Thank you for forwarding the thoughtful letter regarding our paper “Altered resting-state electroencephalogram microstate characteristics in stroke patients” published in the Journal of Integrative Neuroscience. We sincerely appreciate your interest in our work and the time you took to raise these important questions. We welcome the opportunity to address the concerns and questions raised, as they help clarify the study’s methods and also highlight areas where future research could be refined. We have carefully considered the comments and provide the following responses:

Point 1: “…the electrical activity of the cortex can be highly dependent on the location and volume of the stroke [1]. Subcortical strokes can produce a completely different resting electroencephalogram (EEG) pattern from strokes that involve the cortex. Brainstem strokes may present completely differently on the resting EEG from supratentorial strokes involving the cortex. Also, the extent of the stroke nucleus can strongly influence the cortical electrical activity [2].”

Author response to Point 1:

Research published in The New England Journal of Medicine has revealed that similar clinical symptoms can result from lesions in distinct brain locations that share the same brain network [3]. The correlation between the location of lesions and clinical symptoms is complex. Our study aimed to address the same clinical symptoms. Although the location of lesions may vary, it can also lead to similar changes in brain function within a single brain network (as shown in Table 1).

It is more important that the current understanding of how different lesion locations in stroke patients affect microstate dynamics remains limited. Previous research has revealed that the transition probabilities from microstate A and microstate D to microstate C significantly decreased whereas the transition probabilities from microstate A to microstate D and from microstate D to microstate B significantly increased in patients with acute brainstem infarction relative to those in healthy individuals, suggesting an increased tendency to activate microstate B and a corresponding rise in the extent of neural activity over time [1]. The previous studies of simultaneous EEG-functional magnetic resonance imaging (fMRI) have shown a significant correlation between microstate B and the visual network (VN) [2, 4, 5]. The previous studies also revealed that microstate A is related to the sensorimotor network (SMN), microstate C is associated with the salience network (SN), and microstate D is connected with the dorsal attention network (DAN). However, there is still a lack of exploration into the different effects of lesion location (cortical, subcortical, brainstem) on brain dynamics in stroke patients.

This current study included 24 stroke patients, among whom 4 had cortical lesions, 10 had subcortical lesions, 1 had brainstem lesions, 5 had cortical and subcortical lesions, and 4 had subcortical and brainstem lesions. The study did not involve subgroup analyses based on lesion location, which is a limitation. Thus, we are currently unable to determine whether different lesion locations led to distinct EEG microstate patterns. This will be a key question that can be explored from a functional network perspective in our future research. We plan to expand sample size for more detailed stratified analyses. Additionally, if requested, we can perform subgroup analyses with the current data, although the small sample size may limit the statistical power of such analyses.

Point 2: “…whether only patients with an ischemic stroke or also with a hemorrhagic stroke were included, was not mentioned. If patients with hemorrhagic stroke were also included, whether the hemorrhage was accompanied by edema or not should be mentioned, because the degree of perifocal edema can strongly influence cortical electrical activity [4]. It is also important to know how many patients with hemorrhagic stroke had or did not have intraventricular intrusion.”

Author response to Point 2:

As explicitly stated in Section 2.1 Participants, our study enrolled patients with “first-ever ischemic or hemorrhagic stroke”. The study did not involve separate subgroup analyses for ischemic or hemorrhagic stroke, because our primary research objective was to elucidate the global patterns of cortical electrophysiological alterations induced by post-stroke motor dysfunction. Our research enrolled 5 patients with hemorrhage (sub 02, 07, 12, 22, and 23), with onset times of 12 mo, 11 mo, 10 mo, 7 mo, and 8 mo, respectively. We traced the medical history of the patients and confirmed that the most recent neuroimaging prior to study enrollment showed no evidence of perifocal edema or intraventricular intrusion, which suggested that their effect on electrophysiological measurements was likely minimal in our study. As noted by the reader, we acknowledge that perilesional edema may indeed influence electrophysiological activity. We agree that future investigations could incorporate stroke-phase classification, pathological subtypes, and multimodal neuroimaging data (e.g., magnetic resonance imaging (MRI)/computed tomography (CT)) to facilitate more nuanced subgroup analyses.

Point 3: “…the latency between the acute stroke and the EEG recordings was not measured, or was not included in the analysis. The cortical electrical activity may strongly depend on the “age” of the stroke [5]. The “older” the stroke, the more likely it is that cortical activity, and thus EEG activity, will recover. Electrical activity may also depend on the response to stroke rehabilitation. Patients who have full functional and structural recovery may be associated with normal electrical activity, compared to patients who do not fully recover. Therefore, the final outcome of the 19 patients must be included in the analysis.”

