Development and internal validation of a multimodal nomogram integrating clinical, imaging, and laboratory data to predict malignant brain edema after reperfusion in acute ischemic stroke

Introduction

Acute ischemic stroke (AIS) is a major global public health problem due to its high incidence, mortality, and disability rates (1). Anterior-circulation large-vessel occlusion involving the middle cerebral artery (MCA) or internal carotid artery (ICA) is particularly devastating (2). Extensive ischemia can lead to malignant brain edema (MBE), one of the most severe and life-threatening complications of AIS. Without timely and effective intervention, early mortality associated with MBE may reach 80% (3-5). Many patients require intensive care, and severe cases undergo decompressive hemicraniectomy to reduce intracranial pressure and preserve life (6). The early identification of patients at high risk of MBE before clinical deterioration is therefore critical for treatment decision-making. However, the pathophysiology of MBE is incompletely understood, and reliable early prediction tools remain scarce (7). Conventional imaging such as non-contrast computed tomography (CT) and magnetic resonance imaging (MRI) can quantify infarct size but are limited for early MBE prediction and often detect marked edema only after midline shift appears. With the development of functional imaging, CT perfusion (CTP) provides fine-grained hemodynamic quantification. Prior studies have suggested that perfusion metrics—such as infarct core volume, penumbral extent, and infarct-growth rate—are closely related to edema formation. Systemic post-stroke inflammation and metabolic dysregulation are also implicated; blood biomarkers including peripheral leukocyte count and lactate dehydrogenase (LDH) may have predictive value (8,9). Integrating imaging parameters with biomarkers may enable multi-factor models that improve early identification of patients at high risk for MBE. Based on this rationale, we integrated imaging and laboratory parameters to develop and internally validate a clinically practical nomogram for predicting early MBE after reperfusion in AIS, thereby informing decision-making. We present this article in accordance with the TRIPOD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1153/rc).

Methods

Study population

This retrospective single-center cohort study was conducted at Shanghai East Hospital, Tongji University School of Medicine. We screened patients treated for anterior-circulation large vessel occlusion (LVO) stroke between January 2018 and December 2023. The inclusion criteria were as follows: (I) age ≥18 years; (II) clinical AIS confirmed by CTP or MRI within the MCA or ICA territory; (III) onset-to-admission ≤24 hours; (IV) evaluable baseline imaging including non-contrast CT (NCCT), CTA, and CTP; (V) distal ICA or MCA-M1 occlusion on imaging; (VI) mechanical thrombectomy (MT), either direct MT or IV thrombolysis bridging to MT; (VII) no severe comorbidities that would markedly affect prognosis (e.g., refractory heart failure, end-stage liver/kidney disease); and (VIII) complete 90-day follow-up including modified Rankin Scale (mRS) and repeat imaging. The exclusion criteria were as follows: (I) severe artifact or incomplete baseline CTP precluding analysis; (II) bilateral large-artery occlusions; and (III) wake-up stroke or unknown onset time. Of 1,628 consecutive patients with suspected AIS who underwent CTP, 214 met the inclusion criteria and were analyzed. Figure 1 shows the screening process used to identify eligible patients for the study.

Figure 1 Study flow diagram of CTP-screened suspected AIS admissions [2018–2023]. From 1,628 consecutive patients who underwent CTP, exclusions included no mechanical thrombectomy, posterior circulation stroke, wake-up/unknown-onset stroke, poor-quality or incomplete CTP, severe comorbidities, and bilateral large-vessel occlusion, leaving 214 patients for analysis. The cohort was divided into MBE (n=62) and non-MBE (n=152); all completed 90-day follow-up. AIS, acute ischemic stroke; CTP, computed tomography perfusion; MBE, malignant brain edema.

Ethical approval

This retrospective study was approved by the Ethics Committee of Shanghai East Hospital (No. 2024YS-279) , and was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The requirement for informed consent was waived due to the retrospective design.

