Sage Journals: Discover world-class research

Abstract

Background:

Parenchymal hematoma (PH) is a common complication of acute ischemic stroke, particularly following reperfusion therapy.

Objective:

This study aimed to explore the relationship between regional perfusion parameters and PH outcomes in stroke patients treated beyond the conventional time window.

Design:

This retrospective cohort study included patients from the CHinese Acute tissue-Based imaging selection for Lysis In Stroke–Tenecteplase (CHABLIS-T) trials and the Huashan Hospital stroke registry.

Methods:

Regional perfusion parameters were calculated within Alberta Stroke Program Early CT Score (ASPECTS)-defined regions of interest (ROIs). Mirror indices of cerebral blood flow (CBFmi), cerebral blood volume (CBVmi), and mean transit time were derived as the ratios of median perfusion values within ASPECTS-ROIs in the lesion and its contralateral hemisphere. Absolute time to maximum values for symptomatic ASPECTS-ROIs were also recorded. Logistic regression evaluated associations between perfusion parameters and PH outcomes, with predictive performance assessed using receiver operating characteristic (ROC) curves and area under the curve (AUC). Sensitivity analysis was conducted in patients receiving endovascular treatment (EVT) and in the trial-only population.

Results:

Of 1010 patients screened, 313 met the inclusion criteria, and 54 developed PH. Multivariable stepwise logistic regression identified reduced CBFmi (adjusted odds ratios (aOR) = 0.07, 95% confidence interval (CI), 0.02–0.30, p < 0.001) and CBVmi (aOR = 0.11, 95% CI, 0.03–0.45, p = 0.002) in the lentiform nucleus as significant predictors of PH. ROC analysis showed good discriminative performance (AUC: CBFmi 0.71 (95% CI, 0.62–0.80), CBVmi 0.70 (95% CI, 0.61–0.79)). Sensitivity analysis in patients undergoing EVT and trial-only patients drew similar results.

Conclusion:

Decreased CBFmi and CBVmi in the lentiform nucleus were independently associated with an elevated risk of PH, highlighting their potential utility in predicting hemorrhagic complications.

Trial registration:

NCT04086147, NCT04516993.

Keywords

hemorrhage ischemic stroke perfusion imaging reperfusion

Introduction

Intravenous thrombolysis (IVT) and endovascular treatment (EVT) are highly effective therapies for acute ischemic stroke (AIS). However, as a common complication of these treatments, hemorrhagic transformation significantly increases mortality and morbidity. Parenchymal hematoma (PH) is particularly concerning due to its profound impact on prognosis, with PH type II (PH2) being most strongly associated with poor outcomes.^1,2

The pathophysiological mechanisms underlying PH are complex and multifactorial, involving blood–brain barrier disruption, impaired cerebral perfusion, and regional susceptibility to ischemia.^3,4 Literature has demonstrated that perfusion parameters can reliably predict hemorrhagic outcomes and exhibit a strong correlation with blood–brain barrier integrity.^5–8 Since different brain regions show varying levels of vulnerability to ischemia,⁹ hemorrhagic risk varies based on the location of the ischemic infarct.^10,11 Assessing regional sensitivity is therefore crucial for predicting the fate of hypoperfused tissue and its associated hemorrhagic risks. However, neuroimaging markers associated with this spatial heterogeneity remain poorly understood.

Advances in imaging algorithms have extended the reperfusion time window to 24 h from stroke onset. However, the risk of PH in the extended time window remains insufficiently studied. Existing imaging markers may not capture regional variability or adapt to the evolving reperfusion strategies, both of which are essential for accurately predicting hemorrhagic risk.^12,13

In this study, we comprehensively analyzed regional perfusion parameters to evaluate how ischemia severity in different brain regions influences PH risk in acute stroke caused by arterial stenosis or occlusion, treated with reperfusion therapy in an extended time window.

Methods

Study design and participants

A retrospective analysis was performed on patients with AIS who participated in the CHinese Acute Tissue-Based Imaging Selection for Lysis In Stroke-Tenecteplase (CHABLIS-T) clinical trials (CHABLIS-T1 NCT04086147, CHABLIS-T2 NCT04516993)^14,15 or were enrolled in the Huashan Hospital stroke registry between April 2015 and January 2023. An overview of the eligibility criteria for the CHABLIS clinical trials and the Huashan Hospital stroke registry and treatment administration was provided in the Supplemental Materials.

