Abstract
Background:
Recent randomized trials have shown the benefit of mechanical thrombectomy (MT) also in patients with an established large ischemic core.
Aims:
The purpose of this study was to define baseline predictors of clinical outcome in patients with large vessel occlusion (LVO) in the anterior circulation and an Alberta Stroke Program Early CT score (ASPECTS) ⩽ 5, undergoing MT.
Material and methods:
The databases of 16 comprehensive stroke centers were retrospectively screened for patients with LVO and ASPECTS ⩽5 that received MT. Baseline clinical and neuroradiological features, including the differential contribution of all ASPECTS regions to the composite score, were collected. Primary clinical outcome measure was a 90-day modified Rankin Scale (mRS) score of 0–2. Statistical analysis used a logistic regression model and random forest algorithm.
Results:
A total of 408 patients were available for analysis. In multivariate model, among baseline features, lower age (odd ratio (OR) = 0.962, 95% confidence interval (CI) = 0.943–0.982) and lower National Institute of Health Stroke Scale (NIHSS) score (OR = 0.911, 95% CI = 0.862–0.963) were associated with the mRS score 0–2. Involvement of the M2 (OR = 0.398, 95% CI = 0.206–0.770) or M4 (OR = 0.496, 95% CI = 0.260–0.945) ASPECTS regions was associated with an unfavorable outcome. Random forest analysis confirmed that age and baseline NIHSS score are the most important variables influencing clinical outcome, whereas involvement of cortical regions M5, M4, M2, and M1 can have a negative impact.
Conclusion:
Our retrospective analysis shows that, along with age and baseline clinical impairment, presence of early ischemic changes involving cortical areas has a role in clinical outcome in patients with large ischemic core undergoing MT.
Data access statement:
The data that support the findings of this study are available upon reasonable request.
Keywords: Large ischemic core, ASPECTS, large vessel occlusion, mechanical thrombectomy, clinical outcome
Introduction
Mechanical thrombectomy (MT) is the standard of care for large vessel occlusion (LVO) in the anterior circulation but is not yet recommended in patients with a baseline Alberta Stroke Program Early CT score (ASPECTS) ⩽5.1,2 The ASPECTS is a 10-point scoring system used to quantify the extent of early ischemic changes in the middle cerebral artery (MCA) territory on non-contrasted computed tomography (CT) scan, with lower scores indicating a larger infarct core. 3 Although other studies have challenged its accuracy in defining the extension of early ischemic changes compared with other imaging methods, ASPECTS has been widely used to define boundaries of infarct size for enrollment in MT trials.1,4–6 Therefore, the limited number of patients with a baseline ASPECTS ⩽5 included in the main MT trials was insufficient to draw conclusions about the benefit of endovascular treatment in this category of patients.1,7–9 More recently, dedicated randomized trials have shown that MT is beneficial also in patients with baseline ASPECTS ⩽5.10–13
Aims
In current real-world practice, MT is considered for patients with a large ischemic core on a case-by-case basis and after careful evaluation of clinical and radiological features. In a scenario where decision of treatment may be encouraged by the evidence of recent trials, it is helpful to define baseline characteristics that can be of prognostic value. Here, we report a large retrospective multicenter analysis on consecutive patients with LVO in the anterior circulation and an established large ischemic core that received MT. The aim of our study was to evaluate clinical and neuroradiological predictors of outcome in these patients, including the topographic distribution of early ischemic changes on pre-treatment scans.
Methods
Patients and treatment
In this retrospective observational study, the prospective databases of 16 comprehensive Italian stroke centers were screened for consecutive patients with LVO that received MT, between January 2016 and December 2021. This work was approved by the ethics committee of the Fondazione Policlinico Universitario A. Gemelli IRCCS (protocol no. 6410/20, ID 3004) and of all the other participating centers. The need of informed consent was waived due to the retrospective nature of the study.
All patients were diagnosed with an initial non-contrasted CT scan followed by multiphase CT angiography to locate the site of occlusion. Patients with occlusion involving either the internal carotid artery (ICA), the M1 or M2 segments of the MCA or with tandem occlusion were included. The ASPECTS was calculated from two axial CT slices, one at the level of the thalamus and basal ganglia, and one just rostral to the ganglionic structures. 3 Only patients with an ASPECTS ⩽5 were included in this study. Differential contribution of all ASPECTS regions (caudate nucleus, insula, internal capsule, lentiform nucleus, M1–M6 regions) to the composite score was collected for each patient. Demographics, cardiovascular risk factors, and therapeutic procedures of the acute phase were recorded. Patients with a pre-event modified Rankin Scale (mRS) score >1 or with missing clinical outcome data were excluded.
