Cerebral small vessel disease impairs dynamic cerebral autoregulation and mediates outcomes after stroke thrombectomy Authors: Yue Qiao, Ran Liu, Lupei Cai, Min Zhao, Boyi Yuan, Siying Fan, Shengjun Wang, Jiameng Li, Longfei Wu, Yingqi Xing, Qingfeng Ma, Wenbo Zhao Published: 2026 ABSTRACT Background:Cerebral small vessel disease (CSVD) is associated with poor outcomes after endovascular thrombectomy (EVT), but the underlying mechanism remains unclear. Impaired dynamic cerebral autoregulation (dCA), a fundamental function of cerebral microvasculature, has been proposed as a potential mediator but has not been systematically investigated.Objectives:To determine whether CSVD burden influences functional outcomes through alterations in dCA.Design:Observational-cohort-study.Methods:We included 107 patients with anterior circulation large-vessel occlusion and successful EVT (expanded Thrombolysis in Cerebral Infarction ⩾2b). CSVD burden was assessed using standardised neuroimaging markers and dichotomised as none-to-mild (score 0–1) versus moderate-to-severe (score 2–3). dCA was assessed at two time points: early phase (T1, around 24 h post-EVT) and early subacute phase (T2, around 4–5 days post-EVT). Mediation analysis was performed to assess whether dCA changes mediated the association between CSVD burden and functional independence (modified Rankin Scale (mRS) score 0–2) at 90 days.Results:Patients with moderate-to-severe CSVD (n = 38) demonstrated distinct dCA recovery patterns compared with those with none-to-mild CSVD (n = 69). While patients with none-to-mild CSVD exhibited improved very low frequency (VLF, 0.02–0.07 Hz) phase differences from T1 to T2 (+7.67° (−9.73, 35.53)), those with moderate-to-severe CSVD exhibited deterioration (−7.17° (−25.03, −0.74), p = 0.004), which remained significant after multivariable adjustment (β = −20.06, 95% confidence interval (CI): −37.78 to −0.90, p = 0.040). Functional independence was achieved in 73.9% of patients with none-to-mild CSVD versus 52.6% in those with moderate-to-severe CSVD (adjusted OR 0.25, 95% CI 0.09–0.74, p = 0.012). Mediation analysis revealed that VLF phase changes in the affected side mediated 24.48% of the association between CSVD burden and functional outcome (average causal mediation effect: −7%, p < 0.001; total effect: −29%, p = 0.020).Conclusion:Higher CSVD burden was associated with worse functional outcomes after EVT, partially explained by impaired dCA. Patients with moderate-to-severe CSVD exhibited deteriorating dCA, whereas those with none-to-mild CSVD showed progressive improvement. Larger powerful studies are warranted to validate these findings and elucidate underlying mechanisms.Trial registration:ClinicalTrials.gov, NCT06361017. Introduction Cerebral small vessel disease (CSVD) describes a range of pathological processes that damage small end arteries, arterioles, venules and brain capillaries. It is found in over 50% of people aged over 65 years, and is the primary cause of lacunar stroke and vascular dementia.1 –3 Imaging features of CSVD, including white matter hyperintensities (WMH), lacunes, perivascular spaces, cerebral microbleeds and brain atrophy, 2 are also frequently observed in patients with acute large-vessel occlusion. 4 Notably, nearly half of patients undergoing endovascular thrombectomy (EVT) show evidence of CSVD burden, 5 which has consistently been associated with poorer functional recovery.5,6 This association has been attributed to impaired cerebrovascular reserve and increased susceptibility to reperfusion injury.7,8 However, the underlying mechanisms by which CSVD burden contributes to poor post-thrombectomy outcomes remain unclear. Dynamic cerebral autoregulation (dCA) stabilises cerebral perfusion by buffering fluctuations in systemic blood pressure (BP), 9 and impaired dCA has been associated with poorer functional outcomes in acute ischaemic stroke.9,10 It has also been proposed as an independent predictor of outcome following EVT. 11 Physiologically, dCA relies on myogenic responses of small arterial smooth muscle, endothelial regulation, neural modulation and the integrity of the neurovascular unit, all of which are disrupted in CSVD.11 –13 Moreover, EVT itself may exacerbate vascular dysfunction through endothelial injury and inflammatory responses, while patients with a higher CSVD burden are particularly vulnerable to microcirculatory disturbances and reperfusion injury.14,15 These