Assessment of cerebrovascular reactivity in middle cerebral artery stenosis using non-hypercapnic resting-state fMRI: a potential biomarker for hemodynamic alterations Authors: Liqing Zhang, Luoyu Wang, Xue Tang, Yidi Zhu, Rong Wang, Han Zhang, Zhongxiang Ding Published: 2026 ABSTRACT AbstractBackgroundCerebrovascular reactivity(CVR), a key indicator of cerebrovascular reserve, is crucial for evaluating cerebrovascular pathophysiology. This study employed resting-state MRI (rs-MRI) to assess CVR alteration in patients with unilateral middle cerebral artery stenosis or occlusion (MCA-S) under non-hypercapnic conditions, comparing them with healthy controls.MethodsA total of 41 patients with unilateral MCA-S and 50 age-, sex-, and education-matched normal controls (NC). All underwent rs-MRI and neuropsychological assessments. CVR was derived from rs-fMRI frequency band signals, and t-test was conducted to obtain the CVR-differentiated brain regions. Exploratory seed-based FC analysis was further performed to characterize the network context of regions showing altered CVR. Partial correlation analyses explored relationships between these differential brain regions and both neuropsychological assessments and clinical indicators. Discriminative performance of the CVR-related metric between MCA-S and controls was evaluated using receiver operating characteristic (ROC) curves.ResultsCompared with the NC group, patients with MCA-S exhibited increased CVR in the contralesional Cerebellum Crus1 (CC1) and decreased iCVR in the ipsilesional postcentral gyrus (PoCG). When contralesional CC1 served as regions of interest (ROIs), increased FC was observed in the ipsilesional middle frontal gyrus (MFG) and the contralesional precuneus of MCA-S patients. The partial correlation analysis indicated a positive correlation between the FC of the ipsilesional MFG and anxiety scores (r = 0.404, (95%CI: 0.106, 0.631), P = 0.012, P-FDR = 0.030). Using ipsilesional PoCG as the ROI, MCA-S patients showed significantly decreased FC in ipsilesional PoCG, contralesional precentral gyrus (PreCG), and ipsilesional supplementary motor area. The FC of the contralesional PreCG showed a positive correlation with anxiety scores (r = 0.436, (95%CI: 0.142, 0.658), P = 0.006, P-FDR = 0.030). ROC analysis demonstrated strong diagnostic accuracy for CVR in CC1 (AUC = 0.809) and PoCG (AUC = 0.787), with a combined AUC of 0.866.ConclusionNon-hypercapnic rs-MRI effectively evaluates CVR alterations in MCA-S patients and may serve as a complementary physiological biomarker for characterizing hemodynamic alteration.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12880-026-02503-z.Keywords: Resting-state functional magnetic resonance imaging, Cerebrovascular reactivity, Middle cerebral artery, Stenosis, Hemodynamics Background Middle cerebral artery stenosis or occlusion(MCA-S) represents a significant type of ischemic cerebrovascular disease [1–3]. Ischemic stroke caused by MCA-S most prevalent among Asian populations, particularly Chinese [4, 5], and is a common cause of ischemic stroke in young adults [6]. Although clinical symptoms may not be pronounced, patients with MCA-S exhibit measurable structural and functional brain alterations [7]. In one study of asymptomatic MCA-S patients, Huang et al. [7] reported significant cortical thinning in the left superior parietal lobule in patients with left-sides MCA-S, whereas those with right-sided showed thinning in the right superior parietal lobule and the tail of the right middle frontal gyrus. Additionally, a decreasing trend in functional connectivity (FC) within and between hemispheres was also observed. Diffusion spectrum imaging revealed [8] structural network disruption and white matter fiber damage in patients with unilateral MCA-S without stroke, with more severe fiber connection interruptions in the ipsilesional hemisphere. These findings collectively demonstrated the widespread brain changes associated with MCA-S. Nevertheless, the direct hemodynamic consequences of the vascular stenosis and their functional implications require further investigation. Cerebrovascular reactivity (CVR), a key indicator of hemodynamic and cerebrovascular reserve capacity, is defined as the adaptive dilation or constriction of cerebral arteries and arterioles in response to physiological or pathological stimuli, which maintains stable and adequate cerebral blood flow(9). Intracranial artery stenosis typically disrupts cerebral hemodynamics, impairing cerebral autoregulation and CVR, and consequently elevating the risk of ischemic cerebrovascular events [9, 10]. CVR is assessed using several techniques, including CT [11], functional magnetic resonance imaging (fMRI) [9, 12], positron emission tomography (PET) [13, 14], single-photon emission computed tomography (SPECT) [15], and transcranial Doppler ultrasonography (TCD) [4]. These methods generally involve altering blood CO2 concentration via carbon dioxide inhalation tests, breath-holding tests, or hyperventilation tests, or they induce vasodilation through intravenous acetazolamide injection. By measuring changes in cerebral blood flow and velocity, these methods reflect the cerebrovascular response to such stimuli, thereby providing an assessment of CVR. The meta-analysis by King et al. [16] indicated that internal carotid artery stenosis may increase the risk of ischemic stroke by impairing patients’ CVR. Zhang et al. [17] investigated CVR in 73 patients with symptomatic unilateral MCA stenosis ≥ 50%. Their results demonstrated a strong correlation between impaired CVR and the increased stroke risk, which appears to worsen with greater stenosis severity. Existing techniques for measuring CVR, however, largely relied on hypercapnia-based imaging, with the carbon dioxide inhalation test serving as the reference gold standard. This approach necessitates an MRI-compatible gas delivery system. It also requires considerable operator training and practice to ensure accuracy, making the procedure time-intensive. Breath-holding and hyperventilation tests demand high levels of patient cooperation and suffer from lower success rates [18]. Intravenous acetazolamide administration, meanwhile, may