Patients were included if they met the following criteria: (1) presented within 24 h of time of last known well (TLKW); (2) underwent a complete baseline multimodal computed tomography (CT) scan, including non-contrast CT (NCCT), CT angiography (CTA), and CT perfusion (CTP); (3) had complete clinical baseline profiles; (4) exhibited occlusion or severe stenosis (>70% vascular caliber stenosis in the ipsilateral artery on CTA, using the adjacent contact segment diameter as a reference) in the extracranial or intracranial internal carotid artery or the M1/M2 segment of the middle cerebral artery (MCA); and (5) underwent emergent IVT or EVT in an extended time window (>4.5 h). Exclusion criteria were as follows: patients with (1) absent or poor-quality multimodal CT source images and (2) failed post-processing of CTP images due to unsuccessful registration or incomplete field of view.

The reporting of this study conforms to STROBE guidelines.¹⁶ Ethics approval was obtained from the Ethics Committee of each participating center (for details, please see the ethics approval statement at end of the article). All participants or their legally authorized representatives provided written informed consent prior to inclusion, in accordance with the Declaration of Helsinki.

Data collection

Clinical information, including demographics, medical history, and treatment details, was recorded by trained neurologists. Blood pressure and glucose levels measured upon emergency room admission were defined as baseline values. Stroke severity was assessed using the National Institutes of Health Stroke Scale (NIHSS) at baseline. Stroke subtypes were classified according to the Trial of ORG 10172 in Acute Stroke Treatment criteria.¹⁷

Imaging protocol and analysis

All eligible patients received NCCT, CTP, and CTA upon arrival and underwent follow-up NCCT within 24 h after reperfusion therapy. This follow-up imaging was performed in alignment with standardized protocols to ensure consistency across all cases.

NCCT images were evaluated for focal parenchymal low attenuation, diminished gray-white matter differentiation, and sulcal effacement using the Alberta Stroke Program Early CT Score (ASPECTS).¹⁸ Baseline and follow-up NCCT images were independently interpreted by two neurologists (X.L. and X.W.), blinded to clinical data, with a third neurologist (Z.H.) consulted to resolve discrepancies.

Regions of interest (ROIs) for ASPECTS were defined within the Montreal Neurological Institute (MNI) space to construct the MNI-ASPECTS template. CTP source images were aligned with the standardized MNI-ASPECTS template using an affine transformation via the commercial software F-STROKE (Neuroblam, Ltd. Co., Shanghai, China).

To account for inter-individual variability, the mirror indices of cerebral blood flow (CBFmi), cerebral blood volume (CBVmi), and mean transit time (MTTmi) were calculated as the ratios of the median absolute perfusion values of each ASPECTS-ROI in the symptomatic hemisphere to the corresponding ROI in the contralateral hemisphere. In addition, absolute time to maximum (Tmax) values for ASPECTS-ROIs in the symptomatic hemisphere were recorded, which directly reflected the hemodynamic delay in tissue perfusion (Figure S1).

CTP source images were post-processed automatically using the commercial software F-STROKE to generate CBF and Tmax maps. Hypoperfused tissue volume was calculated based on Tmax >6 s. The ischemic core was defined as regions with relative CBF <30% within the hypoperfused lesion. The hypoperfusion intensity ratio (HIR) was computed as the ratio of brain volume with Tmax >10 s to that with Tmax >6 s.

Outcome measurement

PH was evaluated on 24-h follow-up NCCT by two experienced neurologists blinded to clinical data, following the European Cooperative Acute Stroke Study-II protocol.¹⁹ PH was classified into PH type I (PH1) and PH2, with PH2 defined as blood clots occupying more than 30% of the infarcted area with significant mass effect. PH, encompassing both PH1 and PH2, was the primary outcome, with PH2 analyzed separately as severe hemorrhagic transformation. The anatomical location of PH was further identified based on the ASPECTS template to characterize the brain regions involved.

In this study, both NCCT-based ASPECTS and CTP-derived perfusion parameters (CBFmi, CBVmi, MTTmi, Tmax values, hypoperfused tissue volume, ischemic core volume, and HIR) were tested for their associations with PH and PH2.

Statistical analysis

The Shapiro–Wilk test was used to assess the normality of continuous data. Continuous variables were summarized as mean ± standard deviation for normally distributed data or as median (interquartile range, IQR) for non-normal data. Categorical variables were described as frequencies and percentages. Continuous variables between groups were compared using the t test or Wilcoxon rank-sum test, while categorical variables were analyzed with the chi-squared or Fisher’s exact test. These univariate analyses were exploratory, aimed at identifying candidate variables for inclusion in subsequent multivariable regression. Statistical significance was set at a two-tailed p value <0.05 for all tests. Interrater agreement for ASPECTS was evaluated using quadratic weighted Kappa statistics on a randomly selected subset of 63 (20%) baseline NCCT scans. Backward stepwise logistic regression was performed to identify predictors of PH and PH2, including variables with p < 0.05 in univariate analysis or deemed clinically significant (e.g., age, previous stroke or transient ischemic stroke, systolic blood pressure, glucose levels, TLKW to reperfusion, IVT, and EVT). The variance inflation factors (VIFs) were calculated to assess multicollinearity, with VIF <5 considered acceptable. Adjusted odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. Receiver operating characteristic (ROC) curves were plotted to assess the diagnostic performance of each significant variable in distinguishing PH and PH2, with the area under the curve (AUC) calculated. Optimal cutoff values were determined using the Youden index, and sensitivity and specificity at the cutoff were reported. To account for the potential influence of EVT-related procedures on PH and PH2, a sensitivity analysis was conducted in a subset of patients undergoing EVT. In addition, a sensitivity analysis was performed, including only CHABLIS-T trial patients. Statistical analyses and visualizations were conducted using RStudio (version 2024.12.0 Build 467) with R 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org) and Stata v18.0 (StataCorp, College Station, TX, USA).

