., 2024). Combining multiple models with unique strengths allows for a more comprehensive capture of data features, leading to improve overall performance, reduce the risk of overfitting and produce a more reliable result (Rajaraman et al., 2019). Machine learning analysis was performed in MATLAB. Four feature selection methods: Cross-combination Fisher Score (FSCR), Relief-F (RELF-F), Minimum Redundancy Maximum Relevance (MRMR) and Mutual Information Maximization (MIM), were firstly applied to select features from all FCs (both sFCs and dFCs). Four machine-learning classifiers: Decision Tree (DT), Naive Bayers (NB), Random Forest (RF) and Support Vector Machines (SVM) were then trained based on features. Performance of the classifiers were evaluated via 5-fold cross-validation using the area under curve (AUC).

3. Results

3.1. Demographic and clinical characteristics

Demographic and clinical information for all participants was summarized in Table 1. Among the 43 AIS patients after thrombectomy, there were 37 patients with MCA infarctions, 5 with ICA infarctions and 1 with ACA infarctions. There was no significant difference in age (F = 0.531, p = 0.662) or gender (χ2 = 0.188, p = 0.910) between AIS patients and HCs. Furthermore, no significant difference was observed in NIHSS at admission (t = 0.396, p = 0.694), 24 h NIHSS after thrombectomy (t = 0.661, p = 0.512), 90-day NIHSS (t = 0.761, p = 0.451), and 90-day mRS (t = −0.985, p = 0.331) between subgroups patients.

Table 1.

Demographic and clinical characteristics of participants. All values were presented as mean ± standard deviation. a The p-value was obtained by one-way analysis of variance test. b The p-value was obtained by a chi-square test. c The p-value were obtained using two-sample two-side t-tests.

Characteristics AIS (n = 43) GR (eTICI: 2c-3; n = 28) PR (eTICI: 0-2b; n = 15) HCs (n = 37) p Value
Age (years) 70.047 ± 11.286 70.786 ± 9.612 68.667 ± 14.171 67.810 ± 7.832 0.662a
Gender (male/female) 19/24 13/15 6/9 17/20 0.910b
Thrombectomy time after onset 7 h ± 3 h 55 min 7 h ± 4 h 1 min 7 h ± 3 h 43 min − 0.999c
NIHSS at admission 15.023 ± 6.913 14.714 ± 6.749 15.600 ± 7.414 − 0.694c
24 h NIHSS after thrombectomy 12. 139 ± 7.167 11.607 ± 7.455 13. 133 ± 6.728 − 0.512c
90-day NIHSS 3.302 ± 3.543 3.000 ± 3.174 3.867 ± 4.207 − 0.451c
90-day mRS 2.093 ± 1.428 2.250 ± 1.404 1.800 ± 1.474 − 0.331c
Open in a new tab

Abbreviations: AIS, acute ischemic stroke; HCs, healthy controls; NIHSS, National Institute of Health stroke scale; mRS, Modified Rankin Scale; eTICI: expanded Thrombolysis In Cerebral Infarction; GR, good reperfusion; PR, poor reperfusion.

3.2. Between-group differences in static functional connectivity

Significant between-group differences (AIS vs. HCs, GR vs. PR) of sFC were observed in four functional networks (Fig. 2a and Fig. 2b).

Fig. 2.

