A Novel Nomogram for Predicting Malignant Cerebral Edema After Endovascular Thrombectomy in Acute Ischemic Stroke: A Retrospective Cohort Study

Background Malignant cerebral edema (MCE) is a common and feared complication after endovascular thrombectomy (EVT) in acute ischemic stroke (AIS). This study aimed to establish a nomogram to predict MCE in anterior circulation large vessel occlusion stroke (LVOS) patients receiving EVT. Methods In this retrospective cohort study, 381 patients with anterior circulation LVOS receiving EVT were screened from 636 hospitalized patients with LVOS in two stroke medical centers. Clinical baseline data and imaging data were collected within 2–5 days of admission to the hospital. The patients were divided into two groups based on whether MCE occurred after EVT. Multivariate logistic regression analysis was used to evaluate the independent risk factors for MCE and to establish a nomogram


Introduction
Acute ischemic stroke (AIS) is a complicated cerebrovascular disease leading to high mortality and severe long-term disability.Endovascular thrombectomy (EVT) has evolved rapidly in recent years and has gradually become an effective treatment for AIS due to large vessel occlusion stroke (LVOS) [1][2][3][4].
The latest randomized controlled trials have con rmed the e cacy and safety of EVT for the therapy of patients with stroke occlusion of large vessels in the anterior vasculature within 24 hours [1,2,5,6].
Malignant cerebral edema (MCE) is a common and serious complication of acute large-scale cerebral infarction that can lead to cerebral herniation within few days after its onset [7][8][9].These cases occur in approximately 10%-78% of middle cerebral artery infarction [10], and the mortality rate with conservative treatment is as high as 80% [11,12].Although some recent studies have shown that reperfusion therapies could signi cantly decrease the risk of MCE for patients with acute large-scale cerebral infarction to prevent later poor outcome [13,14].However, previous studies have shown that even after implementing EVT or other available reperfusion therapies, more than 10-40% of patients still have MCE [15][16][17].Hence, accurate prediction of MCE after AIS undergoing EVT is essential to prevent poor prognosis.It is well known that several previous predictive instruments for MCE, such as TURN score [18], EDEMA score [19], and modi ed EDEMA score [20], were not speci cally designed for evaluating acute anterior circulation LVOS patients undergoing EVT.At present, there is a lack of fully research evidence related to predictors of MCE in patients with AIS after EVT.Therefore, novel risk models that enable prediction are needed.
The nomogram is an easy-to-use and graphical predictive tool that converts risk factors into a continuous scoring system capable of calculating precise risk probabilities for speci c events in individual patients.In the last few years, some studies have widely constructed nomogram models to predict the complications of AIS [21][22][23].However, the nomogram for MCE in AIS patients with EVT has not been well built until now.In this study, we aimed to develop a novel nomogram for early predicting MCE in AIS patients after EVT.

Inclusion and exclusion criteria of patients:
This study retrospectively analyzed the clinical and imaging ndings of patients with acute anterior circulation LVOS who were hospitalized and received EVT in two comprehensive stroke centers (Southwest Hospital between January 2017 and October 2021, The Third People's Hospital of Zigong City between January 2017 and October 2021).It was approved by the ethics committee of two hospitals [(B)KY2021146 and (2021-01-01)], and all patients or their surrogates signed the informed consent form for the EVT.
All patients who met the following inclusion criteria were included: (1) age > 18; (2) de nite diagnosis of AIS; (3) computed tomography angiography (CTA), magnetic resonance angiography (MRA) or intraoperative digital subtraction angiography (DSA) to determine the presence of large vessel occlusion in the anterior circulation; (4) time from onset to femoral artery puncture within 6 hours, (5) patients who were treated within 6 hours of symptom onset or who had a mismatch between the severity of the clinical de cit and the infarct volume established by MR within 6 to 24 hours.