Clinical and imaging data collection

Baseline clinical data included sex, age, comorbidities such as hypertension and diabetes, admission National Institutes of Health Stroke Scale (NIHSS) and Glasgow Coma Scale (GCS) scores, and systolic and diastolic blood pressure (DBP). Laboratory variables were drawn from the first routine panel after admission and before any reperfusion therapy; for bridging therapy, we used the sample obtained before intravenous thrombolysis. The panel comprised complete blood count measures such as neutrophils, biochemistry including glycosylated hemoglobin, high-density lipoprotein cholesterol (HDL-C), and LDH, and coagulation indices such as D-dimer. For continuous laboratory predictors with <5% missingness, values were imputed using the median within outcome strata, MBE and non-MBE; the outcome MBE was not imputed. Before feature selection, all continuous predictors were Z-standardized to mean 0 and standard deviation (SD) 1 to ensure comparable penalization, and categorical variables had no missing data.

Imaging assessment

Baseline CTA and CTP were assessed. CTP was processed using MIStar (Apollo Medical Imaging Technology, Melbourne, Australia) with delay-corrected singular-value decomposition, automatically outputting voxelwise parameters and volumes (Figure 2). Candidate perfusion variables included:

• Low cerebral blood flow (CBF) volumes: relative CBF (rCBF) <40%, <35%, <30%, <25%, <20%, and <15% relative to normal tissue;
• Low-perfusion volumes: delay time (DT) >2, >3, >4, >6, >8, and >10 s;
• Hypoperfusion intensity ratio (HIR): DT >10 s volume fraction/DT >6 s volume fraction.

Figure 2 Pre reperfusion multimodal CT perfusion maps processed with MIStar. Tmax, rCBF, MTT, and CBV maps reveal an extensive left-MCA perfusion deficit; automated segmentation outlines infarct core (red) and penumbra (green) with a large mismatch, whereas the AIF curve and CTA source image confirm arterial occlusion. AIF, arterial input function; CBV, cerebral blood volume; CT, computed tomography; CTA, computed tomography angiography; MCA, middle cerebral artery; MTT, mean transit time; rCBF, relative cerebral blood flow.

Collateral status was scored with the modified American Society of Interventional and Therapeutic Neuroradiology/Society of Interventional Radiology collateral grading system (ASITN/SIR) 0–4 scale, where higher scores indicate better collateral flow. Two neuroradiologists independently scored CTA while blinded to outcomes and clinical data, and disagreements were resolved by discussion to consensus.

Infarct growth rate (IGR) was defined as the ischemic core volume on the baseline CTP divided by the interval from symptom onset to the baseline CTP time; the interval equals baseline CTP time minus onset time, in mL/h (10).

Assessment of MBE

The occurrence of MBE was assessed within 48–72 hours after MT by two experienced radiologists, who reviewed clinical data and imaging findings independently. The diagnosis of MBE was based on the presence of progressive neurological deterioration in the early postoperative period, such as decreased level of consciousness or pupillary abnormalities suggestive of elevated intracranial pressure, combined with radiological evidence of severe brain edema. Imaging criteria included large territory infarction with significant cerebral swelling, compression of the ventricles, and a midline shift of ≥5 mm (Figure 3). Patients who met both clinical and imaging criteria were classified into the MBE group, whereas those who did not were assigned to the non-MBE group. Neurological functional outcomes were evaluated at 90 days through outpatient visits or structured telephone interviews. An mRS ≤2 was defined as a favorable outcome, and 3–6 as an unfavorable outcome.

Figure 3 Follow-up non-contrast CT demonstrating malignant brain edema. A non-contrast CT scan obtained 24 hours after reperfusion therapy shows diffuse swelling of the left cerebral hemisphere, complete loss of gray-white matter differentiation, compression of the ipsilateral lateral ventricle, effacement of cortical sulci, and a marked midline shift toward the left. These findings are consistent with malignant brain edema. CT, computed tomography.