Results

Baseline characteristics

Among 1010 stroke patients with anterior large vessel severe stenosis or occlusion in the CHABLIS-T clinical trials or Huashan Hospital stroke registry, 313 met the inclusion criteria (Figure 1). Of these, 54 (17.3%) developed PH (Table S1). Among these, most PH occurred in deep structures, including the lentiform nucleus, internal capsule, and caudate, whereas cortical and insular involvement was less frequent (Table S2). The median (IQR) age was 68 (58, 75) years and the baseline median (IQR) NIHSS was 11 (7, 15). Among the patients, 99 (31.6%) were female. A total of 95 (30.4%) patients received IVT alone, 128 (40.9%) received EVT alone, and 90 (28.8%) underwent bridging therapy (combined IVT and EVT). Automated CTP analysis revealed a median (IQR) ischemic core volume of 3 (0, 12) ml and a median (IQR) hypoperfusion lesion volume of 120 (70, 170) ml across all patients. The median (IQR) HIR was 0.17 (0.05, 0.35) and median (IQR) ASPECTS was 8 (7, 9). The inter-rater agreement for ASPECTS was moderate, with a quadratic weighted κ value of 0.58 (95% CI, 0.41–0.75; p < 0.001). Table 1 summarizes the complete baseline demographic, clinical, and imaging characteristics. The distribution of perfusion parameters stratified by PH is provided in Table S3. Compared to non-PH patients, those with PH had a higher prevalence of atrial fibrillation (37.0% vs 19.7%; p = 0.011), a higher baseline NIHSS score (median, 14 vs 11; p < 0.001), and required more EVT passes (median, 2 vs 1; p = 0.002) to reach recanalization. Patients with poorer imaging profiles showed a higher likelihood of developing PH, characterized by lower ASPECTS (median, 7 vs 8; p < 0.001), larger ischemic core volume (median, 9 vs 1 ml; p < 0.001), greater hypoperfusion lesion volume (median, 156 vs 113 ml; p < 0.001), and higher HIR (median, 0.30 vs 0.14; p < 0.001). Further stratification by PH2 and corresponding baseline characteristics is summarized in Table S4.

Figure 1.

Patient inclusion flowchart.

Table 1.

Baseline characteristics of study cases and stratified by PH.

Item	Non PH(n = 259)	PH(n = 54)	p Value
Demographics
Age, years, median (IQR)	68 (57, 75)	71 (61, 76)	0.153
Sex, female, n (%)	78 (30.1)	21 (38.9)	0.260
Medical history, n (%)
Hypertension	160 (61.8)	40 (74.1)	0.119
Atrial fibrillation	51 (19.7)	20 (37.0)	0.011
Diabetes mellitus	73 (28.2)	15 (27.8)	>0.999
Previous stroke or TIA	46 (17.8)	11 (20.4)	0.699
Cardiovascular disease	22 (8.5)	5 (9.3)	0.793
Antiplatelets	35 (13.5)	12 (22.2)	0.140
Smoking	113 (43.6)	23 (42.6)	>0.999
Baseline data
NIHSS, median (IQR)	11 (6, 15)	14 (11, 18)	<0.001
SBP, mmHg, median (IQR)	145 (130, 162)	147 (133, 163)	0.916
DBP, mmHg, median (IQR)	84 (77, 94)	84 (75, 94)	0.767
Glucose, mmol/L, median (IQR)	6.80 (5.90, 8.70)	7.00 (6.28, 8.80)	0.424
IVT, n (%)	157 (60.6)	28 (51.9)	0.287
Drug of IVT, n (%)^a			0.260
Alteplase or urokinase	27 (17.2)	2 (7.1)
Tenecteplase	130 (82.8)	26 (92.9)
Endovascular thrombectomy, n (%)	174 (67.2)	44 (81.5)	0.050
Number of passes, median (IQR)^b	1 (0, 2)	2 (1, 3)	0.002
Angioplasty, n (%)	77 (44.3)	12 (27.3)	0.058
Stenting, n (%)	68 (39.1)	10 (22.7)	0.053
Angioplasty or stenting, n (%)	94 (36.3)	14 (25.9)	0.159
GP2b3a inhibitors, n (%)	66 (37.9)	10 (22.7)	0.076
LMWH, n (%)	34 (19.5)	8 (18.2)	>0.999
TLKW to door, min, median (IQR)	519 (345, 742)	436 (260, 693)	0.185
DNT, min, median (IQR)^a	114 (81, 163)	113 (93, 147)	0.953
DTP, min, median (IQR)^b	171 (128, 227)	159 (127, 196)	0.248
TLKW to treatment, min, median (IQR)	681 (485, 907)	640 (418, 795)	0.127
Imaging profile
Location of occlusion or stenosis, n (%)			0.013
MCA	182 (70.3)	38 (70.4)
ICA	62 (23.9)	7 (13.0)
Tandem	15 (5.8)	9 (16.7)
ASPECTS, median (IQR)	8 (7, 9)	7 (6, 8)	<0.001
Ischemic core volume, ml, median (IQR)	1 (0, 9)	9 (3, 29)	<0.001
Hypoperfusion lesion volume, ml, median (IQR)	113 (63, 158)	156 (105, 194)	0.001
HIR, median (IQR)	0.14 (0.04, 0.33)	0.30 (0.17, 0.44)	<0.001
TOAST, n (%)			0.026
LAA	166 (64.1)	26 (48.1)
CE	55 (21.2)	21 (38.9)
Others	38 (14.7)	7 (13.0)