findings suggest that impaired dCA may represent a key mechanistic link between CSVD and poor functional recovery after EVT. However, whether post-thrombectomy alterations in dCA mediate the association between CSVD burden and clinical outcomes remains unknown. The present study aimed to investigate the impact of pre-existing CSVD burden on dCA following successful stroke thrombectomy. In addition, we determine whether dCA alterations mediate the association between CSVD burden and functional outcomes. Materials and methods Study design and population This study was based on a prospective registry study at the Xuanwu Stroke Centre (ClinicalTrials.gov: NCT06361017), a high-volume stroke centre in China. Information on the registry study, the processes of EVT for AIS, and other information have been described previously. 16 In the present study, all patients who underwent EVT between March 2022 and December 2024 were screened, as systematic dCA assessments in EVT-treated patients were initiated at our centre from March 2022 onwards. The Inclusion criteria were: (1) age ⩾18 years; (2) acute ischaemic stroke due to anterior circulation large-vessel occlusion; (3) achieving successful recanalisation after thrombectomy, defined as expanded Thrombolysis in Cerebral Infarction (eTICI) score ⩾2b50; (4) availability of cranial MRI examination after EVT; and (5) availability of dCA examination. Patients were excluded if they had: (1) pre-existing neurological disability (modified Rankin Scale (mRS) > 2); (2) severe stenosis (>70%) or occlusion of the contralateral internal carotid artery or middle cerebral artery (MCA); (3) other conditions potentially affecting cerebral haemodynamics, such as heart failure, hyperthyroidism or myocardial infarction 17 ; and (4) transfer function analysis (TFA) metrics did not meet quality criteria. Data collection We collected demographic characteristics from the medical records system, including age, sex, body mass index, medical history (hypertension, diabetes mellitus, hyperlipidaemia, coronary heart disease, atrial fibrillation, prior stroke or transient ischaemic attack, use of anticoagulant/antiplatelet drugs, smoking and alcohol consumption status). Clinical data on admission included the National Institutes of Health Stroke Scale (NIHSS) score, Glasgow Coma Scale score, Alberta Stroke Program Early CT Score (ASPECTS), and systolic and diastolic BP. Procedural characteristics included time from last known well to groin puncture, use of intravenous thrombolysis, stroke aetiology according to TOAST classification, EVT details and periprocedural management. Neuroimaging and CSVD burden assessment All included patients underwent 3.0 T brain MRI within 3 days after thrombectomy, including at minimum T1-weighted, T2-weighted, DWI, ADC and FLAIR imaging. MRI was performed on a Discovery MR750 scanner (GE Medical Systems, Milwaukee, WI, USA) using a 32-channel head array coil. CSVD markers were assessed according to Standards for Reporting Vascular changes on Neuroimaging (STRIVE)-2 criteria 1 by two experienced trained readers blinded to clinical details, with a third neuroradiologist consulted in case of disagreement. WMH were evaluated on fluid-attenuated inversion recovery (FLAIR) images and rated using the Fazekas scale. 18 Periventricular WMH and deep WMH were scored separately (0–3 points each). Lacunes were defined as round or ovoid, fluid-filled cavities (with a signal like cerebrospinal fluid on all sequences) up to 15 mm in diameter. The perivascular spaces were assessed on T2-weighted images in the basal ganglia with small punctate (<3 mm) or linear hyperintensities, according to a semi-quantitative rating scale developed by the Edinburgh group. 19 A modified total CSVD burden score was evaluated based on the presence of four standard neuroimaging markers, as previously described. 