Statistical analysis

Analyses used R 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria) with glmnet, rms, pROC, and rmda; Python (Python Software Foundation, Wilmington, DE, USA) was used only to check figures. Tests were two-sided with alpha 0.05. For continuous laboratory predictors with less than 5% missingness, values were imputed using the median within outcome strata, MBE, and non-MBE; the MBE outcome was not imputed, and categorical variables had no missing data. Before feature selection, all continuous variables were Z-standardized to mean 0 and SD 1. Feature selection and multicollinearity handling used least absolute shrinkage and selection operator (LASSO) logistic regression with stratified five-fold cross-validation; the penalty lambda followed the 1-SE rule to yield a parsimonious, stable subset. Selected variables were entered on original measurement scales into a multivariable logistic regression; regression coefficients, odds ratios (ORs), and 95% confidence intervals (CIs) were reported, and a nomogram was constructed. Discrimination and accuracy were evaluated using stratified five-fold out-of-fold predictions with area under the curve (AUC) and Brier score, with fixed random seeds for reproducibility. Calibration used 1,000 bootstrap resamples with reporting of calibration slope and intercept and a bias-corrected calibration curve; a 10-group Hosmer-Lemeshow test was also performed. Clinical utility was examined with decision curve analysis (DCA) across threshold probabilities 0.10 to 0.70, and a clinical impact curve (CIC) was plotted. Robustness was assessed using ROC and calibration curves from out-of-fold predictions and by repeating modeling on complete-case data. The final model included seven predictors: rCBF <40%, DT >6.0 s, IGR, DBP, N, collateral circulation score (CCS), and LDH.

Results

Patient characteristics

We included 214 eligible AIS patients: 130 men (60.7%) and 84 women (39.3%); mean age was 66 years. All underwent endovascular reperfusion for anterior-circulation LVO. Median admission NIHSS was 17 in the MBE group and 14 in the non-MBE group, indicating more severe deficits with MBE (P<0.05), whereas age and sex did not differ significantly. Comorbidity burdens (e.g., hypertension, diabetes) were similar between groups. Notably, admission DBP was higher in the MBE group than it was in the non-MBE group (89 vs. 80 mmHg, P=0.001). The proportion receiving bridging IV thrombolysis did not differ. Baseline clinical and imaging data are shown in Tables 1-3.

Univariate analysis of imaging and laboratory variables

Compared with the non-MBE group, the MBE group had larger infarct core and hypoperfusion volumes. Across DT thresholds (e.g., >4, >6, >8 s), hypoperfused regions were significantly larger in the MBE group (all P<0.001). For DT >6.0 s, median volume was 89.5 mL in MBE versus 36.0 mL in non-MBE (P<0.001). IGR was higher with MBE (median 28.3 vs. 4.95 mL/h, P<0.001), indicating faster expansion. CCS was lower with MBE, typically 1–2 versus 2–3 (P<0.001), consistent with poorer collateral compensation. In laboratories, neutrophil count (N) was higher with MBE (median 9.0 vs. 7.25 ×10⁹/L, P=0.001), and serum LDH was elevated (median 286 vs. 224 U/L, P=0.001). Other indices such as fasting glucose, D-dimer, and C-reactive protein did not differ significantly. These findings suggest that several imaging and inflammatory and metabolic markers are strongly associated with MBE.

Variable selection and final model construction

We applied LASSO logistic regression to select features and address multicollinearity. All continuous predictors were Z-standardized. Using stratified five-fold cross-validation and the one-standard-error rule, LASSO yielded a stable and parsimonious subset. We then fit a multivariable logistic model on original measurement scales and derived a nomogram from the estimated coefficients. The final model retained seven predictors: rCBF <40% volume, DT >6.0 s volume, IGR, DBP, N, CCS, and LDH. Table 4 reports regression coefficients, OR, and 95% CI. P-values are descriptive and did not determine variable retention. Single-predictor ROC AUCs were: rCBF <40% 0.793, IGR 0.788, DT >6.0 s 0.771, LDH 0.657, neutrophils 0.643, DBP 0.640, and CCS 0.322. Figure 4 illustrates the ROC curves for these single predictors. Figure 5 shows the nomogram, which assigns points in proportion to each predictor’s contribution and converts the total to an individualized probability of MBE. This workflow preserves interpretability, constrains model complexity through LASSO, handles correlated metrics, and limits overfitting.