Results are expressed as number (%) or median (IQR).

n = 185.

n = 218.

ASPECTS, Alberta Stroke Program Early CT Score; CE, cardioembolism; DBP, diastolic blood pressure; DNT, door-to-needle time; DTP, door-to-puncture time; HIR, hypoperfusion intensity ratio; ICA, internal carotid artery; IQR, interquartile range; IVT, intravenous thrombolysis; LAA, large-artery atherosclerosis; LMWH, low-molecular-weight heparin; MCA, middle cerebral artery; NIHSS, National Institute of Health stroke scale; PH, parenchymal hematoma; SBP, systolic blood pressure; TIA, transient ischemic stroke; TLKW, time last known well; TOAST, trial of Org 10172 in acute stroke treatment.

Association between perfusion parameters and PH outcomes

Compared to patients without PH, patients with PH exhibited significantly lower CBFmi levels in the lentiform nucleus (L, p < 0.001), insular ribbon (I, p < 0.001), caudate (C, p = 0.030), MCA cortex lateral to I (M2, p = 0.010), posterior MCA cortex (M3, p = 0.003), and anterior MCA territories superior to the anterior MCA cortex (M4, p = 0.043). Similarly, CBVmi levels were lower in L (p < 0.001), I (p < 0.001), M2 (p = 0.017), and M3 (p = 0.005). Moreover, Tmax was prolonged in L (p = 0.001), I (p < 0.001), the anterior MCA cortex (M1, p < 0.001), M2 (p < 0.001), M3 (p < 0.001), and M4 (p = 0.023; Table S3). All of the variables had VIFs of <5, suggesting that there was no significant collinearity among them. Backward stepwise logistic regression analysis, with baseline NIHSS and ASPECTS (both p < 0.05) entered in the model, identified lower L-CBFmi (aOR = 0.07, 95% CI, 0.02–0.30, p < 0.001) and lower L-CBVmi (aOR = 0.11, 95% CI, 0.03–0.45, p = 0.002) as independent predictors of PH (Table 2). Similarly, lower L-CBFmi (OR = 0.01, 95% CI, 0.00–0.03, p < 0.001) and lower L-CBVmi (OR = 0.01, 95% CI, 0.00–0.05, p < 0.001) remained significantly associated with PH2 (Table S5). ROC analysis showed that L-CBFmi (AUC, 0.71, 95% CI, 0.62–0.80) and L-CBVmi (AUC, 0.70, 95% CI, 0.61–0.79) had satisfactory discriminative abilities for PH (Figure 2). The optimal cutoff values were 0.64 for L-CBFmi (sensitivity 56%, specificity 85%) and 0.67 for L-CBVmi (sensitivity 56%, specificity 88%). For PH2, L-CBFmi (AUC, 0.79, 95% CI, 0.70–0.88) and L-CBVmi (AUC, 0.78, 95% CI, 0.67–0.88) also demonstrated strong discriminative performance (Figure S2).

Table 2.

Association between perfusion parameters and PH.