20 One point was allocated for each of the following: (1) any lacunes; (2) WMH burden (perivascular WMH Fazekas 3 or deep-WMH Fazekas ⩾2); (3) any cerebral micro-bleeding; and (4) moderate-to-severe (>10) perivascular spaces in the basal ganglia. Because susceptibility-weighted imaging (SWI) was not routinely available and microbleeds could not be reliably ascertained, the microbleed item was omitted, yielding a 3-item modified score ranging from 0 to 3. We stratified participants into two groups with CSVD score 0–1 and CSVD score ⩾2, consistent with prior thrombectomy and ischaemic-stroke cohorts using composite CSVD scores.14,21 In cases where the acute infarct potentially overlapped with areas of chronic small vessel changes, CSVD burden assessment was performed primarily on the contralateral hemisphere and at least 10 mm away from the margin of the acute ischaemic lesion as defined on DWI images. If lesions appeared on both FLAIR and diffusion-weighted imaging (DWI) images, we excluded areas showing high signals on DWI images, then measured the remaining chronic changes on FLAIR images and compared with CSVD markers on the contralateral hemisphere. dCA evaluation dCA was assessed according to the International Cerebral Autoregulation Research Network guidelines 22 at two predefined time points to capture both the immediate and short-term hemodynamic responses following recanalisation. The early phase assessment (T1) was performed approximately 24 h post-EVT, and the follow-up assessment (T2) during the early subacute phase, around 4–5 days post-EVT. Detailed methods for evaluating dCA have been reported previously.23,24 Patients were examined in the supine position in a quiet, temperature-controlled room (20°C–24°C) with spontaneous breathing. Bilateral MCA flow velocities were recorded simultaneously using the EMS-9D PRO TCD system (Shenzhen Delica Medical, Shenzhen, China) with two 2 MHz probes fixed over the temporal acoustic windows via a head frame, targeting the M1 segment of MCA at a depth of 50–60 mm. Patients with insufficient temporal acoustic windows that prevented stable bilateral MCA flow velocity recording were excluded from the dCA analysis, as reliable measurements could not be obtained in these cases. Continuous non-invasive beat-to-beat BP was measured by servo-controlled finger plethysmography (Finometer; Finapres Medical Systems, Enschede, The Netherlands), calibrated against brachial artery pressure (Omron HBP-1300) before each session. End-tidal CO2 was monitored continuously via nasal cannula. Each recording lasted 10 min; a stable 5-min artefact-free segment was selected for TFA. TFA was performed using the CA-200 TFA module of the EMS-9D PRO system per CARNet guidelines, applying a Hanning window of 100s with 50% superposition. Phase shift, absolute gain (cm/s/mmHg), normalised gain (%/mmHg) and coherence were derived for the very low frequency (VLF; 0.02–0.07 Hz) and low frequency (0.07–0.20 Hz) bands. TFA quantifies the dynamic pressure-flow relationship by treating arterial BP as input and cerebral blood flow velocity (CBFV) as output; phase shift represents the temporal displacement between these signals, with a larger positive value indicating active autoregulatory buffering and a lower value reflecting a more pressure-passive, impaired circulation. According to the recent study, cerebral autoregulation operates most efficiently at frequencies at or below 0.03 Hz 22 ; accordingly, VLF phase was designated as the primary dCA parameter for mediation analysis. Gain gauges the degree to which dCA dampens BP oscillation intensity, with lower values reflecting more efficient autoregulation. Coherence reflects the linearity of the BP-CBFV relationship, approaching one in highly linear systems; only recordings with coherence >0.5 were included in analyses. Outcome assessment The primary outcome of this study was to evaluate the mediating role of dCA changes in the relationship between CSVD burden and functional recovery post-EVT. We employed mediation analysis with functional independence (mRS score 0–2) at 90 days. Secondary outcomes, prespecified to assess the association between CSVD burden and prognosis, included excellent functional outcome (mRS 0–1), ordinal analysis of the full mRS score distribution at 90 days, early neurological improvement, defined as either a ⩾4-point decrease in NIHSS score or achievement of a score of 0, assessed at both 3- and 7-day post-procedure, and adverse imaging events including symptomatic intracranial haemorrhage (sICH) and hematoma expansion. All outcome assessments were evaluated by raters blinded to the dCA results. Serial CT imaging was performed to evaluate hematoma expansion, defined as an absolute volume increase of >12.5 mL or a relative increase of >33% within 3 days of initial presentation or during the follow-up period. 25 Hematoma volumes were calculated using the ellipsoid formula (A × B × C/2, where A and B represent the largest perpendicular diameters on the slice with the largest haemorrhage area, and C represents the number of slices multiplied by slice thickness). 