Fig. 2

Open in a new tab

Between-group differences of functional connectivity in four advanced functional networks. The distribution of 86 brain regions across four networks (DAN with 19 ROIs, VAN with 15 ROIs, ECN with 21 ROIs, and DMN with 31 ROIs) were displayed in upper row. The ROI pairs within different networks showed significant differences of sFC or dFC between AIS and HCs groups (A, C) (permutation test, p < 0.05, FDR corrected), and between GR and PR groups (B, D) (permutation test, p < 0.05). The red connecting line indicates significantly increased connectivity in AIS than HCs as well as in GR than PR, while the blue connecting line indicates decreased FCs in AIS than HCs as well as in GR than PR. Abbreviations: IL, ipsilesional hemisphere; CL, contralesional hemisphere; sFC, static functional connectivity; dFC, dynamic functional connectivity; DAN, Dorsal Attention Network; VAN, Ventral Attention Network; ECN, Executive Control Network; DMN, Default Mode Network; AIS, acute ischemic stroke; HCs, healthy controls; PR, poor reperfusion; GR, good reperfusion; L, left (i.e. ipsilesional hemisphere); R, right (i.e. contralesional hemisphere); I, inferior; S, superior; M, middle; A, anterior; P, posterior; PreCG, Precental gyrus; PoCG, Postcentral gyrus; SMA, Supplementary motor area; SFGdor, Superior frontal gyrus, dorsolateral; SFGmed, Superior frontal gyrus, medial; ORBsupmed, Superior frontal gyrus, orbital part; MFG, Middle frontal gyrus; ORBmed, Middle frontal gyrus, orbital part; IFGoperc, Inferior frontal gyrus, opercular part; IFGtriang, Inferior frontal gyrus, triangular part; SOG, Superior occipital gyrus; MOG, Middle occipital gyrus; SPG, Superior parietal gyrus; IPL, Inferior parietal gyrus; STG, Superior temporal gyrus; TPOsup, Temporal pole: superior temporal gyrus; ITG, Inferior temporal gyrus; MTG, Middle temporal gyrus; ACG, Anterior cingulate and paracingulate gyri; DCG, Median cingulate and paracingulate gyri; PCG, Posterior cingulate gyrus; SMG, SupraMarginal gyrus; FFG, Fusiform gyrus; PCUN, Precuneus; CUN, Cuneus; REC, Gyrus rectus; INS, Insula; ANG, Angular gyrus. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

As compared to HCs, AIS patients exhibited 49 increased sFCs (18 interhemispheric sFCs and 31 intrahemispheric sFCs) within DAN. The increased interhemispheric connectivity mainly included bilateral precuneus (PCUN) and parietal cortex, contralesional temporal cortex and ipsilesional fusiform as well as superior occipital gyrus (SOG). Additionally, AIS patients showed increased connectivity both within ipsilesional hemisphere (PCUN and parietal cortex, PCUN and SOG) and within contralesional hemisphere (parietal cortex and temporal cortex, parietal cortex and PCUN). As compared to GR patients, increased sFCs were observed between ipsilesional PCUN and contralesional perietal cortex, and between contralesional inferior parietal gyrus (IPL) and superior parietal gyrus (SPG) in PR patients.

Compared with HCs, 21 increased sFCs (12 interhemispheric sFCs and 9 intrahemispheric sFCs) of AIS patients in VAN were mainly presented between ipsilesional frontal cortex and contralesional temporal cortex, as well as between contralesional temporal cortex and insula. In addition, GR patients exhibited increased sFCs between median cingulate gyrus (DCG) and superior frontal gyrus (SFG) when compared with PR patients.

For ECN, significant alterations of 44 sFCs (18 interhemispheric sFCs and 26 intrahemispheric sFCs) were observed. AIS patients revealed increased connectivity between ipsilesional parietal cortex and contralesional frontal cortex, between ipsilesional cingulate cortex and contralesional frontal cortex, and between contralesional cingulate cortex and frontal cortex. Whereas, decreased sFCs were observed between ipsilesional IPL and contralesional angular gyrus (ANG), bilateral anterior cingulate gyrus (ACG), and between ipsilesional ACG and contralesional supplementary motor area (SMA). Meanwhile, GR patients exhibited increased sFCs between contralesional IPL and ANG, and between ipsilesional IPL and contralesional ANG.

Post-thrombectomy patients exhibited significant differences in 29 sFCs (15 interhemispheric sFCs and 14 intrahemispheric sFCs) in DMN as compared to HCs. Increased sFCs were presented between ipsilesional ACG and contralesional PCUN, between ipsilesional SFG and contralesional frontal cortex, and between contralesional inferior frontal gyrus (IFG) and temporal cortex in patients. Conversely, decreased sFCs were observed between ipsilesional middle frontal gyrus (MFG) and ACG, and between ipsilesional PCUN and contralesional posterior cingulate gyrus (PCG) in patients. As compared to PR patients, GR patients exhibited increased sFCs between ipsilesional ACG and PCUN, between ipsilesional ACG and rectus, and between bilateral PCUN.