Figure 4 ROC curves showing the discriminative ability of the predictive model for MBE. AUC, area under the curve; CCS, collateral circulation score; DBP, diastolic blood pressure; DT, delay time; IGR, infarct growth rate; LDH, lactate dehydrogenase; MBE, malignant brain edema; N, neutrophil count; rCBF, relative cerebral blood flow; ROC, receiver operating characteristic.

Figure 5 The nomogram model for predicting the probability of malignant brain edema in patients after reperfusion therapy. CCS, collateral circulation score; DBP, diastolic blood pressure; DT, delay time; IGR, infarct growth rate; LDH, lactate dehydrogenase; N, neutrophil count; rCBF, relative cerebral blood flow.

Nomogram performance

In the full modeling cohort (n=214; MBE 62, 29.0%), apparent AUC in the modeling cohort was 0.87 (95% CI: 0.82–0.93); five-fold out-of-fold AUC was 0.860. After 1,000-bootstrap optimism correction, the AUC was 0.85 (95% CI: 0.80–0.91) with a Brier score of 0.13. The calibration slope was ~0.86, the intercept was near 0, and Hosmer-Lemeshow P=0.173. The apparent calibration slope and intercept were 1.000 (0.734–1.309) and 0.000 (−0.396–0.381); optimism-corrected values were 0.859 (0.609–1.123) and −0.007 (−0.402–0.441). The Hosmer-Lemeshow statistic was χ²=11.546, df=8, P=0.173, indicating good fit. Figure 6 shows the calibration curve. The integrated model significantly outperformed the best single predictor (DeLong P=0.003). DCA showed higher net benefit than “treat-all” or “treat-none” at thresholds of 10%, 20%, 40%, and 60% with net benefit per 100 patients of 12.6, 38.8, 50.0, and 58.9, respectively. Figure 7 shows the decision curves. On the CIC, among 100 patients at a 20% threshold, 42.5 were predicted high-risk with 25.2 true MBE cases; at 60%, 20.1 were predicted high-risk with 16.4 true MBE cases. Figure 8 displays the CICs. Overall, the model achieved strong discrimination, acceptable calibration, robust internal validation, and substantial net benefit across clinically relevant thresholds.

Figure 6 Calibration plot showing the agreement between predicted and observed probabilities for MBE. MBE, malignant brain edema.

Figure 7 DCA plot showing the clinical utility of the prediction model for MBE. DCA, decision curve analysis; MBE, malignant brain edema.

Figure 8 CIC illustrating the clinical utility of the prediction model for MBE after reperfusion therapy. CIC, clinical impact curve; MBE, malignant brain edema.

MBE occurrence and clinical outcomes

Follow-up (Figure 9) confirmed that MBE was associated with markedly worse 90-day functional outcomes. Most MBE patients had mRS ≥4 at 90 days, indicating severe disability; the combined proportion of death (mRS =6) or severe disability (mRS =5) exceeded 20%, and fewer than 10% achieved independence (mRS 0–2). In contrast, non-MBE patients fared substantially better: most had mRS ≤3, and more than 50% achieved independence (mRS 0–2). The difference in poor outcomes (mRS 3–6) between groups was highly significant (P<0.001). These findings underscore the importance of early MBE risk identification and targeted intervention to improve overall prognosis in AIS.

Figure 9 Distribution of mRS scores at 90 days between MBE and non-MBE groups. MBE, malignant brain edema; mRS, modified Rankin Scale; NMBE, non-MBE.