carry potential side effects (such as transient ischemic attacks, possibly linked to steal syndrome). Consequently, these established methods are either procedurally complex, reliant on patient compliance, or incur inherent risks due to their invasive nature. Recent research indicated that resting-state fMRI (rs-fMRI) can estimate CVR without requiring hypercapnic breathing or breath-holding maneuvers [19]. This approach leverages spontaneous respiratory fluctuations to alter CO2 levels, the corresponding BOLD signal fluctuations are captured within a specific frequency band (0.02–0.04 Hz) of rs-fMRI data. Based on prior work [19], spontaneous global BOLD fluctuations in the 0.02–0.04 Hz range are thought to be enriched for respiration-related arterial CO₂ variations and have shown spatial agreement with gas-challenge CVR maps in both healthy individuals and cerebrovascular disease cohorts. Here, we define intrinsic CVR (iCVR) [19, 20] as a resting-state, non-hypercapnic estimate of cerebrovascular reactivity derived from spontaneous low-frequency BOLD fluctuations associated with natural CO₂ variations, rather than as a direct replacement for conventional gas-challenge CVR. This method is non-invasive, minimizes patients discomfort and cooperation, and is well-suited for clinical use. Liu et al. [19] were among the first to adopt this method, validating CVR measurements from both hypercapnic and non-hypercapnic methods in 26 healthy controls and 7 patients with moyamoya disease, with consistent results. Our prior research [21] utilized non-hypercapnic rs-fMRI method to explore the mild cognitive impairment and Alzheimer’s disease, finding that CVR showed promise as a non-invasive biomarker for assessing cognitive impairment, due to its clinical applicability and practicality. In other cerebrovascular diseases, Taneja et al. [22] applied the approach to stroke patients, demonstrating that rs-fMRI can allow repeated CVR assessments in patients with cerebrovascular diseases. However, its application in evaluatinf MCA-S patients has not yet been reported. Therefore, this study employed a non-hypercapnic rs-fMRI to measure CVR in patients with MCA-S and normal controls (NC). We compared CVR between the two groups and explored its correlation with clinical indicators, cognitive function, emotional state, and sleep quality in MCA-S patients. Furthermore, we assessed the role of brain regions with CVR differences in the brain network and performed FC analysis using them as regions of interest (ROIs). Group differences in FC were compared, the relationships between FC and clinical scales. The aim was to investigate the potential of CVR as a novel biomarker in MCA-S, which may inform more proactive clinical management of cerebrovascular disease. Materials and methods Participants This study was approved by the Ethics Committee of the Hangzhou First People’s Hospital (IIT-20230822-0181-01). All methods were conducted by the Declaration of Helsinki, and all participants signed informed consent forms. Between September 2023 and December 2024, a total of 90 patients with MCA-S recruited. The inclusion criteria were as follows: (a) Unilateral MCA-M1 stenosis > 50% confirmed by three-dimensional time-of-flight magnetic resonance angiography (3D TOF-MRA), with the degree of intracranial stenosis calculated using the Warfarin-Aspirin Symptomatic Intracranial Disease (WASID) criteria [5, 17], (b) No lesions on DWI, (c) No overt neurological deficit at the time of MRI examination, (d) Ability to complete MRI examination and obtain qualified MRI images, (e) No history of medication use that may affect cognitive function and (f) Right-handedness. Inclusion criteria for HC group were as follows: (a) No evidence of cerebral infarction, hemorrhage, or tumor on MRI imaging, (b) Absence of motor, auditory and visual system disorders, and no language barriers and normal corrected vision, (c) good physical condition, no history of psychiatric disorders, (d) right-handedness. The exclusion criteria were as follows: (a) Any other cerebrovascular stenosis ≥ 30%, (b) History of severe systemic diseases and neuropsychiatric disorders, (c) Incomplete clinical data or inconsistent image quality, (d) MRI detection of acute or chronic cerebral infarction, hemorrhage, tumors, infectious diseases, or metabolic diseases, (e) Head movement, and (f) Contraindications for MRI examination. NC participants were recruited from the community and were required to have no evidence of cerebral infarction, hemorrhage, tumor, or other major structural brain abnormality on MRI, and no relevant intracranial or cervical arterial stenosis on vascular imaging. Additional exclusion criteria were the same as those applied to the patient group where applicable. The final study cohort consisted of 41 patients with MCA-S and 50 healthy subjects matched for age, gender, and education level from the community as NC group (Fig. 1). Flowchart of the present study. MCA-S, Middle cerebral artery stenosis; NC, Normal control Clinical and laboratory data In addition to participants’ age, gender, and years of education, we collected data on body mass index(BMI), smoking and drinking history, and histories of hypertension, hyperlipidemia, diabetes, and coronary heart disease. Additionally, multiple clinical laboratory indicators were collected, including blood pressure (diastolic blood pressure and systolic blood pressure), blood sugar, blood routine examination (platelet count and C-reactive protein), coagulation function(D dimer and fibrinogen), biochemical indices (apolipoprotein A1, apolipoprotein B, lipoprotein A, total cholesterol, low density lipoprotein, high density lipoprotein and triglyceride). Neuropsychological assessments A psychiatrist evaluated the clinical psychological state of all subjects. The Mini-Mental State Examination (MMSE), Montreal cognitive assessment(MoCA), Self-Rating Depression and Anxiety Scales (SDS and SAS), and Pittsburgh Sleep Quality Index(PSQI) were used to assess cognitive function, depression and anxiety symptoms, and sleep quality, respectively. According to Chinese norms, the cutoff score for the SAS is 50, while the cutoff score for the SDS is 53. The PSQI