3.3. Between-group differences in dynamic functional connectivity

When comparing dFC between post-thrombectomy patients and HCs, significant differences were observed in DAN, ECN and DMN (Fig. 2c). In DAN, decreased dFCs in patients were found between ipsilesional SOG and PCUN, between contralesional inferior temporal gyrus (ITG) and PCUN and between contralesional SPG and PCUN. In ECN, patients showed increased dFC between ipsilesional IPL and contralesional ANG. In DMN, increased dFC was found between bilateral PCUN, while decreased dFC was observed between ipsilesional ANG and contralesional IFG in patients. Additionally, GR patients exhibited increased dFC between bilateral PCUN within DMN when compared to PR patients (Fig. 2d).

Notably, in order to verify whether specific infarction areas would influence our findings, we further compared the sFC and dFC in four advanced networks between the AIS patients with MCA infarctions and HCs, and found that between-group differences in connectivity remains unchanged (Table S2). Additionally, no significant correlation was found between the significant FC values and lesion sizes, as well as MRI scan intervals in AIS patients (Table S3).

3.4. Relationship between functional connectivity and clinical assessments

All the results of correction analyses underwent the FDR procedure (p < 0.05). For post-thrombectomy patients, the 90-day NIHSS positively correlated with sFC between ipsilesional SFG and contralesional SMA in VAN (r = 0.531, p < 0.001), and between contralesional DCG and SMA in VAN (r = 0.544, p < 0.001) as shown in Fig. 3a. 90-day NIHSS also positively correlated with sFC between ipsilesional DCG and SFG in VAN in GR patients (r = 0.448, p = 0.019), while 24 h NIHSS after thrombectomy negatively correlated with sFC between ipsilesional IPL and contralesional ANG in ECN in PR patients (r = -0.599, p = 0.01). Meanwhile, 90-day NIHSS significantly correlated with dFC between contralesional ANG and IFG in DMN (r = -0.449, p = 0.003), and between bilateral PCUN in DMN (r = 0.390, p = 0.010), which was also observed in PR patients (r = 0.588, p = 0.021) as shown in Fig. 3b.

Fig. 3.

Fig. 3

Open in a new tab

Relationship between FCs and clinical assessments. Significant correlations (p < 0.05, FDR corrected) between sFC (A) or dFC (B) and clinical assessments (NIHSS scores after thrombectomy or 90-day NIHSS scores) were observed in AIS, PR or GR groups. The red connecting line in brain maps represents significant increased FC in corresponding group, and blue connecting line represents decreased FC in corresponding group. Abbreviations: IL, ipsilesional hemisphere; CL, contralesional hemisphere; sFC, static functional connectivity; dFC, dynamic functional connectivity; VAN, Ventral Attention Network; ECN, Executive Control Network; DMN, Default Mode Network; AIS, acute ischemic stroke; HCs, healthy controls; PR, poor reperfusion; GR, good reperfusion; L, left (i.e. ipsilesional hemisphere); R, right (i.e. contralesional hemisphere); I, inferior; IFGtriang, Inferior frontal gyrus, triangular part; DCG, Median cingulate and paracingulate gyri; SMA, Supplementary motor area; PCUN, Precuneus; ANG, Angular gyrus. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.5. Classifier performance

Classification models were established for classifying post-thrombectomy patients and HCs, and for differentiating patients with different reperfusion conditions. The mean AUC values for all models were illustrated in Fig. 4. For AIS-HCs classification, FSCR selection method combined with RF classifier exhibited the highest performance (AUC=0.7377 ± 0.1135), followed by FSCR combined with DT (AUC=0.7188 ± 0.1907) and RELF-F combined with SVM (AUC=0.7003 ± 0.1426). Meanwhile, for the classification of subgroups in patients, RELF-F selection method combined with NB classifier demonstrated the highest performance (AUC=0.7133 ± 0.2468), followed by FSCR combined with RF (AUC=0.66678 ± 0.2308), and PELF-F combined with SVM (AUC=0.6678 ± 0.2308).

Fig. 4.