Discussion

In this study, we reaffirmed that the development of MBE significantly worsens early outcomes in AIS patients, markedly increasing the risk of early death and severe disability. Therefore, early identification and targeted intervention in high-risk patients based on predictive modeling are critical to improving the overall prognosis in AIS. Furthermore, we developed and internally validated a nomogram integrating multiple clinical, imaging, and laboratory variables to predict MBE after reperfusion therapy in patients with AIS. Based on the finalized model, the nomogram integrates seven independent predictors: rCBF <40% volume, DT >6.0 s volume, IGR, CCS, N, LDH, and DBP. In internal validation it showed strong discrimination and good calibration, supporting early MBE risk stratification. Our findings highlight the key role of ischemic burden and collateral status in the pathophysiology of MBE. IGR reflects the speed of infarct expansion, with higher rates indicating poor collateralization and more rapid ischemic progression, consistent with prior studies suggesting that faster infarct growth correlates with worse outcomes (11-13). Poor CCS indicates insufficient collateral compensation, which accelerates infarct expansion and edema formation. DT >6.0 s hypoperfusion volume quantifies the extent of severely hypoperfused tissue, providing a “substrate” for cerebral edema development. In addition to imaging parameters, several blood biomarkers were significantly associated with MBE. Elevated N likely reflects the intensity of post-stroke inflammatory cascades; massive neutrophil infiltration into ischemic tissue exacerbates blood-brain barrier disruption and edema formation (14-16). Higher LDH levels suggest greater tissue ischemia and necrosis, with more extensive cellular enzyme release, indicating more severe brain injury and edema risk (17,18).

Collectively, prior studies have underlined key determinants of malignant edema while revealing gaps that our model addresses. Simpler clinical scores (e.g., Jo et al.’s MBE score based on NIHSS, ASPECTS, collaterals, and revascularization failure) demonstrated that baseline stroke severity and collateral status are critical (19-22). Likewise, the Western EDEMA score and its Chinese modifications confirmed that adding clinical severity (NIHSS) to imaging signs modestly improves prediction (AUC ~0.7–0.8) (23,24). However, these earlier tools omitted advanced perfusion imaging and biochemical markers. Recent Chinese cohorts have moved toward multimodal integration: He et al. and Wang et al. incorporated CTA collaterals and basic lab indices and achieved higher accuracy (25,26). In parallel, Zhang et al. developed a multi-region NCCT radiomics model that accurately predicts malignant cerebral edema (27), highlighting the added value of quantitative imaging features. Notably, among recent models, only Wang et al. incorporated a perfusion metric (core infarct volume), whereas others relied on proxy measures such as ASPECTS or 24-h CT changes. By contrast, the prospective multicentre INTEP-AR model (28) integrated large infarct, NIHSS, intravenous thrombolysis, endovascular therapy, pneumonia, brain atrophy, and recanalization—highlighting the protective effect of recanalization but not using perfusion imaging. Building on these strands, we unify three domains—clinical variables, quantitative CT perfusion, and laboratory biomarkers—into a single pretreatment nomogram. Specifically, we combined quantitative CTP burden (rCBF <40% volume and DT >6.0 s volume) with a dynamic marker of infarct expansion (IGR), systemic signals (N and LDH), a hemodynamic reserve measure (CCS), and DBP. This yielded excellent discrimination (optimism-corrected AUC ~0.85), on par with or exceeding prior models. Crucially, our nomogram focuses on early prediction before overt edema or extensive 24-hour changes occur, unlike Jiang et al.’s post-thrombectomy score (29) which required post-procedure NIHSS and CT findings. This enables identification of high-risk patients immediately after reperfusion therapy and supports prompt intervention. Furthermore, our use of perfusion-derived infarct growth rate directly captures dynamic ischemic progression that earlier scores could only approximate indirectly, and we are the first to include LDH, informed by evidence that metabolic and cellular injury markers correlate with edema risk. Operationally, after baseline NCCT, CTA, CTP, and the first laboratory panel, the workstation computes DT and rCBF volumes and the IGR using predefined parameters, readers grade CCS with a standardized rubric, and the electronic health record (EHR) returns a probability and risk tier; predefined thresholds map to three pathways—escalation with early neurosurgical evaluation for high risk, timed reassessment for intermediate risk, and routine care for low risk—thereby translating probability into concrete actions on monitoring intensity, imaging cadence, and early decompressive strategies, which is especially valuable in resource-constrained settings to improve allocation and reduce overuse. Notably, we did not include standardized recanalization metrics such as modified thrombolysis in cerebral infarction (mTICI) (30,31); the retrospective single-center design and modest sample size limit generalizability and leave residual overfitting risk; CCS is reader-dependent and inter-rater agreement was not assessed; CTP processing relies on a single platform and may be affected by software settings, delay correction, vessel masking, and input-function selection; the cohort was limited to anterior-circulation LVO patients undergoing reperfusion, so applicability to patients without thrombectomy or with posterior-circulation stroke remains uncertain; and key variables such as recanalization status, stroke etiology, pre-stroke disability, and early infection were not modeled and may confound associations for neutrophils or LDH. Together, these issues define priorities for multicenter prospective validation, workflow standardization, and model refinement.