was used to evaluate overall sleep quality where a score below 5 indicates good sleep quality, 5–8 indicates poor sleep quality and a score exceeding 8 suggests the presence of insomnia or other severe sleep disorders. MRI data acquisition The MRI data were acquired with a 3.0T MRI scanner (SIGNA Premier, GE Healthcare, Waukesha, WI, USA) and a 48-channel phased array head coil. All participants were asked to remain quiet and keep their eyes closed, but not to fall asleep [20, 23, 24]. Subjects were equipped with sponge cushions placed on both sides of the head to minimize head movement, and provided with built-in earplugs and external earmuffs to reduce noise interference and protect hearing. The scanning parameters were as follows: T1-weighted 3D fast spoiled gradient-recalled (3D-T1WI): slice = 192, slice thickness/gap = 1/1 mm, repetition time (TR) = 2616 ms, echo time (TE) = 2.9 ms, inversion time (TI) = 1000 ms, field of view (FOV) = 256 mm × 256 mm, matrix size = 256 × 256, voxel size = 1 mm × 1 mm × 1 mm, flip angle (FA) = 8°. T1-weighted 3D fast spoiled gradient-recalled (3D-T1WI): slice = 192, slice thickness/gap = 1/1 mm, repetition time (TR) = 2616 ms, echo time (TE) = 2.9 ms, inversion time (TI) = 1000 ms, field of view (FOV) = 256 mm × 256 mm, matrix size = 256 × 256, voxel size = 1 mm × 1 mm × 1 mm, flip angle (FA) = 8°. T2 weighted Imaging(T2WI): slice = 20, slice thickness/gap = 6/1 mm, TR = 4753 ms, TE = 120 ms, FOV = 240 mm × 240 mm, matrix size = 352 × 352, voxel size = 0.6 mm × 0.6 mm × 6.0 mm. T2 weighted Imaging(T2WI): slice = 20, slice thickness/gap = 6/1 mm, TR = 4753 ms, TE = 120 ms, FOV = 240 mm × 240 mm, matrix size = 352 × 352, voxel size = 0.6 mm × 0.6 mm × 6.0 mm. T2 fluid attenuated inversion recovery(FLAIR): slice = 20, slice thickness/gap = 6/1 mm, TR = 9000 ms, TE = 100 ms, FOV = 240 mm × 240 mm, matrix size = 288 × 256, voxel size = 0.8 mm × 0.9 mm × 6.0 mm. T2 fluid attenuated inversion recovery(FLAIR): slice = 20, slice thickness/gap = 6/1 mm, TR = 9000 ms, TE = 100 ms, FOV = 240 mm × 240 mm, matrix size = 288 × 256, voxel size = 0.8 mm × 0.9 mm × 6.0 mm. axial diffusion-weighted imaging (DWI): slice = 40, slice thickness/gap = 6/1 mm, TR = 2115 ms, TE = 62 ms, FOV = 240 mm × 240 mm, matrix size = 160 × 168, voxel size = 1.5 mm × 1.4 mm × 6.0 mm, b value = 1000 s/mm2. axial diffusion-weighted imaging (DWI): slice = 40, slice thickness/gap = 6/1 mm, TR = 2115 ms, TE = 62 ms, FOV = 240 mm × 240 mm, matrix size = 160 × 168, voxel size = 1.5 mm × 1.4 mm × 6.0 mm, b value = 1000 s/mm2. axial rs-fMRI: slice = 52, slice thickness = 3 mm, TR = 800ms, TE = 30 ms, FOV = 240 mm × 240 mm, matrix size = 80 × 80, voxel size = 3 mm × 3 mm × 3 mm, flip angle (FA) = 56°, volumes = 600, and parallel acceleration = 4, phase encoding PA direction, and the acquisition duration was 8 min. To perform magnetic susceptibility correction, we also acquired echo-planar imaging data with reversed phase encoding directions using the same acquisition parameters. axial rs-fMRI: slice = 52, slice thickness = 3 mm, TR = 800ms, TE = 30 ms, FOV = 240 mm × 240 mm, matrix size = 80 × 80, voxel size = 3 mm × 3 mm × 3 mm, flip angle (FA) = 56°, volumes = 600, and parallel acceleration = 4, phase encoding PA direction, and the acquisition duration was 8 min. To perform magnetic susceptibility correction, we also acquired echo-planar imaging data with reversed phase encoding directions using the same acquisition parameters. White matter hyperintensity (WMH) analysis Given that WMH may affect cognition and shares vascular risk factors with MCA-S, Fazekas grading was included to evaluate WMH burden as a potential confounding factor. The Fazekas classification acccording to the study [25], and the total score was obtained by adding the scores of deep WMH and periventricular WMH. According to the total score, Fazekas grade 0 was 0; Fazekas grade 1: 1 ~ 2 points; Fazekas 2: 3 ~ 4 points; Fazekas grade 3: 5 ~ 6 points. Fazekas grade was performed independently by two Radiologists (both with more than 10 years of experience in brain MRI diagnosis) who were blinded to all participants’ data. In case of interrater disagreement, the raters met to reach a consent [26]. MRI data processing Image flipping To standardize lesion laterality across patients, following previous studies [20, 23, 27], for patients with right MCA stenosis (23 patients), both the structural T1-weighted images and the resting-state fMRI images were horizontally flipped before preprocessing. The left hemisphere defined as the ipsilesional side and the right hemisphere as the contralesional side. Therefore, all subsequent preprocessing, CVR calculation, and FC analysis were performed in the same standardized orientation. CVR calculation The calculation of CVR was performed using the FSL, ANTs and SPM12 toolboxes, following these steps: (1) Discarding the first 10 time points to ensure subjects’ adaptation to the scanning environment and stability of the MRI signal; (2) Performing head motion correction, and excluding subjects with head motion displacement >3 mm, rotation > 3°, or mean framewise displacement (FD) > 0.5; (3) Correcting spatial distortions and susceptible artifacts; (4) Aligning with structural MRI using tissue boundary maps for precise functional localization; (5) Applying a one-time resampling in each subject’s native space, which aimed to combine motion correction, distortion correction, and structural alignment into a single spline interpolation step, thereby minimizing repeated interpolation and reducing unnecessary image blurring. (6) Spatial normalization to the MNI standard space and resampling to 3 × 3 × 3 mm3 voxels. (7) Spatial smoothing was applied using a 6 mm FWHM isotropic Gaussian kernel to reduce noise and improve the signal-to-noise ratio. (8) Linear detrending and band-pass filtering (0.02–0.04 Hz) were carried out. Based on previous research by Liu et al. [19] global BOLD signal fluctuations within this frequency range are most correlated with natural arterial CO2 variations during spontaneous breathing. (9) A voxel-wise analysis was conducted using the general linear model (GLM), with the reference time course as the independent variable. The