Fig. 4

Open in a new tab

The AUC Heatmap of different machine learning models. The performance of all the crossed models (AUC) combining four feature selection (in rows) and four classification (in columns) methods were depicted. (A) The mean AUC of classifiers distinguished AIS from HCs. (B) The mean AUC of classifiers distinguished two subgroups of AIS patients (GR and PR). Abbreviations: AUC, Area under curve; FSCR, Fisher Score; RELF-F, Relief-F; MRMR, Minimum Redundancy Maximum Relevance; MIM, Mutual Information Maximization; DT, Decision Tree; NB, Naive Bayers; RF, Random Forest; SVM, Support Vector Machines.

4. Discussion

In this study, we utilized rs-MRI to investigate FC alterations in four advanced functional networks, and further explored their dependence on the degree of reperfusion. We found that connectivity patterns within four networks reorganized in post-thrombectomy patients. Notably, the reorganizations were found to be affected by the degree of reperfusion after thrombectomy. Additionally, the connectivity changes correlated with functional outcome in patients, and could differentiate stroke patients from HCs and distinguished patients with different reperfusion conditions.

Our connectivity analyses revealed that post-thrombectomy patients exhibited increased connectivity within DAN and VAN. DAN and VAN are two key networks to capture attention information exogenously and deploy attention system endogenously (Spreng et al., 2013). DAN directs the attention of brain to crucial signals while VAN is responsible for shifting our attention between various tasks (Leonards et al., 2000). The increased sFCs within DAN were mainly among temporal cortex and posterior brain regions, and hyperconnectivity was commonly reported in neurological disorders (Hillary et al., 2015). Posner and his colleagues (Posner et al., 1984) observed that damage of parietal cortex in DAN could lead to a phenomenon known as attention disengagement, challenging to focus attention on specific tasks. To compensate for this functional impairment, short-term increase of connectivity in corresponding affected regions could be observed (Hong et al., 2022). Thus, we speculate the damage to attention in post-thrombectomy patients could lead to corresponding compensation mechanism within DAN. Moreover, our dynamic analyses found significant decreased dFCs between ITG and PCUN, and SPG in DAN. As a supplementary measure to sFC, dFC depicts fluctuations in brain connectivity over time. The reduced dFCs in DAN further validated functional impairments existed, affecting the attention function of stroke patients.

In VAN, increased sFCs between DCG and SMA, and between MTG and insula were observed in post-thrombectomy patients. The top-down processing in VAN might be compromised after stroke, resulting in spatial neglect or inability to perceive sensory information on contralesional side (Passingham et al., 2014). However, post-stroke alterations within VAN are currently far less extensive than that within DAN, and studies of post-stroke brain function within VAN primarily focused on the domain of spatial neglect (Barrett et al., 2019, Bian et al., 2023). Barrett and his colleagues (Barrett et al., 2019) found decreased sFC between bilateral ventral frontal cortices and precentral gyrus in post-stroke spatial neglect patients, which was contrary to our findings that increased sFC between frontal cortex and precentral gyrus was observed in post-thrombectomy patients. This may because that thrombectomy therapy restores blood supply of patients, but the recovery of cerebral function lags behind the restoration of blood supply. Thus, we speculate that the abnormally increased connectivity may represent that brain mobilize additional resources to sustain function operation after stroke.

As a network responsible for integrating information and making decisions, ECN primarily handles working memory and task-related consideration (Zhang et al., 2023). We discovered increased sFC between extensive frontal regions in ECN of patients. Frontal cortex in ECN plays a predominant role in planning and decision-making processes (Brown et al., 2019, Liu et al., 2021). Previous research revealed that stroke patients with damage in frontal gyrus within ECN exhibited deficits in executive control (LaCroix et al., 2020). Furthermore, there is an enhanced dFC between ANG and IPL in patients, suggesting a potential compensatory mechanism within brain network.

As one of the most widely studied network during resting-state (Buckner et al., 2008), DMN is associated with emotional and cognitive processing (Mantini and Vanduffel, 2013). We found significant increase in sFC between PCUN and prefrontal cortex, as well as dFC of PCUN in post-thrombectomy patients. PCUN is involved in fundamental brain functions, such as emotional state observation and self-related mental representations (Mantini et al., 2007). In normal circumstances, neural activities of DMN maintain relatively quiescent state, yet in AIS patients, there is an increased strength in sFC within DMN (Ding et al., 2014). The changes are found associated with emotional disorders, especially depression (Zhang et al., 2018), and cognitive impairment (Tuladhar et al., 2013). These findings indicate that connectivity disruptions within DMN still persisted even though blood supply restored in AIS patients after thrombectomy.