26 sICH was assessed based on clinical deterioration (⩾4-point increase in NIHSS scores) accompanied by radiological evidence of new or expanded haemorrhage. Collateral circulation was assessed using computed tomography angiography (CTA) and graded as good or poor based on the Tan scale. 27 All imaging analyses were performed by two experienced neurologists blinded to the clinical and dCA data. Statistical analysis Continuous variables were expressed as mean ± SD or median (IQR), and categorical variables as frequencies (percentages). Between-group comparisons used t-test or Mann–Whitney U test for continuous variables and chi-square or Fisher’s exact test for categorical variables. Associations between CSVD burden and dCA parameters were analysed using linear regression models with progressive adjustment: Model 1 adjusted for age and sex; Model 2 additionally adjusted hypertension, diabetes, previous stroke, occlusion site and intravenous thrombolysis. Logistic regression analysis was used to evaluate the association between CSVD burden and functional outcomes. Receiver operating characteristic (ROC) curve analysis was performed to assess the predictive value of VLF Δphase for functional independence. Subgroup analyses were conducted using forest plots to evaluate the consistency of associations across different baseline characteristics. Causal mediation analysis using the counterfactual framework examined whether dCA mediated the CSVD-outcome relationship. Nonparametric bootstrap (1000 simulations) estimated the average causal mediation effect, average direct effect and proportion mediated. Effect estimates were expressed on the probability scale as percentage changes in the likelihood of functional independence. All analyses were performed using R software (version 4.3.0; R Foundation for Statistical Computing, Vienna, Austria), with statistical significance set at two-tailed p < 0.05. Results Baseline characteristics A total of 107 patients were included in this study, and the flow chart of this study is showed in Supplemental Figure 1 in the materials. The mean age of this study population was 58 (±15), 18 (16.8%) of them were female. Patients were stratified by CSVD burden into none-to-mild (score 0–1, n = 69) and moderate-to-severe (score 2–3, n = 38) groups. Baseline characteristics of these two groups of patients are summarised in Table 1. The moderate-to-severe CSVD group was significantly older (66 ± 11 vs 55 ± 15 years) and had higher prevalence of diabetes mellitus (36.8% vs 11.6%), coronary heart disease (28.9% vs 10.1%), previous ischaemic stroke (42.1% vs 13.0%) and prior antiplatelet use (28.9% vs 5.8%). Additionally, patients with moderate-to-severe CSVD demonstrated poorer collateral circulation (31.6% vs 53.6% good collaterals) and different final eTICI score distribution compared to those with none-to-mild CSVD. Table 1.Baseline characteristics of patients stratified by CSVD burden.VariablesNone-to-mild CSVD (n = 69)Moderate-to-severe CSVD (n = 38)p ValueAge (years), mean (SD)55 ± 1566 ± 11<0.001Sex (male), n (%)56 (81.2)33 (86.8)0.63BMI (kg/m2), median (IQR)24.8 (23.8–27.0)24.3 (23.4–28.4)0.742Medical history Hypertension, n (%)38 (55.1)28 (73.7)0.092 Diabetes mellitus, n (%)8 (11.6)14 (36.8)0.004 Hyperlipidaemia, n (%)11 (15.9)11 (28.9)0.179 Coronary heart disease, n (%)7 (10.1)11 (28.9)0.027 Atrial fibrillation, n (%)14 (20.3)11 (28.9)0.439 Previous ischaemic stroke, n (%)9 (13.0)16 (42.1)0.002 Current smoking, n (%)36 (52.2)21 (55.3)0.917 Current drinking, n (%)27 (39.1)18 (47.4)0.534 Prior antiplatelet use, n (%)4 (5.8)11 (28.9)0.003 Prior anticoagulant use, n (%)5 (7.2)5 (13.2)0.425Baseline BP Systolic BP (mmHg)148.0 ± 23.7153.7 ± 22.50.230 Diastolic BP (mmHg), median (IQR)83.0 (73.0–96.0)82.0 (75.3–97.5)0.914NIHSS score at baseline, median (IQR)12.0 (10.0–15.0)14.5 (9.3–16.0)0.144GCS at baseline, median (IQR)13.0 (10.0–14.0)11.0 (10.0–14.0)0.059ASPECTS, median (IQR)8.0 (7.0–9.0)8.0 (7.0–9.0)0.708Time from last known well to groin puncture (min), median (IQR)525.0 (345.0–728.0)585.5 (386.3–815.0)0.347Time from first TCD examination to thrombectomy completion (h), median (IQR)18.0 (11.0–32.0)21.0 (13.5–26.8)0.984Time from second TCD examination to thrombectomy completion (h), median (IQR)102.0 (67.0–148.0)96.0 (63.5–128.8)0.340IVT, n (%)25 (36.2)8 (21.1)0.159Occlusion site, n (%)0.735 ICA28 (40.6)13 (34.2) MCA-M136 (52.2)21 (55.3) MCA-M25 (7.2)4 (10.5)Good collaterals, n (%)37 (53.6)12 (31.6)0.047Stroke aetiology, n (%)0.172 Large artery atherosclerosis39 (56.5)23 (60.5) Cardioembolism24 (34.8)15 (39.5) Others6 (8.7)0 (0.0)Stenting, n (%)20 (29.0)14 (36.8)0.536Final eTICI score, n (%)0.007 2b23 (33.3)8 (21.1) 2c2 (2.9)9 (23.7) 344 (63.8)21 (55.3)ASPECTS, Alberta Stroke Program Early CT Score; BMI, body mass index; BP, blood pressure; CSVD, cerebral small vessel disease; eTICI, expanded thrombolysis in cerebral infarction; GCS, Glasgow Coma Scale; ICA, internal carotid artery; IVT, intravenous thrombolysis; MCA, middle cerebral artery; NIHSS, National Institutes of Health Stroke Scale; TCD, transcranial Doppler. Baseline characteristics of patients stratified by CSVD burden. ASPECTS, Alberta Stroke Program Early CT Score; BMI, body mass index; BP, blood pressure; CSVD, cerebral small vessel disease; eTICI, expanded thrombolysis in cerebral infarction; GCS, Glasgow Coma Scale; ICA, internal carotid artery; IVT, intravenous thrombolysis; MCA, middle cerebral artery; NIHSS, National Institutes of Health Stroke Scale; TCD, transcranial Doppler. Association between CSVD burden and dCA parameters All patients completed dCA early phase assessment (median 19 h (IQR: 11–30) post-EVT), and 70 (65.4%) underwent a follow-up assessment at a median 101 h (IQR: 64–146); comparisons between patients with and without T2 dCA assessment are presented in Supplemental Table 1. At the early phase assessment, patients with moderate-to-severe CSVD showed significantly higher VLF phase differences in both affected (45.92° vs 33.56°, p = 0.026) and unaffected hemispheres (59.15° vs 48.23°, p = 0.011) compared to the none-to-mild group. The association remained significant for the unaffected hemisphere after multivariable adjustment (β = 15.63, 95% confidence interval (CI): 3.92–27.35, p = 0.009). Notably, the change in VLF phase differences from initial to follow-up assessment revealed opposite patterns between groups (Figure 1). In the affected hemisphere, the none-to-mild CSVD group showed improvement in phase differences (+7.67° (−9.73, 35.53)), while the moderate-to-severe group demonstrated deterioration (−7.17° (−25.03, −0.74), p = 0.004). This difference persisted after adjustment in Model 2 (β = −20.06, 95% CI: −37.78 to −0.90, p = 0.040). Similar patterns were observed in the unaffected hemisphere (Table 2). The association between the change in VLF phase differences and individual CSVD markers showed consistent patterns across different small vessel disease features (Supplemental Figure 2). VLF phase differences between CSVD burden groups. VLF phase differences at initial assessment (T1) and follow-up assessment (T2) in (a) affected and (b) unaffected hemispheres between none-to-mild and moderate-to-severe CSVD groups. CSVD, cerebral small vessel disease; VLF, very low frequency. Table 2.Comparison of dCA parameters between non-mild and moderate-severe CSVD burden score groups.Comparison of dCA parameters between non-mild and moderate-severe CSVD burden score groupsUnadjustedModel 1Model 2VariablesNone-to-mild CSVD (n = 69)Moderate-to-severe CSVD (n = 38) p β (95% CI) P β (95% CI) p β (95% CI) p Early phase assessment (T1) Phase in affected side (degree)33.56 (22.58, 48.01)45.92 (30.26, 57.52)0.0269.76 (0.46, 19.10)0.0408.48 (−1.63, 18.60)0.0996.99 (−3.95, 17.94)0.208 Phase in unaffected side (degree)48.23 (36.66, 58.88)59.15 (44.27, 78.29)0.01114.50 (4.22, 24.90)0.00617.80 (6.68, 28.80)0.00215.63 (3.92, 27.35)0.009 Gain in affected side (%/mmHg)1.10 (0.80, 1.50)1.21 (0.72, 1.56)0.711−2.03 (−7.74, 3.67)0.481−3.73 (−9.88, 2.43)0.232−3.44 (−10.09, 3.21)0.307 Gain in unaffected side (%/mmHg)1.25 (1.02, 1.67)1.23 (0.90, 1.64)0.619−0.05 (−0.28, 0.17)0.640−0.02 (−0.26, 0.23)0.884−0.09 (−0.34, 0.17)0.516Early subacute phase assessment (T2) a Phase in affected side (degree)41.99 (26.90, 57.51)39.50 (24.22, 48.30)0.402−7.28 (−19.30, 4.78)0.230−6.16 (−19.60, 7.26)0.363−7.35 (−22.10, 7.41)0.323 Phase in unaffected side (degree)51.03 (41.86, 73.53)44.83 (38.45, 56.89)0.116−6.92 (−19.00, 5.16)0.257−7.32 (−20.70, 6.05)0.278−12.17 (−26.62, 2.28)0.097 Gain in affected side (%/mmHg)1.10 (0.95, 1.49)1.09 (0.80, 1.59)0.8270.03 (−0.20, 0.28)0.7560.08 (−0.18, 0.35)0.5230.11 (−0.18, 0.39)0.458 Gain in unaffected side (%/mmHg)1.19 (0.84, 1.58)1.33 (1.04, 1.73)0.3980.19 (−0.13, 0.50)0.2420.27 (−0.08, 0.62)0.1300.32 (−0.03, 0.68)0.075Changes from