results were normalized by dividing by the global mean, yielding the relative CVR index for each voxel. FC calculation based on ROI The ROI-based FC calculation was performed using Data Processing & Analysis of Brain Imaging(DPABI) software as follows: (1) As mentioned above, preprocessing involved discarding initial time points, motion correction, distortion correction, structural alignment, and one-time resampling for functional localization; (2) Utilizing ICA AROMA for noise identification, primarily to detect motion-related artifacts. (3) Conducting a one-time regression on identified noise components, Friston 24 motion parameters, linear trends, and signals from white matter and cerebrospinal fluid to purge confounds from the data. (4) Normalizing based on tissues segmented by ANTs, aligning individual brain images with MNI space according to tissue segmentation, and then resampling to a 3 × 3 × 3 mm3 voxels size to ensure anatomical consistency across subjects for accurate group analysis. (5) Applying a band-pass filter (0.01–0.08 Hz). (6) Using brain regions with statistically significant differences in CVR between the two groups as ROIs, and extracting mean time courses from the preprocessed rs-fMRI data. (7) Calculating Pearson correlation coefficients between the mean time course of each ROI and all other voxels in the brain to obtain correlation maps. (8) Converting correlation maps to z-score maps using Fisher’s r-to-z transformation. Cerebellar anatomical validation To improve the anatomical specificity of the cerebellar finding, we performed a SUIT-based anatomical validation analysis [28]. The suprathreshold cerebellar statistical cluster identified in the main voxel-wise CVR analysis was transformed into SUIT template space and overlaid with the SUIT cerebellar lobular atlas. We then determined whether the cluster peak was located within the SUIT-defined Crus 1 region and quantified the proportion of suprathreshold cluster voxels overlapping with the SUIT-defined Crus I region. Statistical analysis Statistical analysis was conducted with SPSS version 26.0 (SPSS Inc., Chicago, IL, USA), GraphPad Prism 9.0 and MedCalc(version 20.0.27). Cohen’s Kappa test was used to assess the consistency of white matter hyperintensity grading parameters measured by two observers. Categorical data (such as gender) were presented as frequencies and analyzed with the Chi-square (χ2) test. The normality of continuous variables (age, education, and neuropsychological test scores) was evaluated with the Shapiro-Wilk test. Normal distribution data were expressed as mean ± standard deviation (mean ± SD) and were analyzed with two independent sample t-test. Head motion was quantified as framewise displacement (FD) [29], with mean FD compared between groups via two independent sample t-tests. Cerebrospinal fluid and other nuisance signals were regressed out in DPABI using a whole-brain template. Age, gender, years of education, and head motion were included as covariates. Voxel-based two sample t-tests were employed to analyze CVR differences between the two groups. Seed ROIs for the FC analysis were defined from all clusters showing significant between-group differences in the voxel-wise CVR analysis after GRF correction, and no additional atlas-wide ROI screening procedure was performed. Where group differences were detected, multiple comparison corrections were performed using Gaussian random field theory (GRF), with thresholds set at voxel P < 0.001 and cluster P < 0.05. To evaluate the robustness of the voxel-wise CVR findings, we additionally performed permutation-based Threshold-Free Cluster Enhancement (TFCE) [30] analysis with 5,000 permutations and family-wise error correction at P < 0.05. Partial correlations, with gender, age, and years of education as covariates, assessed the relationships between neuropsychological test scores (MMSE, MoCA, SAS, SDS, and PSQI) and the significant CVR and FC measures, as well as between clinical indicators and significant CVR. To assess the specificity of the observed brain-cognition relationships, partial correlation analyses were performed in the HC group, and also included age, gender, years of education, and head motion as covariates. These partial correlations were corrected using False Discovery Rate (FDR, P < 0.05) for multiple comparisons. Receiver Operating Characteristic (ROC) curve analysis evaluated discriminative performance of the CVR-related metric between MCA-S and controls, recording the area under the curve (AUC), 95% CI, specificity and sensitivity. To explicitly assess the influence of psychometric variables, the voxel-wise two-sample t-test for CVR was repeated as a secondary sensitivity analysis with age, sex, years of education, head motion, SAS, SDS, and PSQI included as covariates. In addition, the partial correlation analyses between significant FC measures and SAS scores were repeated after additionally controlling for SDS and PSQI. Results Demographics and clinical characteristics Table 1 displayed the demographic and clinical characteristics of the participants. This study enrolled 41 patients with MCA-S (25 males and 16 females, aged 58.51 ± 14.82 years) and 50 age-, sex-, and education-matched normal controls (30 males and 20 females, aged 59.38 ± 12.25 years). All MCA-S patients exhibited unilateral onset (23 right and 18 left). The MCA-S group had a higher BMI and a higher proportion of smoking, drinking, hypertension, hyperlipidemia, diabetes, and coronary heart disease compared to the NC group. Notably, hypertension showed a significant difference between the two groups (P = 0.010). The consistency analysis between two observers for Fazekas grading yielded a Cohen’s kappa of 0.79. No significant differences were observed in white matter hyperintensity grading between the two groups. Compared to the NC group, the MCA-S group had lower scores on both MMSE and MoCA, with a statistically significant difference in MMSE scores between the two groups (P = 0.003). However, there were no statistically significant differences in anxiety, depression, or sleep status between the two groups. No significant difference in mean FD was observed between the MCA-S and control