Perfusion parameters and PH	B	Adjusted OR (95% CI, p)
CBFmi
L-CBFmi	−2.65	0.07 (0.02–0.30, p < 0.001)
NIHSS	0.06	1.07 (1.02–1.12, p = 0.010)
ASPECTS	−0.26	0.77 (0.63–0.94, p = 0.009)
CBVmi
L-CBVmi	−2.17	0.11 (0.03–0.45, p = 0.002)
NIHSS	0.06	1.07 (1.01–1.12, p = 0.011)
ASPECTS	−0.27	0.76 (0.63–0.93, p = 0.007)

ASPECTS, Alberta Stroke Program Early CT Score; CBFmi, cerebral blood flow mirror index; CBVmi, cerebral blood volume mirror index; CI, confidence interval; L, lentiform nucleus; NIHSS, National Institute of Health stroke scale; OR, odds ratio; PH, parenchymal hematoma.

Figure 2.

ROC curves for predicting PH. AUC of L-CBFmi was 0.71 (95% CI, 0.62–0.8); AUC of L-CBVmi was 0.7 (95% CI, 0.61–0.79).

Sensitivity analysis

For the sensitivity analysis in patients treated with EVT, 218 patients were included in the analysis. With baseline NIHSS, ASPECTS, and the number of EVT passes being entered in the model as covariates, lower L-CBFmi (aOR = 0.12, 95% CI, 0.02–0.67, p = 0.017) and lower L-CBVmi (aOR = 0.14, 95% CI, 0.02–0.73, p = 0.014) remained significant independent predictors of PH (Table 3). For PH2, only L-CBFmi (OR = 0.01, 95% CI, 0.00–0.04, p < 0.001) and L-CBVmi (OR = 0.01, 95% CI, 0.00–0.04, p < 0.001) remained significantly associated with the outcome (Table S6). A sensitivity analysis including only CHABLIS-T trial patients was also performed, with results consistent with the main analysis (Table S7).

Table 3.

Association between perfusion parameters and PH in patients treated with EVT.

Item	Unadjusted OR (95% CI, p)	Adjusted OR (95% CI, p)
Patients with EVT (n = 218)
CBFmi
L-CBFmi	0.03 (0.01–0.15, p < 0.001)	0.12 (0.02–0.67, p = 0.017)^a
CBVmi
L-CBVmi	0.03 (0.01–0.14, p < 0.001)	0.14 (0.02–0.73, p = 0.022)^a

Adjusted for NIHSS, ASPECTS, and number of passes.

ASPECTS, Alberta Stroke Program Early CT Score; CBFmi, cerebral blood flow mirror index; CBVmi, cerebral blood volume mirror index; CI, confidence interval; EVT, endovascular thrombectomy; L, lentiform nucleus; NIHSS, National Institute of Health stroke scale; OR, odds ratio; PH, parenchymal hematoma.

Discussion

Our study is one of the first studies to explore the role of regional perfusion parameters in the extended time window, identifying L-CBFmi and L-CBVmi as independent predictors of PH outcomes.

PH is a frequent and serious complication of reperfusion therapy for AIS, significantly limiting treatment efficacy and leading to poor outcomes.²⁰ Accurate prediction of PH is essential to evaluate hemorrhagic risk, guide clinical decision-making, and optimize the management of patients during the hyperacute phase of stroke. Advances in imaging have extended the therapeutic window for reperfusion therapy beyond the traditional 4.5-h limit, enabling treatment decisions based on tissue viability rather than strict time constraints. This evolution underscores the need for precise tools to predict complications such as PH, particularly in extended time windows. Regional perfusion heterogeneity has been increasingly recognized as a critical determinant of hemorrhagic complications. Certain brain regions exhibit disproportionate susceptibility to ischemia and subsequent PH due to their distinct anatomical and metabolic properties.^10,11 However, despite advancements in imaging modalities, neuroimaging markers that can reliably capture this spatial variability remain underexplored. Our study aimed to bridge this gap by evaluating the role of regional perfusion parameters in predicting PH. The findings indicate that reduced L-CBFmi and L-CBVmi were consistently associated with an increased risk of PH and PH2, emphasizing the importance of assessing regional ischemic heterogeneity in clinical practice.