Moreover, post-thrombectomy patients with better reperfusion exhibited increased sFCs within four advanced networks as compared to patients with poor reperfusion. Interestingly, between-group differences in FC were primarily associated with connections in PCUN of advanced networks, especially DAN and DMN. The significant dFC in subgroups of patients was only found between bilateral PCUN within DMN. As a functional hub in DMN, PCUN is involved in episodic memory, visuospatial processing, reflections upon self and aspects of consciousness (Margulies et al., 2009, Zhang et al., 2012). Dysfunction in PCUN was frequently reported in neurological and psychiatric disorders including Alzheimer’s disease (Frings et al., 2010, Perez et al., 2015) and depression (Delaveau et al., 2014, Kito et al., 2014). The alteration of PCUN connectivity, particularly within DMN, may serve as an index to reflect functional state in post-thrombectomy patients.

The significantly different sFC within VAN and dFC within DMN were primarily associated with 90-day NIHSS scores. Specifically, enhanced FC exhibited positive correlation with NIHSS scores, while reduced FC showed negative correlation with NIHSS scores. That is, the better functional outcome in patients, the higher relevant FC values and vice versa, implying that functional alterations within network can be used to monitor functional outcome. Future longitudinal studies are required to elucidate the roles played by FCs within VAN and DMN in functional outcomes of stroke.

Using different machine learning models, we found the connectivity within four networks could not only distinguish patients from HCs but also differentiate patients with different reperfusion conditions. Machine learning models can improve clinical treatment and rehabilitation based on individual imaging information, providing better care plans with more accurate and quantitative prognosis (Zhang et al., 2017). In this study, good discriminant performance (AUC>0.7) was achieved in models, suggesting that FCs within functional networks may have the potential to serve as imaging markers for assisting diagnosis and classification of post-thrombectomy patients.

Several limitations should be considered in this study. Firstly, the findings of subgroup comparisons did not survive multiple comparison correction. Presumably, this may be due to small sample size in each subgroup. Thus, the reproducibility and reliability of our subgroup results need to be examined in future studies by recruiting more patients in each subgroup. Secondly, due to the small sample size of AIS patients with ICA infarctions and ACA infarctions, we did not test whether ICA or ACA infarctions would influence our findings. Therefore, our results need to be further validated in post-thrombectomy AIS patients with different infarction regions, especially in ICA and ACA. Besides, we failed to record the vascular stenosis history of AIS patients, which may potentially affect our findings. Future studies with patients’ histories of vascular stenosis collected could help clarify how the duration and severity of vascular stenosis may influence the functional reorganization in thrombectomy-treated AIS patients. Moreover, while brain networks tend to independently carry out their respective functions, there is also a cooperation among advanced networks (Zhao et al., 2018). Complex network analysis can reflect various properties of complex systems by describing topological features of networks. Therefore, future research efforts may explore the topological properties of functional networks in post-thrombectomy patients.

Funding

This work was supported by the Zhejiang Provincial Natural Science Foundation of China (No. LGJ22H180001); Zhejiang Medical and Health Science and Technology Project (No. 2021KY249).

Author contributions

Conception and design of the study: Z.D. and Y.L. Acquisition and analysis of data: J.X., T.L, L.W., X.L, W.C., L.Z., G.N., Y.Z., Z.D. and Y.L. Drafting the manuscript or figures: T.L., L.W. and Y.L.

CRediT authorship contribution statement

Tongyue Li: Writing – review & editing, Writing – original draft, Visualization, Formal analysis. Jiaona Xu: Resources, Methodology, Investigation, Data curation. Luoyu Wang: Methodology, Formal analysis. Kang Xu: Methodology, Formal analysis. Weiwei Chen: Methodology, Formal analysis. Liqing Zhang: Formal analysis. Guozhong Niu: Supervision, Resources, Project administration. Yu Zhang: Formal analysis. Zhongxiang Ding: Supervision, Resources, Project administration, Methodology. Yating Lv: Writing – review & editing, Supervision, Project administration, Methodology, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.nicl.2024.103648.