Author response to Point 3:

This is a significant issue. As presented in Table 1 of our response, we have documented the latency between acute stroke and the EEG recordings, which ranged from 1 to 12 mo across participants. Table 1 also details motor function of the affected limb in stroke patients using the Fugl-Meyer assessment (FMA) and the action research arm test (ARAT) scales. We explored correlations between temporal characteristics (duration, occurrence, and coverage) and transition probabilities (TP) of four microstates and clinical outcomes, including the FMA and ARAT scores, in stroke patients. We found that the TP from microstate A to microstate D had a significant positive correlation with the Fugl-Meyer assessment of lower extremity (FMA-LE) scores, which did not survive false discovery rate (FDR) adjustment. This indicated a trend where the higher the TP from microstate A to D was, the higher the patient’s FMA-LE scores would be, revealing that these networks were continuously switched. Although all patients showed motor dysfunction based on the scoring criteria and none achieved full recovery, the varying degrees of dysfunction may still have been a source of EEG heterogeneity. Additionally, as Fox has highlighted, lesion network mapping focuses on the spatial component of lesion-induced symptoms, but the temporal component may be equally significant [3]. Therefore, future studies can employ subgroup analyses based on severity levels of dysfunction and longitudinal follow-up designs to further elucidate the relationship between latency, deficits and EEG activity.

Point 4: “….a previous stroke was an exclusion criterion, but the exclusion was based on history only, suggesting that patients with a previous subclinical stroke on imaging were included in the study. This issue should be clarified.”

Author response to Point 4:

Although subclinical stroke (also termed silent cerebral infarction) presents no neurological symptoms, emerging evidence has suggested that it may still exert effects on electrophysiological activity of brain [6]. Yang et al. [6], found that patients with silent cerebral infarction had lower P300 amplitude and longer latency than did healthy individuals. Therefore, we fully concur with the reader’s comment. A limitation of our research was the absence of neuroimaging (e.g., MRI) to exclude participants with subclinical stroke, which may have ​​had a confounding effect​​ on the observed outcomes. Thus, it is significant to incorporate neuroimaging techniques​​ (e.g., MRI) to screen for and exclude individuals with a history of silent cerebral infarction in order to enhance methodological rigor in future investigations.

Point 5: “…stroke may be manifested not only by limb weakness but also by dysarthria, aphasia, or dysphagia. However, the FMA does not record and assess these features, which is why the severity of the deficits may have been misclassified. Furthermore, the ARTA test is inadequate to assess the bulbar symptoms of stroke patients.”

Author response to Point 5:

The diversity of clinical manifestations of stroke, such as dysarthria, aphasia, or dysphagia, may indeed influence the comprehensive evaluation of global neurological function in patients. However, our study was specifically designed to investigate the underlying mechanism of brain motor reorganization, with a primary focus on the motor function of hemiplegic limbs and EEG activity. Thus, we selected two movement-specific assessment scale: FMA and ARAT. FMA is considered by many in the field of stroke rehabilitation to be one of the most comprehensive quantitative measures of motor impairment after stroke [7]. ARAT is a reliable, valid measure of arm motor status after stroke and has established value for characterizing clinical states [8]. Since all enrolled patients in this study presented with limb-movement disorders as the primary clinical manifestation (as explicitly specified in the inclusion criteria), prioritizing motor function assessment was justified. However, we must acknowledge that the lack of assessment for other neurological symptoms (e.g., dysarthria, aphasia, or dysphagia) may have resulted in an incomplete characterization of patients’ overall neurological deficits. Future studies could incorporate comprehensive scales or domain-specific assessment tools, such as the Western Aphasia Battery (WAB) and the Water-swallowing test (WST), to fully delineate the spectrum of neurological impairments in stroke patients [9, 10].

Point 6: “…patients taking anti-seizure and antipsychotic drugs were excluded, but not patients taking sedatives or hypnotics. Since the latter can greatly reduce cortical activity at rest, we should know how many of the patients suffered from insomnia, anxiety, or depression, and required appropriate medication. Those patients also need to be excluded from the analysis.”

Author response to Point 6:

It is indeed a methodological limitation of our study that we did not exclude participants using sedatives or hypnotics. Specifically, we neither systematically collected data on such medication use nor assessed related psychological symptoms (e.g., insomnia, anxiety, or depression). Previous studies examining EEG activity demonstrated that sedatives or hypnotics can affect EEG activity [11, 12]. For example, in a report that evaluated the power spectral profiles of various sleep agents in healthy individuals, zolpidem reduced activity in the lower frequency bands and increased activity in the middle frequency bands, whereas suvorexant had no effects in any of the frequency bands [13]. To address these limitations, future studies should implement stricter screening protocols to exclude participants using sedative/hypnotic medications, while incorporating standardized psychometric assessments to comprehensively evaluate psychological status.