first to second time point a ΔVLF phase in affected side7.67 (−9.73, 35.53)−7.17 (−25.03, −0.74)0.004−22.00 (−37.80, −6.19)0.007−19.80 (−37.40, −2.28)0.027−20.06 (−37.78, −0.90)0.040 ΔVLF phase in unaffected side5.18 (−17.80, 26.72)−18.13 (−33.46, −3.51)0.003−25.9 (−43.4, −8.4)0.004−30.5 (−49.7, −11.4)0.002−29.19 (−50.52, −7.86)0.008 ΔLF phase in affected side7.81 (−15.16, 30.58)2.70 (−11.16, 16.05)0.343−11.10 (−36.40, 14.20)0.384−13.80 (−42.00, 14.30)0.331−16.06 (−46.78, 14.66)0.300 ΔLF phase in unaffected side9.68 (−4.14, 23.65)−2.28 (−9.02, 16.60)0.189−7.54 (−25.9, 10.8)0.414−13.4 (−33.9, 7.05)0.195−11.10 (−33.35, 11.15)0.322 ΔVLF gain in affected side−0.01 (−0.30, 0.32)0.01 (−0.54, 0.18)0.375−0.15 (−0.40, 0.09)0.213−0.04 (−0.31, 0.22)0.756−0.08 (−0.37, 0.22)0.601 ΔVLF gain in unaffected side−0.17 (−0.38, 0.16)0.07 (−0.12, 0.29)0.0620.16 (−0.05, 0.37)0.1290.18 (−0.05, 0.41)0.1170.16 (−0.10, 0.41)0.223 ΔLF gain in affected side0.01 (−0.26, 0.37)0.01 (−0.24, 0.15)0.406−0.11 (−0.43, 0.21)0.500−0.01 (−0.36, 0.33)0.943−0.07 (−0.45, 0.31)0.714 ΔLF gain in unaffected side0.03 (−0.32, 0.30)0.16 (−0.13, 0.37)0.3430.05 (−0.26, 0.36)0.7480.10 (−0.24, 0.44)0.5640.08 (−0.29, 0.44)0.668Model 1: adjusted for age and sex; Model 2: adjusted for age, sex, hypertension, diabetes, previous stroke, occlusion site and intravenous thrombolysis.aData from 70 patients who completed both first and second time point TCD examinations.Δ, Changes from first to second time point; CI, confidence interval; CSVD, cerebral small vessel disease; dCA, dynamic cerebral autoregulation; LF, low frequency; VLF, very low frequency. Comparison of dCA parameters between non-mild and moderate-severe CSVD burden score groups. Model 1: adjusted for age and sex; Model 2: adjusted for age, sex, hypertension, diabetes, previous stroke, occlusion site and intravenous thrombolysis. Data from 70 patients who completed both first and second time point TCD examinations. Δ, Changes from first to second time point; CI, confidence interval; CSVD, cerebral small vessel disease; dCA, dynamic cerebral autoregulation; LF, low frequency; VLF, very low frequency. Impact of CSVD burden on clinical outcomes Functional independence at 90 days was achieved in 73.9% of patients with none-to-mild CSVD compared with 52.6% of those with moderate-to-severe CSVD (adjusted OR 0.25, 95% CI, 0.09–0.74, p = 0.012). The distribution of 90-day mRS scores differed significantly between groups, with higher scores observed in the moderate-to-severe CSVD group (adjusted OR 0.37, 95% CI, 0.16–0.86, p = 0.022). Excellent functional outcome was less frequent in the moderate-to-severe CSVD group (21.1% vs 47.8%, adjusted OR 0.26, 95% CI, 0.09–0.76, p = 0.015). No significant differences were observed in early neurological improvement or haemorrhagic complications between groups (Table 3). Table 3.Clinical outcomes stratified by CSVD burden.VariablesNone-to-mild CSVD (n = 69)Moderate-to-severe CSVD (n = 38)Unadjusted analysisAdjusted analysis a Effect size (95% CI) p Effect size (95% CI) p Primary outcome 90-Day mRS 0–251 (73.9%)20 (52.6%)0.39 (0.17–0.90)0.0280.25 (0.09–0.74)0.012Secondary outcomes mRS at 90 days, median (IQR)2 (0–3)3 (2–4)0.37 (0.18–0.76)0.0070.37 (0.16–0.86)0.022 90-Day mRS 0–133 (47.8%)8 (21.1%)0.29 (0.12–0.72)0.0080.26 (0.09–0.76)0.015 Early neurological improvement within 3 days b 37 (53.6%)22 (57.9%)1.19 (0.53–2.64)0.6711.30 (0.51–3.32)0.577 Early neurological improvement within 7 days b 50 (72.5%)28 (73.7%)1.06 (0.43–2.60)0.8921.21 (0.42–3.52)0.720 Any ICH within 24 h24 (34.8%)12 (31.6%)0.87 (0.37–2.01)0.7371.01 (0.37–2.76)0.980 Hematoma2 (2.9%)3 (7.9%)2.87 (0.46–18.00)0.2607.02 (0.70–70.04)0.068 sICH0 (0%)2 (5.3%)––––aAdjusted for age, sex, hypertension, diabetes, previous stroke, occlusion site and intravenous thrombolysis.bEarly neurological improvement defined as NIHSS decrease ⩾4 or score of 0. Hematoma defined as absolute increase >12.5 mL or relative increase >33% within 3 days or follow-up.CI, confidence interval; CSVD, cerebral small vessel disease; ICH, intracranial haemorrhage; mRS, modified Rankin Scale; sICH, symptomatic intracranial haemorrhage. Clinical outcomes stratified by CSVD burden. Adjusted for age, sex, hypertension, diabetes, previous stroke, occlusion site and intravenous thrombolysis. Early neurological improvement defined as NIHSS decrease ⩾4 or score of 0. Hematoma defined as absolute increase >12.5 mL or relative increase >33% within 