The lentiform nucleus, a critical subcortical structure within the basal ganglia, has been identified as a high-risk region for ischemia and subsequent PH. The susceptibility of the lentiform nucleus can be attributed to its unique vascular supply and limited collateral circulation.^9,21 The region is primarily supplied by the lenticulostriate arteries (LSAs), small perforating branches that originate from the M1 segment of MCA. These arteries are considered end arteries, with limited potential for collateral flow to compensate for ischemia.²² Therefore, the tissue-protective effects of collateral circulation, which mitigate ischemic damage in cortical regions, are notably weaker in deep structures like the lentiform nucleus.²³ In line with this, PH was frequently involved in the lentiform nucleus in our cohort. This vulnerability is reflected in the high prevalence of early ischemic changes and obscuration of the lentiform nucleus observed on imaging, both of which are associated with severe infarctions and an increased hemorrhagic risk.²⁴ Moreover, the structural fragility of the LSAs renders them prone to being damaged under conditions of ischemia-reperfusion, particularly when exposed to hemodynamic stress. Studies have shown that vertical branching of the LSAs from the M1 segment increases their vulnerability to high flow velocities during reperfusion, further predisposing the lentiform nucleus to PH.^25,26 Previous studies have demonstrated a relationship between poor perfusion and infarction in subcortical structures, particularly the lentiform nucleus, and hemorrhagic outcomes.^10,26,27 Our findings are consistent with these observations, demonstrating that reduced L-CBFmi and L-CBVmi independently predict PH. These results highlight the lentiform nucleus as a critical region of interest in assessing hemorrhagic risk for patients undergoing reperfusion therapy beyond the traditional time window. The derived cutoff values of L-CBFmi and L-CBVmi may assist in identifying patients at higher risk of PH, although these thresholds require validation in larger, independent cohorts.

Consistent with previous studies,⁶ our findings suggest that CBFmi and CBVmi are more reliable predictors of PH risk compared to other perfusion metrics such as MTTmi or Tmax. This can be attributed to their ability to directly reflect microcirculatory integrity during the vasogenic edema phase,²⁸ whereas Tmax and MTT primarily capture hemodynamic alterations in vascular status. Notably, CBF shows a strong inverse correlation with blood–brain barrier disruption,²⁹ highlighting its relevance in assessing microvascular injury. Furthermore, the strong association between CBV-ASPECTS and final infarct volume supports the pathophysiological relevance of CBV as a direct indicator of tissue viability,³⁰ which may explain their established predictive value for hemorrhagic outcomes.³¹ The mirror index, used to normalize regional perfusion values against the contralateral side, was instrumental in reducing inter-individual variability and enhancing the sensitivity of these measurements. However, it is important to note that the utility of the mirror index may be limited in cases of bilateral cerebral involvement, necessitating alternative approaches for such scenarios.

EVT is now a routine practice for acute stroke patients with large vessel occlusions. However, EVT itself may introduce additional risks for PH. The mechanical nature of EVT, particularly when multiple passes are required, can cause direct injury to the arterial intima, disrupt the blood–brain barrier, and initiate a cascade of reperfusion-related injury.³² Notably, our study found that L-CBFmi and L-CBVmi remained significant predictors of PH even when procedural factors such as the number of device passes were entered in the stepwise model. This suggests that regional perfusion heterogeneity, particularly in the lentiform nucleus, plays a robust and independent role in determining hemorrhagic outcomes, irrespective of procedural variables.

The ability to accurately predict PH has important implications for optimizing stroke care. Our findings suggested that incorporating regional perfusion parameters such as L-CBFmi and L-CBVmi into clinical workflows could enhance risk stratification for patients undergoing reperfusion therapy, apart from the previously acknowledged risk factors, such as baseline NIHSS and ASPECTS.^20,33 Specifically, patients with significantly reduced perfusion in the lentiform nucleus may benefit from enhanced monitoring and individualized therapeutic strategies aimed at mitigating hemorrhagic risk. In addition, the findings reinforce the importance of integrating advanced imaging modalities into routine stroke care, particularly for patients treated in extended time windows. Tissue-based imaging not only guides reperfusion decisions but also provides valuable insights into regional ischemic heterogeneity, offering an insightful understanding of patient-specific risks and outcomes.

Despite its strengths, our study had several limitations. First, the data were derived from different sources: the Huashan Hospital stroke registry and the CHABLIS-T clinical trials. The differences in baseline characteristics, intervention timing, and underlying etiologies between these cohorts may introduce bias. To address this, a sensitivity analysis including only CHABLIS-T trial patients was performed, and the findings were consistent with our main analysis. Second, our study did not account for certain modifiable factors, such as perioperative blood pressure control and antithrombotic therapy, which might influence hemorrhagic outcomes. Incorporating these factors in future research would provide a more comprehensive understanding of PH risk. Third, the CTP source images were registered to the MNI-ASPECTS template as part of our methodology. Given the small ROIs used in ASPECTS scoring, this step might introduce errors in both registration and measurement, potentially affecting the reliability of our findings. To enhance accuracy, further development of the template and registration methods is warranted. Fourth, symptomatic intracranial hemorrhage (sICH) was not included as an outcome because the number of sICH events was too small to allow for reliable statistical analysis. While imaging-defined PH remains clinically meaningful and widely used in reperfusion studies, future prospective research with larger sample sizes and systematic sICH evaluation is warranted to enhance comparability and translational relevance. Fifth, the post-processing methods employed in this study rely on specialized software, which may limit their clinical applicability. However, with ongoing technological advancements, we anticipate broader adoption of automated imaging post-processing, enabling more efficient assessments and assisting clinicians in rapid risk stratification. Finally, while our analysis focused on specific brain regions, we did not assign differential weights to regions across the entire brain. A comprehensive, weighted analysis could provide deeper insights into the relative contributions of different regions to PH risk.