groups (MCA-S: 0.141 ± 0.062 mm; NC: 0.131 ± 0.090 mm; P = 0.311). Table 1Demographic and clinical characteristics of MCA-S and NC groupsMCA-S (n = 41)NC (n = 50)χ2/tP valueGender Male25300.0090.925 Female1620Age(years)58.51 ± 14.8259.38 ± 12.252.0810.153Education6.80 ± 4.118.30 ± 5.231.5840.211Fazekas grade5.3020.141031112631295333BMI24.85 ± 2.5223.71 ± 2.680.0010.976Smoking, n(%)19(46%)18(36%)0.9990.318Drinking, n(%)14(34%)14(28%)0.4000.527Hypertension, n(%)32(78%)26(52%) 6.613 0.010 Hyperlipidemia, n(%)20(49%)16(32%)2.6530.103Diabetes, n(%)15(37%)10(20%)3.1100.078Coronary heart disease, n(%)5(12%)9(18%)0.5830.445MMSE*26.46 ± 4.9428.02 ± 2.32 9.307 0.003 MoCA*23.86 ± 6.3824.87 ± 4.402,5350.115SAS*30.16 ± 6.1429.26 ± 5.371.3200.254SDS*33.32 ± 7.6531.40 ± 5.732.4180.124PSQI*6.51 ± 4.155.60 ± 3.571.4900.226MCA-S, Middle cerebral artery stenosis; MMSE, Mini-mental state examination; MoCA, Montreal cognitive assessment; NC, Normal control; PSQI, Pittsburgh sleep quality index; SAS, Self rating anxiety scale; SDS, Self rating depression scale Demographic and clinical characteristics of MCA-S and NC groups MCA-S, Middle cerebral artery stenosis; MMSE, Mini-mental state examination; MoCA, Montreal cognitive assessment; NC, Normal control; PSQI, Pittsburgh sleep quality index; SAS, Self rating anxiety scale; SDS, Self rating depression scale CVR between MCA-S and NC groups The results showed that compared with the NC group, MCA-S patients showed increased CVR in the contralesional Cerebellum_Crus1 (CC1) (peak Montreal Neurological Institute (MNI) coordinates: 27, -84, -33) and decreased CVR in the ipsilesional postcentral gyrus (PoCG) (peak MNI coordinates: -54, -12, 48) (P < 0.001, cluster P < 0.05) (Fig. 2; Table 2). After additionally including SAS, SDS, and PSQI as covariates, the spatial pattern of CVR differences remained consistent with the primary analysis, including increased CVR in contralesional CC1 and decreased CVR in the ipsilesional PoCG (Figure S2 and Table S2). Differences in CVR between MCA-S and NC groups. Post hoc comparisons of t test. Compared to NC, increased CVR was observed in the contralateral Cerebellum Crus1(Cerebellum Crus1 CON) and decreased CVR was found in the ipsilateral Postcentral gyrus (Postcentral IL) in MCA-S group (voxel P < 0.001, cluster P < 0.05, GRF correction.*** indicates a significant level of P < 0.001, Bonferroni correction). CON, Contralateral; CVR, Cerebrovascular reactivity; IL, ipsilateral; MCA-S, Middle cerebral artery stenosis; NC, Normal control Table 2Significant differences in the CVR between MCA-S and NC groupsBrain regionsPeak MNI coordinatesCluster size(voxels)Peak intensity P XYZCerebellum Crus1 CON27-84-33414.523< 0.001Postcentral IL-54-124852-5.4896< 0.001CON, Contralesional; CVR, Cerebrovascular reactivity; IL, ipsilesional; MCA-S, Middle cerebral artery stenosis; MNI, Montreal neurological institute; NC, Normal control Significant differences in the CVR between MCA-S and NC groups CON, Contralesional; CVR, Cerebrovascular reactivity; IL, ipsilesional; MCA-S, Middle cerebral artery stenosis; MNI, Montreal neurological institute; NC, Normal control Correlation analysis between significant CVR and scales、clinical indicators The results of partial correlation analysis showed a positive correlation between the CVR of the contralesional CC1 and depression scores (r = 0.361, (95%CI: 0.058, 0.599), P = 0.026, P-FDR = 0.520). Additionally, the CVR of the ipsilesional PoCG had a negatively correlation with C-reactive protein(CRP) (r=-0.342, (95%CI: -0.585, -0.037), P = 0.035, P-FDR = 0.520) (Table 3, Figure S1). However, these correlations did not survive FDR correction for multiple comparisons (P-FDR > 0.05). No significant correlations between the different brain regions of CVR and all scales in NC group (Table S1). Table 3The correlation coefficients and the corresponding P values between CVR and scales, Clinical indicatorsBrain regionsCerebellum Crus1 CONPostcentral IL r P P-FDR r P P-FDRMMSE0.1120.5340.871-0.2920.0990.444MoCA0.0970.5910.871-0.3020.0880.444SAS0.2560.1510.8710.0580.7500.929SDS 0.361 0.026 0.5200.0350.8470.941PSQI0.2060.2490.871-0.0480.7900.929Diastolic blood pressure-0.1170.4850.871-0.2810.0870.444Systolic blood pressure-0.0610.7150.888-0.2190.1870.534BMI0.1880.2570.871-0.2310.1620.534Blood Sugar0.1940.2440.8710.0790.6360.929Apolipoprotein A1-0.0450.7900.888-0.0900.5890.929Apolipoprotein B0.0660.6940.888-0.1420.3950.929Lipoprotein A-0.0850.6100.871-0.0100.9510.951Total cholesterol0.1260.4510.871-0.0450.7890.929Low density lipoprotein0.1220.4660.871-0.0620.7090.929High density lipoprotein-0.0180.9140.914-0.0190.9080.951Triglyceride0.1160.4870.8710.0550.7440.929Platelet0.0280.8680.914-0.1300.4370.929C-reactive protein-0.1320.4300.871 -0.342 0.035 0.444D dimer0.0430.7990.888-0.0770.6470.929Fibrinogen0.0920.5830.871-0.2630.1110.444CON, Contralesional; CVR, Cerebrovascular reactivity; FDR, False Discovery Rate; IL, Ipsilesional; MCA-S, Middle cerebral artery stenosis; MMSE, Mini-mental state examination; MoCA, Montreal cognitive assessment; NC, Normal control; PSQI, Pittsburgh sleep quality index; SAS, Self rating anxiety scale; SDS, Self rating depression scale The correlation coefficients and the corresponding P values between CVR and scales, Clinical indicators CON, Contralesional; CVR, Cerebrovascular reactivity; FDR, False Discovery Rate; IL, Ipsilesional; MCA-S, Middle cerebral artery stenosis; MMSE, Mini-mental state examination; MoCA, Montreal cognitive assessment; NC, Normal control; PSQI, Pittsburgh sleep quality index; SAS, Self rating anxiety scale; SDS, Self rating depression scale FC between MCA-S and NC groups The results indicated significant CVR changes in the contralesional CC1 and the ipsilesional PoCG between the MCA-S and the NC groups. These two brain regions were defined as the regions of interest (ROIs) for FC analysis. When using the contralesional CC1 as the ROI, the results revealed compared to NC group, MCA-S group had increased FC in the ipsilesional Frontal_Mid (MFG) (peak MNI coordinates: -36, 48, 0) and the contralesional Precuneus(PCUN) (peak MNI coordinates: 3, -75, 48) (P < 0.001, cluster P < 0.05) (Fig. 3; Table 4). Cerebellum Crus1 CON-ROI FC results between MCA-S and NC groups. Post hoc comparisons of t test. Compared to NC, increased FC was observed in the ipsilateral Frontal Mid (Frontal Mid IL) and contralateral Precuneus (Precuneus CON) in MCA-S group (voxel P < 0.001, cluster P < 0.05, GRF correction.