Contributor Information

Zhongxiang Ding, Email: hangzhoudzx73@126.com.

Yating Lv, Email: lvyating@hznu.edu.cn.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Supplementary Data 1
mmc1.docx (85.2KB, docx)

Data availability

Data will be made available on request.

References

  1. Albers G.W., Marks M.P., Kemp S., et al. Thrombectomy for stroke at 6 to 16 hours with selection by perfusion imaging. N. Engl. J. Med. 2018;378(8):708–718. doi: 10.1056/NEJMoa1713973. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bakasa W., Viriri S. Stacked ensemble deep learning for pancreas cancer classification using extreme gradient boosting. Front. Artif. Intell. 2023;6 doi: 10.3389/frai.2023.1232640. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Barrett A.M., Boukrina O., Saleh S. Ventral attention and motor network connectivity is relevant to functional impairment in spatial neglect after right brain stroke. Brain Cogn. 2019;129:16–24. doi: 10.1016/j.bandc.2018.11.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Berkhemer O.A., Fransen P.S.S., Beumer D., et al. A randomized trial of intraarterial treatment for acute ischemic stroke. N. Engl. J. Med. 2015;372(1):11–20. doi: 10.1056/NEJMoa1411587. [DOI] [PubMed] [Google Scholar]
  5. Bian R., Huo M., Liu W., et al. Connectomics underlying motor functional outcomes in the acute period following stroke. Front. Aging Neurosci. 2023;15 doi: 10.3389/fnagi.2023.1131415. Accessed December 25, 2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Biswal B., Zerrin Yetkin F., Haughton V.M., Hyde J.S. Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magnet. Reson. Med. 1995;34(4):537–541. doi: 10.1002/mrm.1910340409. [DOI] [PubMed] [Google Scholar]
  7. Bonkhoff A.K., Espinoza F.A., Gazula H., et al. Acute ischaemic stroke alters the brain’s preference for distinct dynamic connectivity states. Brain. 2020;143(5):1525–1540. doi: 10.1093/brain/awaa101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bonkhoff A.K., Schirmer M.D., Bretzner M., et al. Abnormal dynamic functional connectivity is linked to recovery after acute ischemic stroke. Hum. Brain Mapp. 2021 doi: 10.1002/hbm.25366. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brown C.A., Schmitt F.A., Smith C.D., Gold B.T. Distinct patterns of default mode and executive control network circuitry contribute to present and future executive function in older adults. Neuroimage. 2019;195:320–332. doi: 10.1016/j.neuroimage.2019.03.073. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Buckner R.L., Andrews-Hanna J.R., Schacter D.L. The brain’s default network. Ann. N. Y. Acad. Sci. 2008;1124(1):1–38. doi: 10.1196/annals.1440.011. [DOI] [PubMed] [Google Scholar]
  11. Campbell B.C.V., Mitchell P.J., Kleinig T.J., et al. Endovascular therapy for ischemic stroke with perfusion-imaging selection. N. Engl. J. Med. 2015;372(11):1009–1018. doi: 10.1056/NEJMoa1414792. [DOI] [PubMed] [Google Scholar]
  12. Chen H., Shi M., Zhang H., et al. Different patterns of functional connectivity alterations within the default-mode network and sensorimotor network in basal ganglia and pontine stroke. Med. Sci. Monit. 2019;25:9585–9593. doi: 10.12659/MSM.918185. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Delaveau P., Jabourian M., Lemogne C., et al. EPA-1086 - Dorsomedial and precuneus activation during self referential processing predict long term remission to agomelatine in major depression. Eur. Psychiatry. 2014;29:1. [Google Scholar]
  14. Ding X., Li C.Y., Wang Q.S., et al. Patterns in default-mode network connectivity for determining outcomes in cognitive function in acute stroke patients. Neuroscience. 2014;277:637–646. doi: 10.1016/j.neuroscience.2014.07.060. [DOI] [PubMed] [Google Scholar]
  15. Frings L., Dressel K., Abel S., et al. Reduced precuneus deactivation during object naming in patients with mild cognitive impairment, Alzheimer’s disease, and frontotemporal lobar degeneration. Dement. Geriatr. Cogn. Disord. 2010;30(4):334–343. doi: 10.1159/000320991. [DOI] [PubMed] [Google Scholar]