Conclusion

Our study demonstrated that L-CBFmi and L-CBVmi independently predicted PH in patients treated beyond conventional therapeutic windows. The results highlighted the lentiform nucleus as a critical region in the pathogenesis of PH and emphasized the utility of regional perfusion parameters for risk stratification. These findings provide preliminary evidence for predicting PH risk, which may help guide clinical decision-making in AIS patients undergoing reperfusion therapy.

Supplemental Material

sj-docx-1-tan-10.1177_17562864251406032 – Supplemental material for Regional perfusion parameters as potential indicators of parenchymal hematoma risk following reperfusion therapy for acute ischemic stroke in the extended time window

Supplemental material, sj-docx-1-tan-10.1177_17562864251406032 for Regional perfusion parameters as potential indicators of parenchymal hematoma risk following reperfusion therapy for acute ischemic stroke in the extended time window by Xinyu Liu, Lan Hong, Guangjian Zhao, Zhijiao He, Xinru Wang, Juehua Zhu, Siyuan Li, Anqi Zhang, Nan Cao, Yifeng Ling, Xiangdi Chen, Ying Guo, Qi Fang, Ziran Wang, Qiang Dong and Xin Cheng in Therapeutic Advances in Neurological Disorders

Supplemental Material

sj-docx-2-tan-10.1177_17562864251406032 – Supplemental material for Regional perfusion parameters as potential indicators of parenchymal hematoma risk following reperfusion therapy for acute ischemic stroke in the extended time window

Supplemental material, sj-docx-2-tan-10.1177_17562864251406032 for Regional perfusion parameters as potential indicators of parenchymal hematoma risk following reperfusion therapy for acute ischemic stroke in the extended time window by Xinyu Liu, Lan Hong, Guangjian Zhao, Zhijiao He, Xinru Wang, Juehua Zhu, Siyuan Li, Anqi Zhang, Nan Cao, Yifeng Ling, Xiangdi Chen, Ying Guo, Qi Fang, Ziran Wang, Qiang Dong and Xin Cheng in Therapeutic Advances in Neurological Disorders

Footnotes

Acknowledgements

We thank every CHABLIS Collaborations investigator for participating in the trials.

Declarations

ORCID iDs

Xinyu Liu

Lan Hong

Zhijiao He

Nan Cao

Qi Fang

Xin Cheng

Supplemental material

Supplemental material for this article is available online.

References

Heit

Mlynash

Christensen

, et al. What predicts poor outcome after successful thrombectomy in late time windows? J Neurointerv Surg 2021; 13: 421–425.

Ospel

Qiu

Menon

, et al. Radiologic patterns of intracranial hemorrhage and clinical outcome after endovascular treatment in acute ischemic stroke: results from the ESCAPE-NA1 trial. Radiology 2021; 300: 402–409.

Spronk

Sykes

Falcione

, et al. Hemorrhagic transformation in ischemic stroke and the role of inflammation. Front Neurol 2021; 12: 661955.

Arba

Piccardi

Palumbo

, et al. Blood–brain barrier leakage and hemorrhagic transformation: the Reperfusion Injury in Ischemic StroKe (RISK) study. Euro J Neurol 2021; 28: 3147–3154.

Langel

Popovic

KS.

Infarct-core CT perfusion parameters in predicting post-thrombolysis hemorrhagic transformation of acute ischemic stroke. Radiol Oncol 2018; 53: 25–30.

Renú

Laredo

Tudela

, et al. Brain hemorrhage after endovascular reperfusion therapy of ischemic stroke: a threshold-finding whole-brain perfusion CT study. J Cereb Blood Flow Metab 2017; 37: 153–165.

Souza

LCS

Payabvash

Wang

, et al. Admission CT perfusion is an independent predictor of hemorrhagic transformation in acute stroke with similar accuracy to DWI. Cerebrovasc Dis 2012; 33: 8–15.

Yassi

Parsons

Christensen

, et al. Prediction of poststroke hemorrhagic transformation using computed tomography perfusion. Stroke 2013; 44: 3039–3043.

Payabvash

Souza

LCS

Wang

, et al. Regional ischemic vulnerability of the brain to hypoperfusion. Stroke 2011; 42: 1255–1260.

10.

G-D

Hang

, et al. Association between infarct location and hemorrhagic transformation of acute ischemic stroke following successful recanalization after mechanical thrombectomy. AJNR Am J Neuroradiol 2023; 44: 54–59.