*** indicates a significant level of P < 0.001, Bonferroni correction). CON, Contralateral; FC, Functional connectivity; IL, ipsilateral; MCA-S, Middle cerebral artery stenosis; NC, Normal control Table 4Significant differences in the FC between MCA-S and NC groupsROIBrain regionsPeak MNI coordinatesCluster size(voxels)Peak intensity P XYZCerebellum Crus1 CONFrontal Mid IL-36480534.757< 0.001Precuneus CON3-75483525.3006< 0.001Postcentral ILPostcentral IL-21-3069205-4.7398< 0.001Precentral CON63627286-5.1101< 0.001Supp Motor Area IL0-126090-4.489< 0.001CON, Contralesional; FC, Functional connectivity; IL, Ipsilesional; MCA-S, Middle cerebral artery stenosis; MNI, Montreal neurological institute; NC, Normal control; ROI, Regions of interest Significant differences in the FC between MCA-S and NC groups CON, Contralesional; FC, Functional connectivity; IL, Ipsilesional; MCA-S, Middle cerebral artery stenosis; MNI, Montreal neurological institute; NC, Normal control; ROI, Regions of interest When using the ipsilesional PoCG as the ROI, the results demonstrated a significant decrease in FC in the ipsilesional PoCG (peak MNI coordinates: -21, -30, 69), the contralesional Postcentral(PreCG) (peak MNI coordinates: 63, 6, 27), and the ipsilesional Supp_Motor_Area (SMA) (peak MNI coordinates: 0, -12, 60) (P < 0.001, cluster P < 0.05) (Fig. 4; Table 4). Postcentral IL-ROI FC results between MCA-S and NC groups. Post hoc comparisons of t test. Compared to NC, decreased FC was found in the ipsilateral Postcentral (Postcentral IL), Contralateral Precentral (Precentral CON) and ipsilateral Supp Motor Area(Supp Motor Area IL) in MCA-S group (voxel P < 0.001, cluster P < 0.05, GRF correction.*** indicates a significant level of P < 0.001, Bonferroni correction). CON, Contralateral; FC, Functional connectivity; IL, ipsilateral; MCA-S, Middle cerebral artery stenosis; NC, Normal control Correlation analysis between significant FC and scales Furthermore, the partial correlation analysis indicated a positive correlation between the FC of the ipsilesional MFG and anxiety scores (r = 0.404, (95%CI: 0.106, 0.631), P = 0.012, P-FDR = 0.030) when using the contralesional CC1 as the ROI. When using the ipsilesional PoCG as the ROI, the FC of the contralesional PreCG showed a positive correlation with anxiety scores (r = 0.436, (95%CI: 0.142, 0.658), P = 0.006, P-FDR = 0.030) (Fig. 5A-B and Table 5). The FC-SAS associations also remained significant after additionally controlling for SDS and PSQI (Table S3). No significant correlations were observed with the other scales. No significant correlations between the different brain regions of FC and all scales in NC group (Table S1). Correlation between FC measures and SAS in MCA-S patients. Partial correlation between FC of Frontal Mid IL and SAS(A), between FC of Precentral CON and SAS(B). CON, Contralateral; FDR, False Discovery Rate; IL, ipsilateral; MCA-S, Middle cerebral artery stenosis; SAS, Self rating anxiety scale Table 5The correlation coefficients and the corresponding P values between FC and different scalesROICerebellum Crus1 CONPostcentral ILBrain regionsFrontal Mid ILPrecuneus CONPostcentral ILPrecentral CONSupp Motor Area ILMMSE r 0.0910.063-0.089-0.115-0.069 P 0.6140.7280.6220.5240.703P-FDR0.7280.7280.7280.7280.728MoCA r 0.1110.1590.0260.1680.019 P 0.5380.3780.8870.3510.914P-FDR0.8970.8970.9140.8970.914SAS r 0.404 -0.0970.325 0.436 0.135 P 0.012 0.5930.065 0.006 0.453P-FDR 0.030 0.5930.108 0.030 0.566SDS r 0.287-0.0290.170.3240.122 P 0.1050.8740.3430.0660.498P-FDR0.2630.8740.5720.2630.623PSQI r 0.1470.101-0.0270.137-0.035 P 0.4130.5770.880.4470.846P-FDR0.8800.8800.8800.8800.880CON, Contralesional; FC, Functional connectivity; IL, Ipsilesional; MMSE, Mini-mental state examination; MoCA, Montreal cognitive assessment; PSQI, Pittsburgh sleep quality index; ROI, Regions of interest; SAS, Self rating anxiety scale; SDS, Self rating depression scale The correlation coefficients and the corresponding P values between FC and different scales CON, Contralesional; FC, Functional connectivity; IL, Ipsilesional; MMSE, Mini-mental state examination; MoCA, Montreal cognitive assessment; PSQI, Pittsburgh sleep quality index; ROI, Regions of interest; SAS, Self rating anxiety scale; SDS, Self rating depression scale Discriminative performance of CVR for MCA-S patients ROC analysis was conducted on CVR in brain regions with differences. The area under the curve (AUC) for the contralesional CC1 was 0.809 (95% CI: 0.713–0.884), with a sensitivity of 85.37% and a specificity of 70.00%. The AUC for the ipsilesional PoCG was 0.787 (95% CI: 0.688–0.866), with a sensitivity of 87.80% and a specificity of 62.00%. When combining both, the AUC increased to 0.866 (95% CI: 0.778–0.928), with a sensitivity of 73.17% and a specificity of 92.00% (Fig. 6). Receiver operating characteristics (ROC) curves of the CVR value for MCA-S patients. CON, Contralateral; CVR, Cerebrovascular reactivity; IL, ipsilateral; MCA-S, Middle cerebral artery stenosis Cerebellar anatomical validation SUIT-based anatomical validation showed that the peak coordinate of the significant cerebellar cluster was located within the SUIT-defined contralesional Crus 1 region. In addition, approximately 60% of suprathreshold voxels within this cluster overlapped with the SUIT-defined Crus 1 region. These findings support the anatomical localization of the cerebellar cluster as contralesional Crus 1 using a cerebellum-specific atlas (Figure S3). TFCE validation The TFCE-corrected analysis showed that compared with the NC group, MCA-S patients showed increased CVR in contralesional CC1 (peak MNI coordinates: 27, -84, -33) and decreased CVR in the ipsilesional PoCG (peak MNI coordinates: -42, -27, 66) (P-TFCE < 0.05) (Figure S4, Table S4). Discussion CVR refers to the ability of cerebral blood vessels to dilate or constrict in response to stimuli or stress.We assessed intrinsic CVR using a