  16. Hillary F.G., Roman C.A., Venkatesan U., Rajtmajer S.M., Bajo R., Castellanos N.D. Hyperconnectivity is a fundamental response to neurological disruption. Neuropsychology. 2015;29(1):59–75. doi: 10.1037/neu0000110. [DOI] [PubMed] [Google Scholar]
  17. Hong W., Du Y., Xu R., et al. Altered cerebellar functional connectivity in chronic subcortical stroke patients. Front. Hum. Neurosci. 2022;16 doi: 10.3389/fnhum.2022.1046378. Accessed December 25, 2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Jia X.Z., Wang J., Sun H.Y., et al. RESTplus: an improved toolkit for resting-state functional magnetic resonance imaging data processing. Sci. Bull. 2019;64(14):953–954. doi: 10.1016/j.scib.2019.05.008. [DOI] [PubMed] [Google Scholar]
  19. Kito S., Pascual-Marqui R.D., Hasegawa T., Koga Y. High-frequency left prefrontal transcranial magnetic stimulation modulates resting EEG functional connectivity for gamma band between the left dorsolateral prefrontal cortex and precuneus in depression. Brain Stimul. 2014;7(1):145–146. doi: 10.1016/j.brs.2013.09.006. [DOI] [PubMed] [Google Scholar]
  20. LaCroix A.N., Baxter L.C., Rogalsky C. Auditory attention following a left hemisphere stroke: comparisons of alerting, orienting, and executive control performance using an auditory attention network test. Auditory Percept. Cogn. 2020;3(4):238–251. doi: 10.1080/25742442.2021.1922988. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Leonards U., Sunaert S., Van Hecke P., Orban G.A. Attention mechanisms in visual search—an fMRI study. J. Cogn. Neurosci. 2000;12(Supplement 2):61–75. doi: 10.1162/089892900564073. [DOI] [PubMed] [Google Scholar]
  22. Liu W., Liu L., Cheng X., et al. Functional integrity of executive control network contributed to retained executive abilities in mild cognitive impairment. Front. Aging Neurosci. 2021;13 doi: 10.3389/fnagi.2021.710172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lv Y., Margulies D.S., Craddock R.C., et al. Identifying the perfusion deficit in acute stroke with resting-state functional magnetic resonance imaging. Ann. Neurol. 2013;73(1):136–140. doi: 10.1002/ana.23763. [DOI] [PubMed] [Google Scholar]
  24. Mantini D., Vanduffel W. Emerging roles of the brain’s default network. Neuroscientist. 2013;19(1):76–87. doi: 10.1177/1073858412446202. [DOI] [PubMed] [Google Scholar]
  25. Mantini D., Perrucci M.G., Del Gratta C., Romani G.L., Corbetta M. Electrophysiological signatures of resting state networks in the human brain. Proc. Natl. Acad. Sci. 2007;104(32):13170–13175. doi: 10.1073/pnas.0700668104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mao R., Zong N., Hu Y., Chen Y., Xu Y. Neuronal death mechanisms and therapeutic strategy in ischemic stroke. Neurosci. Bull. 2022;38(10):1229–1247. doi: 10.1007/s12264-022-00859-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Margulies D.S., Vincent J.L., Kelly C., et al. Precuneus shares intrinsic functional architecture in humans and monkeys. Proc. Natl. Acad. Sci. 2009;106(47):20069–20074. doi: 10.1073/pnas.0905314106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Nachev P., Coulthard E., Jäger H.R., Kennard C., Husain M. Enantiomorphic normalization of focally lesioned brains. Neuroimage. 2008;39(3):1215–1226. doi: 10.1016/j.neuroimage.2007.10.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Nogueira R.G., Jadhav A.P., Haussen D.C., et al. Thrombectomy 6 to 24 hours after stroke with a mismatch between deficit and infarct. N. Engl. J. Med. 2018;378(1):11–21. doi: 10.1056/NEJMoa1706442. [DOI] [PubMed] [Google Scholar]