11.

Phan

Donnan

Koga

, et al. The ASPECTS template is weighted in favor of the striatocapsular region. Neuroimage 2006; 31: 477–481.

12.

Campbell

BCV

Ringleb

, et al. Extending thrombolysis to 4·5–9 h and wake-up stroke using perfusion imaging: a systematic review and meta-analysis of individual patient data. Lancet 2019; 394: 139–147.

13.

Campbell

BCV

Parsons

, et al. Thrombolysis guided by perfusion imaging up to 9 hours after onset of stroke. N Engl J Med 2019; 380: 1795–1803.

14.

Cheng

Hong

Churilov

, et al. Tenecteplase thrombolysis for stroke up to 24 hours after onset with perfusion imaging selection: the umbrella phase IIa CHABLIS-T randomised clinical trial. Stroke Vasc Neurol 2024; 9(5): 551–559.

15.

Cheng

Hong

Lin

, et al. Tenecteplase thrombolysis for stroke up to 24 hours after onset with perfusion imaging selection: the CHABLIS-T II randomized clinical trial. Stroke 2025; 56(2): 344–354.

16.

von Elm

Altman

Egger

, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med 2007; 147: 573–577.

17.

Adams

Bendixen

Kappelle

, et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 1993; 24: 35–41.

18.

Barber

Demchuk

Zhang

, et al. Validity and reliability of a quantitative computed tomography score in predicting outcome of hyperacute stroke before thrombolytic therapy. Lancet 2000; 355: 1670–1674.

19.

Hacke

Kaste

Fieschi

, et al. Randomised double-blind placebo-controlled trial of thrombolytic therapy with intravenous alteplase in acute ischaemic stroke (ECASS II). Lancet 1998; 352: 1245–1251.

20.

Kaesmacher

Maegerlein

, et al. Hemorrhagic transformations after thrombectomy: risk factors and clinical relevance. Cerebrovasc Dis 2017; 43: 294–304.

21.

Verma

Gralla

Klinger-Gratz

, et al. Infarction distribution pattern in acute stroke may predict the extent of leptomeningeal collaterals. PLoS One 2015; 10: e0137292.

22.

Marinkovic

Milisavljevic

Kovacevic

, et al. Perforating branches of the middle cerebral artery. Microanatomy and clinical significance of their intracerebral segments. Stroke 1985; 16: 1022–1029.

23.

Rosso

Blanc

, et al. Impact of infarct location on functional outcome following endovascular therapy for stroke. J Neurol Neurosurg Psychiatry 2019; 90: 313–319.

24.

Scott

Buchan

Sevick

RJ.

Correlation of neurologic dysfunction with CT findings in early acute stroke. Can J Neurol Sci 1999; 26: 182–189.

25.

Yokogami

Nakano

Ohta

, et al. Prediction of hemorrhagic complications after thrombolytic therapy for middle cerebral artery occlusion: value of pre- and post-therapeutic computed tomographic findings and angiographic occlusive site. Neurosurgery 1996; 39: 1102–1107.

26.

Horie

Morofuji

Iki

, et al. Impact of basal ganglia damage after successful endovascular recanalization for acute ischemic stroke involving lenticulostriate arteries. J Neurosurg 2020; 132: 1880–1888.

27.

Mokin

Morr

Fanous

, et al. Correlation between cerebral blood volume values and outcomes in endovascular therapy for acute ischemic stroke. J Neurointerv Surg 2014; 7: 705–708.

28.

Bisdas

Hartel

Cheong

, et al. Detection of early vessel leakiness in acute ischemic stroke using computed tomography perfusion may indicate hemorrhagic transformation. Acta Radiol 2007; 48: 341–344.

29.

Churilov

Yassi

, et al. Microvascular dysfunction in blood-brain barrier disruption and hypoperfusion within the infarct posttreatment are associated with cerebral edema. Stroke 2022; 53: 1597–1605.

30.

Chu

X-Q

, et al. Total and regional ASPECT score for non-contrast CT, CT angiography, and CT perfusion: inter-rater agreement and its association with the final infarction in acute ischemic stroke patients. Acta Radiol 2022; 63: 1093–1101.

31.

Lin

Zink

Tsiouris

, et al. Risk assessment of hemorrhagic transformation of acute middle cerebral artery stroke using multimodal CT. J Neuroimaging 2010; 22: 160–166.

32.

Sun

Lam

Christie

, et al. Risk factors of hemorrhagic transformation in acute ischaemic stroke: a systematic review and meta-analysis. Front Neurol 2023; 14: 1079205.

33.

Hao

Yang

Wang

, et al. Predictors for symptomatic intracranial hemorrhage after endovascular treatment of acute ischemic stroke. Stroke 2017; 48: 1203–1209.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.86 MB

0.03 MB