3 days or follow-up. CI, confidence interval; CSVD, cerebral small vessel disease; ICH, intracranial haemorrhage; mRS, modified Rankin Scale; sICH, symptomatic intracranial haemorrhage. Association between dCA changes, CSVD burden and functional outcomes The ROC curve for change in VLF phase differences in the affected hemisphere predicting 90-day functional independence is shown in Supplemental Figure 3 (AUC = 0.72, 95% CI: 0.59–0.85). The optimal cutoff value was −10.5°, with a sensitivity of 81.3% and specificity of 54.5%. In the adjusted mediation analysis, the total effect of CSVD burden on 90-day functional independence (−29%, p = 0.020) is composed of the direct effect of CSVD burden (−22%, p = 0.060) and the mediated indirect effect of change in VLF phase differences (−7%, p < 0.001). The average causal mediation effect demonstrated that approximately 24.48% of the total effect of CSVD on functional independence was mediated through change in VLF phase differences (Figure 2 and Supplemental Table 2). The interaction analysis between CSVD burden and VLF phase changes showed no significant interaction effect on 90-day functional independence (p for interaction = 0.395, Supplemental Figure 4). Based on the subgroup analysis, the interaction analyses revealed no significant effect modification by most clinical variables (all p for interaction > 0.05), except for stroke aetiology (p for interaction = 0.002, Supplemental Figure 5). In a sensitivity analysis treating the modified CSVD score as continuous, the association was attenuated and non-significant (adjusted OR 0.74, p = 0.311; Supplemental Table 3). Mediation analysis of the relationship between CSVD burden and 90-day functional independence through changes in VLF phase differences in affected side. (a) Unadjusted model. (b) Adjusted model. The adjusted model is adjusted for age, sex, hypertension, diabetes, previous stroke, occlusion site and intravenous thrombolysis. βDir, direct effect; βIndir, indirect effect; βTotal, total effect; CSVD, cerebral small vessel disease; VLF, very low frequency. Discussion The present study demonstrated that in patients with anterior circulation stroke who achieved successful recanalisation after thrombectomy, CSVD burden influences dCA temporal evolution, with moderate-to-severe CSVD patients showing deteriorating VLF phase differences from the early phase (around 24 h) to the early subacute phase (around 4–5 days), contrasting with the recovery patterns observed in none-to-mild CSVD patients. Meanwhile, higher CSVD burden was associated with worse functional outcomes at 90 days, and mediation analysis revealed that this association was partially mediated by dCA impairment. It is well recognised that dCA plays a crucial role in cerebral hemodynamic regulation, and patients with CSVD typically exhibit impaired autoregulatory function.28 –30 Previous studies have shown that patients undergoing EVT develop bilateral dCA impairment with gradual recovery over 5–7 days.31,32 Additionally, our previous work demonstrated that changes in dCA are important predictors of both parenchymal haemorrhage risk and 90-day functional outcomes in patients treated with EVT.23,24 In the present study, patients with moderate-to-severe CSVD showed higher absolute VLF phase at T1 but subsequent deterioration, whereas those with none-to-mild CSVD showed progressive improvement. The higher early phase in the moderate-to-severe group may reflect transient haemodynamic changes after EVT, rather than superior autoregulatory function. Therefore, the longitudinal VLF phase trajectory from T1 to T2 better reflects post-EVT autoregulatory recovery and suggests impaired dCA restoration in patients with greater CSVD burden. Importantly, these changes were observed bilaterally, consistent with prior evidence that CSVD represents a global cerebrovascular disorder rather than a focal pathology.33,34 VLF phase change may be particularly relevant to CSVD related autoregulatory impairment because slow BP oscillations are more closely linked to dynamic vasomotor buffering, and phase reflects the temporal separation between BP and CBFV. 22 CSVD-associated endothelial dysfunction, arteriolar smooth muscle degeneration, and microvascular stiffness could plausibly reduce this buffering capacity. In our cohort, longitudinal change in VLF phase, but not LF phase or gain, differed significantly between CSVD burden groups, supporting its use as the main hemodynamic marker linking CSVD burden to impaired cerebral perfusion regulation after thrombectomy. 