Acute Ischemic Stroke Management And Endovascular Procedure:
After receiving computed tomography (CT) or CTA in the emergency department, it was clear that there was no obvious bleeding focus.Patients within 4.5 hours were treated with intravenous recombinant tissue plasminogen activator (rt-PA) and endovascular thrombectomy (EVT).Patients who had onset within 6 hours or 6-24 hours with mismatch of clinical symptoms and imaging had EVT directly applied [1,2].The EVT device was a Solitaire AB stent-like embolization device.The anesthesia was general anesthesia or local anesthesia, and the choice of the method was determined by the interventional doctor.
Under anesthesia, the femoral artery was punctured, the catheter sheath was introduced, 6F or 8F guide catheter were selected, and the microcatheter and the microguide wire were sent into the guide catheter; rst, the micro guide wire was used in cooperation with the contrast catheter to pass through the occluded section and the microcatheter.The tip was sent to the distal end of the occlusion, and the microguide wire was withdrawn.The embolization stent was pushed through the microcatheter to the occlusion site, the microcatheter was slowly withdrawn, the embolization stent was released; then, 3 to 5 minutes were allowed to pass for the embolization stent to fully expand, the blood ow of the blood vessel was immediately checked, and then the embolus was retracted, thereby, completing the thrombectomy process.If a thrombectomy was not successful, thrombectomy was repeated.If more than 4 thrombus removals were unsuccessful, salvage treatment was used, such as balloon dilation, stent If the infarction in the territory of the MCA was more than 50% of the territory of the MCA and also an acute onset of corresponding clinical signs and symptoms, this kind of infarction was de ned as large cerebral infarction in the hemisphere (LHI) [27].
The De nition And Evaluation Of Malignant Cerebral Edema: Based on the Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS-MOST) protocol [28], this study divided cerebral edema into three levels: grade 0 was assigned if there was no swelling, grade 1 was assigned if the cerebral swelling compromised less than one-third of the hemisphere, grade 2 was assigned if the swelling comprised more than one-third of the hemisphere, and grade 3 was assigned if the midline shift occurred.This study de ned midline shift > 5 mm on CT as malignant cerebral edema (MCE) based on grade 3 [14,17].All patients were evaluated using a CT procedure.

Statistical analysis:
The SPSS (V.22; SPSS Inc., Chicago, IL, USA) statistical software package was used to statistically analyze the baseline data.Measurement data with normal distributions are presented as the mean ± standard deviation, the data with nonnormal distributions are presented as the median (interquartile range, IQR), and the qualitative data are presented as the number of patients (%).Patients were divided into MCE groups and non-MCE groups based on whether MCE occurred within 2-5 days after EVT.To compare the data in the MCE group and non-MCE groups, Student's t test (normal distribution) or Mann-Whitney t test (nonnormal distribution) was used for continuous variables, and the chi-square test or Fisher's exact test was used for grade data.Univariate and multivariate logistic regression analyses were established for MCE to evaluate independent predictors (the entry and removal limits were set to 0.05 and 0.1, respectively, and the variables with P < 0.1 in the univariate analysis were included).According to the multivariate logistic regression outcomes and clinical signi cance, a nomogram model for the risk prediction of MCE occurrence in AIS patients after EVT was constructed using the R software (R 4.1.2;http://www.Rproject.org)rms program package.And calibration plot along with Hosmer-Lemeshow tests were performed to assess the goodness-of-t of the nomogram.Internal validation of the nomogram model was performed by repeated sampling 1000 times using the Bootstrap method.The receiver operating characteristic (ROC) curves were used to calculate the area under the curve (AUC), and to evaluate the discrimination abilities of this nomogram.In order to quantify the net bene t of patients under different threshold probabilities, the decision curve analysis (DCA) was used to determine the clinical application value of nomogram.The rmda package was used to construct the DCA.In the bilateral test, the difference was statistically signi cant (P < 0.05).

Results
A total of 636 patients with AIS undergoing EVT were initially enrolled.After excluding of 176 patients with posterior circulation infarction, 54 patients with sICH in acute anterior circulation stroke, 2 patients with a life expectancy of less than 3 months, and 23 patients with missing data, 381 patients were included for the nal analysis in this study.