Conclusions

We developed a nomogram that integrates clinical, imaging, and laboratory variables to predict the risk of malignant MBE after reperfusion therapy in AIS. Internal validation demonstrated strong predictive accuracy and practical clinical applicability. The nomogram may assist clinicians in the early identification of patients at high risk for MBE and in implementing targeted interventions to reduce stroke-related mortality and disability. Before clinical adoption, prospective external validation in independent, multicenter cohorts is required to confirm reliability and generalizability.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1153/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1153/dss

Funding: This work was supported by the Discipline Construction of Health System in Pudong New Area, Shanghai (No. PWzbr2022-14). The article processing charge and manuscript services were covered by this grant. The funder had no role in study design, data collection, data analysis, or the decision to publish.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1153/coif). All authors report that this study was supported by the Discipline Construction of Health System in Pudong New Area, Shanghai (No. PWzbr2022-14). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. It was approved by the Ethics Committee of Shanghai East Hospital (No. 2024YS-279), and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Feigin VL, Norrving B, Mensah GA. Global Burden of Stroke. Circ Res 2017;120:439-48. [Crossref] [PubMed]
2. Nogueira RG, Jadhav AP, Haussen DC, Bonafe A, Budzik RF, Bhuva P, et al. Thrombectomy 6 to 24 Hours after Stroke with a Mismatch between Deficit and Infarct. N Engl J Med 2018;378:11-21. [Crossref] [PubMed]
3. Pu M, Chen J, Chen Z, Li Z, Li Z, Tang Y, Li Q. Predictors and outcome of malignant cerebral edema after successful reperfusion in anterior circulation stroke. J Stroke Cerebrovasc Dis 2023;32:107139. [Crossref] [PubMed]
4. Wang C, Zhu Q, Cui T, Wang L, Yang T, Hao Z, Wu S, Zheng H, Hu F, Wu B. Early Prediction of Malignant Edema After Successful Recanalization in Patients with Acute Ischemic Stroke. Neurocrit Care 2022;36:822-30. [Crossref] [PubMed]
5. Kasner SE, Demchuk AM, Berrouschot J, Schmutzhard E, Harms L, Verro P, Chalela JA, Abbur R, McGrade H, Christou I, Krieger DW. Predictors of fatal brain edema in massive hemispheric ischemic stroke. Stroke 2001;32:2117-23. [Crossref] [PubMed]
6. Chen Z, Zhang X, Liu C. Outcomes of therapeutic hypothermia in patients treated with decompressive craniectomy for malignant Middle cerebral artery infarction: A systematic review and meta-analysis. Clin Neurol Neurosurg 2020;188:105569. [Crossref] [PubMed]
7. Wu S, Yuan R, Wang Y, Wei C, Zhang S, Yang X, Wu B, Liu M. Early Prediction of Malignant Brain Edema After Ischemic Stroke. Stroke 2018;49:2918-27. [Crossref] [PubMed]
8. Pacinella G, Ciaccio AM, Tuttolomondo A. Molecular Links and Clinical Effects of Inflammation and Metabolic Background on Ischemic Stroke: An Update Review. J Clin Med 2024;13:7515. [Crossref] [PubMed]