31 Consistent with previous studies, this study demonstrates that higher CSVD burden scores are associated with worse functional outcomes at 90 days following EVT.14,35,36 While this effect has been well established,5,14,35,36 this study provides one potential explanation: impaired dCA, which is commonly exists in and may result from CSVD, may represent a key mechanism contributing to these adverse outcomes. The finding that VLF phase changes statistically explained approximately one-quarter of the total CSVD effect suggests that autoregulatory dysfunction is an important but not exclusive mechanism. Other pathways, including compromised blood-brain barrier integrity, and increased susceptibility to reperfusion injury, likely contribute to the remaining effect.5,37 –39 The deteriorating dCA patterns observed in moderate-to-severe CSVD patients likely reflect multiple interconnected pathophysiological mechanisms. CSVD-associated microvascular alterations, including endothelial dysfunction, smooth muscle cell degeneration and basement membrane thickening, fundamentally compromise the structural and functional integrity of the cerebral microvasculature.40,41 These changes impair vasomotor reactivity and reduce microvascular compliance, limiting the capacity for dynamic autoregulatory responses even after restoration of proximal blood flow through mechanical thrombectomy, and finally compromising the recovery of neurological deficits. This study has several limitations. First, the relatively small sample size may have reduced the statistical power of our analyses, and its single-centre design and post-hoc investigation may restrict the generalisability of the findings; 37 (34.6%) patients did not complete the T2 dCA assessment, which may have reduced statistical power for longitudinal analyses and introduced attrition bias. Second, cerebral microbleeds were not included in the CSVD burden assessment because of the unavailability of SWI, which may have resulted in an underestimation of CSVD severity in some patients. Third, dCA was only evaluated within approximately 5 days after EVT, and longer-term changes beyond this period were not captured, which could be relevant for understanding sustained autoregulatory adaptation. Finally, although mediation analysis was used to explore whether dCA change statistically explained part of the association between CSVD burden and functional outcome, causal interpretation is limited by the observational design, modest sample size, incomplete follow-up dCA assessments and potential residual confounding. Future studies with larger sample sizes and prespecified mediation frameworks are needed to validate these findings. In conclusion, in patients with anterior circulation stroke who achieved successful thrombectomy recanalisation, higher CSVD burden was associated with worse functional outcomes and divergent trajectories of dCA recovery, with moderate-to-severe CSVD showing impaired autoregulation compared with none-to-mild CSVD. These dCA alterations partially mediated the adverse impact of CSVD on functional outcomes. Validation in larger, adequately powered studies and further exploration of the underlying mechanisms are warranted. Supplemental Material sj-docx-1-tan-10.1177_17562864261470197 – Supplemental material for Cerebral small vessel disease impairs dynamic cerebral autoregulation and mediates outcomes after stroke thrombectomy Supplemental material, sj-docx-1-tan-10.1177_17562864261470197 for Cerebral small vessel disease impairs dynamic cerebral autoregulation and mediates outcomes after stroke thrombectomy by Yue Qiao, Ran Liu, Lupei Cai, Min Zhao, Boyi Yuan, Siying Fan, Shengjun Wang, Jiameng Li, Longfei Wu, Yingqi Xing, Qingfeng Ma and Wenbo Zhao in Therapeutic Advances in Neurological Disorders sj-docx-2-tan-10.1177_17562864261470197 – Supplemental material for Cerebral small vessel disease impairs dynamic cerebral autoregulation and mediates outcomes after stroke thrombectomy Supplemental material, sj-docx-2-tan-10.1177_17562864261470197 for Cerebral small vessel disease impairs dynamic cerebral autoregulation and mediates outcomes after stroke thrombectomy by Yue Qiao, Ran Liu, Lupei Cai, Min Zhao, Boyi Yuan, Siying Fan, Shengjun Wang, Jiameng Li, Longfei Wu, Yingqi Xing, Qingfeng Ma and Wenbo Zhao in Therapeutic Advances in Neurological Disorders