The baseline characteristics and clinical outcome of the patients are presented in Table 1.The average age was 69.6 ± 12.64 years (mean ± SD), 203 patients (53.3%) were males, and 20 patients with decompressive hemicraniectomy (DHC).There were 66 patients in the MCE group and 315 patients in the non-MCE group.The admission NIHSS scores ≥ 16 in the MCE group was signi cantly higher than that in the non-MCE group (72.7% vs. 59.0%,P = 0.038).And patients with MCE had lower baseline ASPECT score (median 6 [4,8] vs. 8 [7,9]; P < 0.001).The higher status of collateral circulation (ASITN/SIR > 2) was rare in the MCE group (25.6% vs. 68.9%;P < 0.001).The internal carotid artery occlusion was more prevalent in patients with MCE than in patients without MCE (28.8% vs. 16.8%;P = 0.024).And successful recanalization (mTICI 2b/3) was more common in patients without MCE than in patients with MCE (93.3% vs. 75.8%;P < 0.001).The comparison of modi ed Rankin scale (mRS) scores between the two groups after 3 months showed that the incidence of a good outcome (mRS < 2) in the non-MCE group was 63.5%, which was much higher than that in the MCE group (24.2%) (Fig. 2), and the difference in incidence of good outcomes between the two groups was statistically signi cant (P < 0.001) (as shown in Table 1).Furthermore, patients with asymptomatic intracerebral hemorrhage (aICH) (65.2% vs 28.9%, P < 0.001), deep middle cerebral artery territory (93.9% vs 45.1%, P < 0.001) and large cerebral infarction in the hemisphere (LHI) (90.9% vs 25.7%, P < 0.001) were more likely to develop into MCE (as shown in Table 3).2).
Next, the nomogram based on the multivariate logistic regression was formed using the seven signi cant risk factors for predicting MCE (Figure .3).A total score was calculated with the use of admission NIHSS scores ≥ 16, ASPECT score, right hemisphere, collateral circulation (ASITN/SIR > 2), mTICI 2b/3, Hct and serum glucose.The value of each of these variables was given a corresponding score on the point scale axis.Depending on the basic condition of each patient with AIS at the time of admission, the total score could be easily calculated by adding each corresponding score, and by projecting the total score to a lower total score scale, we were able to estimate the probability of developing an MCE in AIS patient after EVT.According to the results of the ROC analysis (Figure .4), the nomogram showed a good discrimination, with an area under the ROC curve of 0.901 (95% CI:0.848-0.940;P < 0.001).As shown in Fig. 5 was the calibration curve of the nomogram model, which shows that the MCE probabilities predicted by the nomogram agreed well with the actual probabilities.And the Hosmer-Lemeshow test results were not signi cant (P = 0.685), demonstrating no deviation from a perfect t.
Furthermore, we also constructed a decision curve to guide clinical application of the nomogram (Figure .6).The decision curve indicated that if the threshold probabilities were > 1% and < 71%, the nomogram could be used to predict the risk of MCE in AIS patients after EVT.DCA indicating that the nomogram to predict MCE produced higher net bene ts than the strategy of assuring that all AIS patients after EVT will occur MCE and the strategy of assuring that no AIS patients after EVT will occur MCE.In summary, these results veri ed the high predictive ability of this nomogram.

Discussion
In the current study, we created a simple intuitive graph of a nomogram model that quanti ed the risk of development of MCE in anterior circulation large vessel occlusion stroke patients after EVT.This nomogram demonstrated that admission NIHSS scores ≥ 16, ASPECT score, right hemisphere, collateral circulation (ASITN/SIR > 2), successful recanalization (mTICI 2b/3), hematocrit and serum glucose were an independent predictor of early MCE in AIS patients after EVT.