  30. Passingham R.E., Chung A., Goparaju B., Cowey A., Vaina L.M. Using action understanding to understand the left inferior parietal cortex in the human brain. Brain Res. 2014 doi: 10.1016/j.brainres.2014.07.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Perez S.E., He B., Nadeem M., et al. Resilience of precuneus neurotrophic signaling pathways despite amyloid pathology in prodromal Alzheimer’s disease. Biol. Psychiatry. 2015;77(8):693–703. doi: 10.1016/j.biopsych.2013.12.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Posner M.I., Walker J.A., Friedrich F.J., Rafal R.D. Effects of parietal injury on covert orienting of attention. J. Neurosci. 1984;4(7):1863–1874. doi: 10.1523/JNEUROSCI.04-07-01863.1984. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Puig J., Blasco G., Alberich-Bayarri A., et al. Resting-state functional connectivity magnetic resonance imaging and outcome after acute stroke. Stroke. 2018;49(10):2353–2360. doi: 10.1161/STROKEAHA.118.021319. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Rajaraman S., Jaeger S., Antani S.K. Performance evaluation of deep neural ensembles toward malaria parasite detection in thin-blood smear images. PeerJ. 2019;7:e6977. doi: 10.7717/peerj.6977. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Schaefer A., Kong R., Gordon E.M., et al. Local-global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI. Cereb. Cortex. 2018;28(9):3095–3114. doi: 10.1093/cercor/bhx179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Solihin M.I., Yuan C.J., Hong W.S., et al. Spectroscopy data calibration using stacked ensemble machine learning. IIUM Eng. J. 2024;25(1):208–224. [Google Scholar]
  37. Spreng R.N., Sepulcre J., Turner G.R., Stevens W.D., Schacter D.L. Intrinsic architecture underlying the relations among the default, dorsal attention, and frontoparietal control networks of the human brain. J. Cogn. Neurosci. 2013;25(1):74–86. doi: 10.1162/jocn_a_00281. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Tuladhar A.M., Snaphaan L., Shumskaya E., et al. Default mode network connectivity in stroke patients. PLoS One. 2013;8(6):e66556. doi: 10.1371/journal.pone.0066556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Van den Heuvel M.P., Hulshoff Pol H.E. Exploring the brain network: A review on resting-state fMRI functional connectivity. Eur. Neuropsychopharmacol. 2010;20(8):519–534. doi: 10.1016/j.euroneuro.2010.03.008. [DOI] [PubMed] [Google Scholar]
  40. Yeo B.T., Krienen F.M., Sepulcre J., et al. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J. Neurophysiol. 2011;106(3):1125–1165. doi: 10.1152/jn.00338.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Zhang H., Di X., Rypma B., Yang H., Meng C., Biswal B. Interaction between memory load and experimental design on brain connectivity and network topology. Neurosci. Bull. 2023;39(4):631–644. doi: 10.1007/s12264-022-00982-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Zhang B., He X., Ouyang F., et al. Radiomic machine-learning classifiers for prognostic biomarkers of advanced nasopharyngeal carcinoma. Cancer Lett. 2017;403:21–27. doi: 10.1016/j.canlet.2017.06.004. [DOI] [PubMed] [Google Scholar]
  43. Zhang S., Li C., Shan R. Functional connectivity mapping of the human precuneus by resting state fMRI. Neuroimage. 2012;59(4):3548–3562. doi: 10.1016/j.neuroimage.2011.11.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Zhang P., Wang J., Xu Q., Song Z., Dai J., Wang J. Altered functional connectivity in post-ischemic stroke depression: A resting-state functional magnetic resonance imaging study. Eur. J. Radiol. 2018;100:156–165. doi: 10.1016/j.ejrad.2018.01.003. [DOI] [PubMed] [Google Scholar]
  45. Zhao Z., Wu J., Fan M., et al. Altered intra- and inter-network functional coupling of resting-state networks associated with motor dysfunction in stroke. Hum. Brain Mapp. 2018;39(8):3388–3397. doi: 10.1002/hbm.24183. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data 1
mmc1.docx (85.2KB, docx)

Data Availability Statement

Data will be made available on request.

Articles from NeuroImage : Clinical are provided here courtesy of Elsevier

RESOURCES