9. Peng G, Lu W, Chen K, Yang X, Wei J, Wang B, Huang B. Study on collateral circulation level and prognosis of acute ischemic stroke by 4D CTA-CTP integrated technology and serum S100B. Microvasc Res 2022;140:104270. [Crossref] [PubMed]
10. Pensato U, Demchuk AM, Menon BK, Nguyen TN, Broocks G, Campbell BCV, Gutierrez Vasquez DA, Mitchell PJ, Hill MD, Goyal M, Ospel JM. Cerebral Infarct Growth: Pathophysiology, Pragmatic Assessment, and Clinical Implications. Stroke 2025;56:219-29. [Crossref] [PubMed]
11. Wang Y, Binkley MM, Qiao M, Pardon A, Keyrouz S, Dhar R, Ford AL. Rate of Infarct-Edema Growth on CT Predicts Need for Surgical Intervention and Clinical Outcome in Patients with Cerebellar Infarction. Neurocrit Care 2022;36:1011-21. [Crossref] [PubMed]
12. Regenhardt RW, Singhal AB, He J, Gonzalez RG, Lev MH. Percent Insular Ribbon Infarction for Predicting Infarct Growth Rate and 90-Day Outcomes in Large-Vessel Occlusive Stroke: Secondary Analysis of Prospective Clinical Trial Data. AJR Am J Roentgenol 2023;221:103-13. [Crossref] [PubMed]
13. Sarraj A, Hassan AE, Grotta J, Blackburn S, Day A, Abraham M, et al. Early Infarct Growth Rate Correlation With Endovascular Thrombectomy Clinical Outcomes: Analysis From the SELECT Study. Stroke 2021;52:57-69. [Crossref] [PubMed]
14. Li Z, Sun S, Xiao Q, Tan S, Jin H, Hu B. Neuron Derived Cold-Inducible RNA-Binding Protein Promotes NETs Formation to Exacerbate Brain Endothelial Barrier Disruption after Ischemic Stroke. Aging Dis 2024; Epub ahead of print. [Crossref]
15. Zheng S, Jiang X, Guo S, Shi Z. Postoperative neutrophil-to-lymphocyte ratio predicts malignant cerebral edema following endovascular treatment for acute ischemic stroke. Front Neurol 2024;15:1394550. [Crossref] [PubMed]
16. Liu C, Li F, Liu S, Chen Q, Sang H, Yang Q, Zhou K, Zi W. Neutrophil Count Predicts Malignant Cerebellar Edema and Poor Outcome in Acute Basilar Artery Occlusion Receiving Endovascular Treatment: A Nationwide Registry-Based Study. Front Immunol 2022;13:835915. [Crossref] [PubMed]
17. Jin H, Bi R, Hu J, Xu D, Su Y, Huang M, Peng Q, Li Z, Chen S, Hu B. Elevated Serum Lactate Dehydrogenase Predicts Unfavorable Outcomes After rt-PA Thrombolysis in Ischemic Stroke Patients. Front Neurol 2022;13:816216. [Crossref] [PubMed]
18. Liu Y, Yang P, Liu HC, Sun S, Zhang JL, Kang J. The Significance of The Detection of Serum Lactate Dehydrogenase, Hypersensitive C-Reactive Protein, and N-Terminal Pro-Brain Natriuretic Peptide for the Evaluation of the Severity and Progression of Pediatric Patients with Traumatic Brain Injury. Curr Neurovasc Res 2022;19:219-24. [Crossref] [PubMed]
19. Jo K, Bajgur SS, Kim H, Choi HA, Huh PW, Lee K. A simple prediction score system for malignant brain edema progression in large hemispheric infarction. PLoS One 2017;12:e0171425. [Crossref] [PubMed]
20. Tang A, Ma X, Ren Y, Zhang H, Xie W, Liu M, Sheng S, Mao K. External validation and comparison of MBE, EDEMA, and modified EDEMA scores for predicting malignant cerebral EDEMA in Chinese patients with large hemispheric infarction patients without revascularization. J Clin Neurosci 2024;122:66-72. [Crossref] [PubMed]