non-invasive MRI technique to, which provided greater convenience, lower risk, and fewer patient cooperation requirements than traditional methods like hypercapnia or intravenous acetazolamide injection, lower risk, and fewer patient cooperation requirements than traditional methods like hypercapnia or acetazolamide challenge. Notably, the unique contribution of CVR in the study of MCA-S lies in its ability to capture subtle functional changes in cerebral blood flow that are not detectable through structural imaging alone, offering a more detailed understanding of the vascular dysfunction and its role in ischemic events. This study suggested decreased ipsilesional PoCG CVR and increased contralesional CC1 CVR in MCA-S patients. The TFCE-corrected results were convergent with the original GRF-based findings, supporting the robustness of the main anatomical pattern of CVR differences. In this exploratory FC analysis, changes in FC appeared potentially linked to CVR alterations. Partial correlation analysis revealed a positive correlation between anxiety scores and FC of ipsilesional MFG and contralesional PreCG. Additionally, the potential to assess MCA-S using significant CVR may have better discriminative performance. CVR serves as a key indicator of cerebrovascular health. Various methods exist for its assessment, with prior studies typically measuring blood flow in major cerebral arteries or brain tissue perfusion before and after vasodilation stimulation. Yang et al. [11] evaluated CVR in the cerebral hemispheres of 31 patients with unilateral MCA-S, finding a significantly reduced CVR% in the MCA territory that correlated strongly with stenosis severity, while CVR% in the anterior and posterior cerebral artery territories remained unchanged. Hartkamp et al. [31] reported a similarly significant CVR reduction in the affected MCA territory of patients with unilateral internal carotid artery stenosis or occlusion compared to healthy controls. In the present study, CVR was also decreased in the ipsilesional PoCG, a region supplied by the MCA, aligning with the findings of Yang and Hartkamp. Earlier work largely focused on manually delineated regions of interest, using cerebral blood flow differences to represent CVR across the entire MCA territory, an approach potentially susceptible to subjective variation. In the present study, we analyzed CVR at the voxel level, which allowed for sensitive detection of CVR alterations in various brain regions. Additionally, we found elevated CVR in the contralesional CC1 of MCA-S patients compared to NC. This finding contradicted previous studies by Sebők et al. [32, 33] who documented decreased CVR in the affected cerebellar hemisphere among patients with symptomatic unilateral cerebrovascular stenosis or occlusion or unilateral anterior circulation stroke. Bieberstein et al. [34] from the same team reported analogous results in patients with ischemic stroke caused by ICA stenosis, attributing these changes to cerebellar diaschisis as the underlying mechanism. Of note, those prior studies employed terminal CVR as the indicator, which assesses CVR at a single time point, this approach may introduce temporal bias and lacks the capacity for dynamic CVR characterization. Dogra S et al. [12] noted that biases arising from terminal CVR can be unpredictable and may affect both normal and abnormal tissues. They developed a new computational framework using principal component analysis combined with random matrix theory to denoise BOLD MRI data, significantly improving the signal-to-noise ratio. This method allows for a more tailed dynamic analysis of CVR signal time series, capturing CVRmax at each BOLD repetition time, which is more accurate than traditional terminal CVR. Later, their team [35] applied CVRmax to analyze unilateral stenotic or occlusive cerebrovascular diseases, and some cases showed increased CVR in the contralesional cerebellar hemisphere, consistent with our research. This finding may involve cerebrocerebellar network-related or adaptive hemodynamic mechanisms; however, this interpretation remains speculative. The initial results of partial correlation analysis showed a positive correlation between CVR in the contralesional CC1 and depression, a negative correlation between ipsilesional PoCG CVR and CRP. However, these uncorrected associations did not survive FDR correction and should therefore be interpreted cautiously. This may be attributed to the limited sample size (which reduces statistical power to detect small effects after strict correction). Therefore, these findings may be interpreted as exploratory findings requiring validation in future studies with larger sample sizes. The BOLD MRI method employed here measured intrinsic CVR at the tissue level. This technique is non-invasive, requires minimal patient cooperation, and is broadly applicable compared to alternative CVR assessments. The ROC analysis indicated that the discriminative performance of CVR in the two significant brain regions ranges from 0.787 to 0.809, while their combination showed even higher (AUC = 0.866). Consequently, CVR may have potential as a complementary physiological biomarker for characterizing hemodynamic alteration. Functional connectivity (FC) describes the temporal correlation between spatially separated brain regions. Because the seed regions in this study were defined from CVR-differentiated clusters, the FC analysis was exploratory and was intended to characterize the network context of altered CVR rather than to provide statistically independent validation of the CVR findings. When contralesional CC1 was used as the ROI, MCA-S patients showed increased FC with the ipsilesional MFG and contralesional PCUN. CC1 is located in the posterior cerebellar hemisphere and is increasingly recognized as a non-motor cerebellar region involved in higher-order cerebellar-cortical networks, including frontoparietal and default-mode-related circuits. Previous FC studies showed that the cerebellar involvement in multiple high-level brain networks [36–38], where crus I and lobule IX participate primarily in the default mode network (DMN) and are intrinsically linked to other DMN regions [39–41]. These patterns