The comparison of the severity based on NIHSS scores showed that the MCE group was in a more severe condition than the non-MCE group.Higher admission NIHSS scores ≥ 16 represent severe neurological de cits in patients with AIS that is related to a large infarct size and/or poor collateral circulation, indicating severe ischemic necrosis of local cerebral tissue and these patients are more likely to develop MCE.The ASPECT scores of the MCE group were lower than that of the non-MCE group.This nding is in agreement with a retrospective study that showed the relationship between early CT signs of infarction (lower ASPECTS < 7) and MCE occurrence [29].We can comprehensively and systematically identify the extent of dead brain tissue and cytotoxic edema on CT for AIS patients by ASPECT score tool.This study once again con rmed that lower ASPECT scores on admission in AIS patients can be closely related to the occurrence of MCE.
In this study, we found that poor collateral circulation scores (ASITN/SIR,0-2) were more common in MCE group.To the best of our current knowledge, collateral circulation is an important determinant of clinical outcome in AIS patients.Previous several studies have demonstrated that good collateral circulation scores (ASITN/SIR > 2) were associated with increased volume of salvageable tissue, smaller infarct volumes and better clinical outcome in AIS patients successfully recanalized with EVT [30,31].On the contrary, poor collateral circulation scores (ASITN/SIR,0-2) were related to larger nal infarct volumes and worse functional outcomes, which were more likely to increase the degree of cerebral edema after cerebral infarction [32].
Furthermore, our results showed that successful recanalization (mTICI 2b/3) could reduce the occurrence of MCE after EVT.These ndings are consistent with the previous studies reporting that successful recanalization was associated to a reduced midline shift [33].A recent registry study revealed that patients with ischemic stroke have a lower risk for early cerebral edema after successful recanalization [34].Thus, successful recanalization may give a more favorable clinical outcome in patients with AIS and reduce the risk of developing postprocedural MCE.
In patients with MCE, levels of admission blood glucose were higher than those in patients without MCE.
Our ndings were consistent with those studies describing the impact of higher blood glucose on cerebral edema in AIS patients [35].On the one hand, the elevated admission blood glucose level was related to increased early cerebral edema and poor clinical outcome mediated by collateral status [30].Previous studies have documented that higher blood glucose might reduce the establishment of new collateral channels after acute arterial occlusion [36].On the other hand, elevated blood glucose could enhance the expression of matrix metalloproteinase-9 (MMP-9) [37], which directly destroys the blood-brain barrier (BBB) and causes the occurrence of MCE [38,39].
The results of this study showed that among AIS patients lower serum hematocrit levels were associated with an increased risk of MCE after EVT.A previous observational study suggested that the low and further decreasing hematocrit level could be an indicator to predict poor outcome in AIS patients [40].What's more, another study further con rmed that lower hematocrit levels might be associated with infarct growth in AIS patients [41].In this study, we found MCE group with lower serum hematocrit levels, which re ecting cerebral oxygen transport failure after AIS [40].Low oxygen levels may lead to larger cerebral infarct cores [42].And larger cerebral infarct volume could aggravate the development of cerebral edema [43].Therefore, in this study, hematocrit might simply be an indirect predictor of the occurrence of MCE in AIS patients after EVT.However, the speci c mechanisms underlying the relationship between lower serum hematocrit levels in AIS patients and the development of MCE in AIS patients after EVT were not fully understood.
Despite no statistical signi cance in univariate analyses, we found an association between right hemisphere and MCE in multivariate analyses.In this study, a higher incidence rate of MCE was observed in patients with right hemisphere involvement, compared to patients with AIS in the left hemisphere.Our ndings were consistent with those studies describing that malignant middle cerebral artery infarctions (MMCA) were relatively common in the right hemisphere than the left hemisphere in AIS patients [44].We speculated that the possible reason was that the right hemisphere plays a key role in cardiovascular regulation [45].Therefore, patients with AIS in the right hemisphere were more likely to cause norepinephrine secretion and sympathetic activation in the autonomic nervous system compared to the left hemisphere.The released noradrenaline could further increase the BBB permeability, leading to abnormal aggregation of water molecules in the intercellular uid and induce cytotoxic edema [46].In this study, right hemisphere might simply be a risk factor of MCE, and the speci c mechanism remains unclear.