21. Miao J, Song X, Sun W, Qiu X, Lan Y, Zhu Z. Predictors of malignant cerebral edema in cerebral artery infarction: A meta-analysis. J Neurol Sci 2020;409:116607. [Crossref] [PubMed]
22. Sun W, Li G, Song Y, Zhu Z, Yang Z, Chen Y, Miao J, Song X, Lan Y, Qiu X, Zhu S, Fan Y. A web based dynamic MANA Nomogram for predicting the malignant cerebral edema in patients with large hemispheric infarction. BMC Neurol 2020;20:360. [Crossref] [PubMed]
23. Cheng Y, Wu S, Wang Y, Song Q, Yuan R, Wu Q, Zhang S, Zhang S, Wu B, Liu M. External Validation and Modification of the EDEMA Score for Predicting Malignant Brain Edema After Acute Ischemic Stroke. Neurocrit Care 2020;32:104-12. [Crossref] [PubMed]
24. Ong CJ, Gluckstein J, Laurido-Soto O, Yan Y, Dhar R, Lee JM. Enhanced Detection of Edema in Malignant Anterior Circulation Stroke (EDEMA) Score: A Risk Prediction Tool. Stroke 2017;48:1969-72. [Crossref] [PubMed]
25. He Y, Zuo M, Huang J, Jiang Y, Zhou L, Li G, Chen L, Liu Q, Liang D, Wang Y, Wang L, Zhou Z. A Novel Nomogram for Predicting Malignant Cerebral Edema After Endovascular Thrombectomy in Acute Ischemic Stroke: A Retrospective Cohort Study. World Neurosurg 2023;173:e548-58. [Crossref] [PubMed]
26. Wang H, Xu C, Xiao Y, Shen P, Guo S, Shang Y, Chen X, Xu J, Li C, Tan M, Zhang S, Geng Y. A dynamic nomogram for predict individual risk of malignant brain edema after endovascular thrombectomy in acute ischemic stroke. Sci Rep 2024;14:9529. [Crossref] [PubMed]
27. Zhang L, Zhang Y, Yang C, Zhang Y, Xie G, Wang D, Li K. Predicting malignant cerebral edema after acute ischemic stroke: a machine-learning model with multi-region radiomics. Quant Imaging Med Surg 2025;15:5188-203. [Crossref] [PubMed]
28. Wu S, Wang Y, Yuan R, Guo F, Yang D, Li Z, Wu B, Wang C, Duan J, Ling T, Zhang H, Zhang S, Wu B, Anderson CS, Liu M. Predicting the emergence of malignant brain oedema in acute ischaemic stroke: a prospective multicentre study with development and validation of predictive modelling. EClinicalMedicine 2023;59:101977. [Crossref] [PubMed]
29. Jiang QM, Yu S, Dong XF, Wang HS, Hou J, Huang ZC, Guo ZL, You SJ, Xiao GD. Predictors and Dynamic Nomogram to Determine the Individual Risk of Malignant Brain Edema After Endovascular Thrombectomy in Acute Ischemic Stroke. J Clin Neurol 2022;18:298-307. [Crossref] [PubMed]
30. von Braun MS, Starke K, Peter L, Kürsten D, Welle F, Schneider HR, Wawrzyniak M, Kaiser DPO, Prasse G, Richter C, Kellner E, Reisert M, Klingbeil J, Stockert A, Hoffmann KT, Scheuermann G, Gillmann C, Saur D. Prediction of tissue and clinical thrombectomy outcome in acute ischaemic stroke using deep learning. Brain 2025;148:2348-60. [Crossref] [PubMed]
31. Zaidat OO, Yoo AJ, Khatri P, Tomsick TA, von Kummer R, Saver JL, et al. Recommendations on angiographic revascularization grading standards for acute ischemic stroke: a consensus statement. Stroke 2013;44:2650-63. [Crossref] [PubMed]