may reflect the increased FC in the MFG and PCUN, which are part of the DMN [42], and may suggest a compensatory increase in DMN connectivity. When the ipsilesional PoCG was used as the ROI, MCA-S patients showed decreased FC in the ipsilesional PoCG, contralesional PreCG, and ipsilesional SMA, all of which are components of the sensorimotor network(SMN). A study [43] in asymptomatic MCA-S has similarly reported widespread FC reductions, particularly in interhemispheric connections, aligning with our observations. Partial correlation analysis further showed positive associations between SAS scores and FC involving ipsilesional MFG and contralesional PreCG. Animal experiments by Wenzel et al. [44] support this, demonstrating that CVR impairment can induce anxiety, potentially via related endothelial damage. These associations suggest a possible link between altered cerebrovascular/network physiology and anxiety symptoms, but they should be interpreted cautiously because of the exploratory nature of the FC analysis. Epidemiological studies [45, 46] identified hypertension, diabetes, hyperlipidemia, metabolic syndrome, smoking, and alcohol consumption as risk factors for stroke. In our study, the MCA-S group had higher proportions of hypertension, diabetes, hyperlipidemia, smoking, and alcohol consumption. Although we did not formally assess metabolic syndrome, the higher BMI observed in the MCA-S group relative to the NC group indirectly reflects a higher risk in this patient population, consistent with prior research. In an investigation of factors affecting cerebral perfusion, Liu et al. [47] reported hypertension as an independent factor for hypoperfusion in patients with middle cerebral artery plaque stenosis. This aligned with our finding of a statistically significant difference in hypertension prevalence between the groups, which was higher in the MCA-S group. Hypertension is known cause of microvascular structural abnormalities, which closely related to the development of arteriosclerosis. As most clinical cases of middle cerebral artery stenosis arise from atherosclerosis, this provides a pathological basis for the higher prevalence of hypertension in such patients. The other clinical factors included in our study showed no statistical differences, a finding potentially attributable to our relatively small sample size. Our results also revealed a statistically significant difference in MMSE scores between the two groups. Although scores in both groups exceeded 26, patients with MCA-S had lower MMSE scores, suggesting a relationship between MCA-S and cognition. Huang et al. [7] reported no significant difference in MMSE scores between patients with asymptomatic MCA-S and NC groups. In their study, overall cognition and other multi-domain neurobehavioral assessments did not show a significant decline in the patient group. However, the patient group had lower scores on the Auditory Verbal Learning Test-Huashan Version, indicating a decline in memory. The absence of differences in MMSE in their study may stem from the inclusion of subjects without WMH. Although we found no statistically significant difference in WMH between our groups, prior studies [48, 49] shown that WMH may be associated with cognitive dysfunction. The study has several limitations. Firstly, we employed rs-fMRI to measure the intrinsic CVR of MCA-S patients. A key limitation of this study is that the resting-state CVR-related metric was not directly validated against hypercapnic CVR imaging in the same MCA-S cohort. Therefore, the present findings should be interpreted as evidence of altered non-hypercapnic CVR-related physiology rather than definitive proof of conventional stimulus-evoked CVR impairment. Although this approach has been successfully applied in atherosclerotic arterial stenosis, moyamoya disease, Alzheimer’s disease, WMHs, and normal controls, accurately reflecting the spatial variation of CVR among subjects. Secondly, this study was cross-sectional and limited by a relatively small sample size; no subgroup analyses(such as by laterality or severity) were performed. Potential differences among these subgroups therefore remain unexplored. Future work will address these through expanded recruitment and longitudinal follow-up. Thirdly, because the seed regions for FC analysis were defined from rs-CVR group differences within the same resting-state dataset, and because the frequency ranges used for rs-CVR and FC partially overlap, these FC results should be considered exploratory and hypothesis-generating rather than statistically independent validation of the rs-CVR findings. Conclusion In summary, non-hypercapnic rs-MRI effectively evaluates CVR changes in MCA-S patients and may serve as a complementary physiological marker for characterizing hemodynamic alteration. This may provide a new basis for individualized management of patients with clinical cerebrovascular disease. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Acknowledgements Thanks to all the researchers who participated in this study. Author contributions L.Z. and Z.D: study concept and design. X.T., R.W. and Y.Z: data collection and curation. L.W and R.W: data analysis and interpretation. L.Z: manuscript writing. H.Z., L.W. and Z.D: manuscript revision. Z.D: supervising the study. All authors approved the final manuscript version. Funding This work was funded by Medical and Health Science Program of Zhejiang Province (No. 2021KY861, 2025HY0621), Zhejiang Provincial Traditional Chinese Medicine Project (No.2021ZQ072), and the Construction Fund of Key Medical Disciplines of Hangzhou - Rare Disease (Motor Neuron Disease) − 2025HZZD09. Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of the Hangzhou First People’s Hospital (IIT-20230822-0181-01) and conducted by the Declaration of Helsinki, and all participants signed informed consent forms. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Liqing Zhang and Luoyu Wang have contributed equally to this work and share first authorship. Contributor Information Luoyu Wang, Email: wangluoyu0827@hotmail.com. Zhongxiang Ding, Email: hangzhoudzx73@126.com. References Associated Data Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.