Although    The ROC curve analysis of the predictive value of the nomogram for the risk of malignant cerebral edema

Table 2
Multivariate Regression Analysis for malignant cerebral edema in the Derivation Cohort Admission NIHSS scores ≥ 16, baseline ASPECT score, vascular occlusion site, successful recanalization (mTICI 2b/3), good collateral circulation (ASITN/SIR > 2), serum NLR levels, serum LMR levels, Hct and serum glucose could be used as independent risk factors or biomarkers for MCE in AIS patients, and the variables with p < 0.1 in the univariate analysis were subsequently examined by binary logistic regression analysis.On the basis of univariate and multivariate logistic regression analyses, the variables with statistical signi cance between the non-MCE and MCE groups were screened, and results suggested that Admission NIHSS scores ≥ 16, ASPECT score, right hemisphere, successful recanalization (mTICI 2b/3), good collateral circulation (ASITN/SIR > 2), Hct and serum glucose were independent predictors of MCE in AIS patients (all P < 0.05).The results indicated signi cant differences in Admission NIHSS scores ≥ 16 (adjusted odds ratio (OR), 5.342; 95% con dence interval (CI): 1.
occurrence of MCE.Moreover, this study constructed a nomogram to further evaluate the predictive value of particular variables for MCE during hospitalization in AIS patients after EVT.Importantly, almost all patients with AIS were routinely tested for whole blood cells on admission, which are inexpensive tests that can provide results within approximately an hour.Therefore, hematocrit and glucose were easily obtained in routine laboratory examinations.It may be easier and more comfortable to use these biomarkers to predict the occurrence of MCE in patients with AIS than other biomarkers.In addition, it is common to assess clinical and radiographic characteristics of admitted patients with AIS, including ASPECT scores, collateral circulation and infarcted hemisphere, which were also taken into account in the clinical prediction nomogram.Therefore, this nomogram had great predictive value for the occurrence of MCE in AIS patients after EVT.This study also had some limitations.At rst, the limited sample size and the ease of operation of the model were taken into account in the process of model construction, and more variables were not included.Secondly, this study validated results only by internal validation and not in a large external population; for this reason, which may cause the overestimated performance of the nomogram.Thirdly, because this real-world study was not a blind study, which could have affected the reliability of the research results.Finally, this study was only two stroke centers retrospective study.In the future, more prospective randomized controlled studies are needed to further study the prognostic value of this nomogram and nding potential marker to predict MCE.In summary, this study showed that MCE during hospitalization in AIS patients after EVT treatment can signi cantly increase the risk of death within 3 months.The ASPECT scores, right hemisphere, good collateral circulation (ASITN/SIR > 2), hematocrit and serum glucose are independent risk factors for MCE in AIS patients.The nomogram constructed based on the above variables has a certain predictive value for the risk of MCE.Nonstandard Abbreviations And AcronymsAIS= acute stroke; EVT= endovascular thrombectomy; LVOS= large vessel occlusion stroke; MCE= malignant cerebral edema; CTA= computed tomography angiography; MRA= magnetic resonance angiography; DSA= digital subtraction angiography; sICH= symptomatic intracranial haemorrhage; rt-PA= recombinant tissue plasminogen activator; CT= computed tomography; NIHSS=National Institutes of Health Stroke Scale; ASPECT=Alberta Stroke Program Early CT; mTICI= modi ed Thrombolysis in Cerebral Infarction; ASITN/SIR= The American Society of Interventional and Therapeutic Neuroradiology/Society of Interventional Radiology; WBC= white blood cell; NLR= neutrophil-tolymphocyte ratio; LMR, lymphocyte-to-monocyte ratio; PLT= platelets; Hct= hematocrit; Fib= brinogen; aICH, deep middle cerebral artery territory and LHI were signi cant related to MCE, these imaging indicators weren't evaluated within 24 hours after EVT.So those factors were not suitable for early predictive MCE.In this study, we found that the clinical, laboratory, and imaging features were highly indicated the LDL= low density lipoprotein; Glu= glucose; mRS= modi ed Rankin Scale score; OR=odds ratio; ROC= receiver operating characteristic; AUC= area under the curve; DCA= decision curve analysis Declarations