Intravenous thrombolysis (IVT) was administered when appropriate. 2 MT was performed with a stent-retriever and proximal guide catheter aspiration, direct contact aspiration, or a combination of stent-retriever and distal aspiration, at the discretion of each interventionalist. Successful recanalization was a score of 2b–3 in the modified Treatment In Cerebral Infarction (mTICI) scale. 14 Procedures were conducted under general anesthesia or local anesthesia/conscious sedation at the discretion of the operating physicians or according to the local protocol.
The presence of hemorrhagic transformation was assessed by CT or magnetic resonance imaging between 24 and 72 h after MT and rated according to the Heidelberg classification as follows: (1) parenchymal hematoma (PH) type 1, hematoma occupying less than 30% of the infarcted region without mass effect and (2) PH type 2, hematoma occupying more than 30% of the infarcted region with mass effect and leading to deterioration of clinical condition. 15 Patients presenting small petechiae, either scattered or confluent, were included in the group of patients without any type of PH.
In each participating center, 2 neuroradiologists with more than 5 years of experience and blinded to clinical information reviewed the diagnostic radiological and angiographic data of all patients. In cases without a consensus regarding ASPECTS region scoring, agreement was reached by consultation in a subsequent common session.
Measures of outcome
Clinical outcome was measured with the mRS score acquired at 90 days. Primary outcome measure was a 90-day mRS score 0–2 (functional independence), whereas a score 0–3 (independent ambulation) was the secondary outcome measure.
Statistical analysis
Univariate and multivariate analyses
We used Mann–Whitney U-test/Student’s T-test and Pearson’s chi-square test as appropriate, to assess differences in demographics, baseline clinical, and neuroradiological features between patients with the chosen favorable and unfavorable outcome. Significance threshold was set at p-value < 0.05. Variables with p-value ⩽ 0.1 in univariate analysis were entered in the multivariate analysis. This was performed with a logistic regression model using the chosen outcome measure as dependent variable.
Inter-rater reliability for ASPECTS was evaluated with the Intraclass Correlation Coefficient (ICC). The original scores of the two evaluators were sent by each participating center to the coordinating center and the ICC was therefore calculated centrally on the whole cohort of patients, using a two-way random effects model, single measures, absolute agreement. The interpretation of the ICC value was performed according to the guidelines suggested by Cicchetti and colleagues (<0.40 = poor; 0.40–0.59 = fair; 0.60–0.74 = good; 0.75–1.00 = excellent). 16
Machine learning analysis
The random forest algorithm was used to further explore the effect of topographic distribution of early ischemic changes on clinical outcome. 17 This method differs from the classical single classification tree, where only one tree is constructed and used for predictions. Indeed, while classification trees are easy to interpret, they can produce overfitting when dealing with large or complex data sets. Random forest mitigates this problem by building many decision trees on random subsets of training data and test data. The decisions of each tree are then combined into the final prediction, thereby improving generalization of results. The performance of the model was assessed with the Out-of-Bag (OOB) score that provides an unbiased estimate of error rate. Features relevant for outcome prediction were ranked using the Mean Decreased Gini (MeanDecGini) index, that measures the importance of a variable in terms of how often it appears in the decision trees and how effective it is at splitting the data into the respective classes. Finally, to increase explicability of our model, we applied the Stable and Interpretable RUle Set (SIRUS) method, a classification algorithm inspired by random forests that produces a concise list of rules. 18 The optimal number of rules was determined through cross-validation, with the aim of minimizing the error while maximizing the stability of the method. This approach can provide a list of easily understandable rules useful to support decisions in a clinical setting. Modeling was performed using R software (version 4.3.1) with the party, caret, randomForest, and sirus packages. 19
Results
Four hundred and eight patients with LVO in the anterior circulation and a baseline ASPECTS ⩽5 subjected to MT were available for analysis. Females were 48.3%, and the median (interquartile range, IQR) age was 75 (63–81) years. Patients aged >75 years were 59.3%. The baseline median (IQR) National Institute of Health Stroke Scale (NIHSS) score was 19 (15–22). The median (IQR) ASPECTS was 5 (4–5) and patients with an ASPECTS = 3–5 were 91.7%. Efficient reperfusion was achieved in 81.1% of patients. Rates of patients with a 90-day mRS score of 0–2 or 0–3 were 24.5% and 37.0%, respectively. Rates of PH type 1 and type 2 were 22.0% and 7.8%, respectively. Demographics and relevant clinical, radiological, and procedural data are listed in Table 1.
Table 1.
Demographics and relevant clinical, radiological, and procedural data of patients.
| Demographics | |
| Female, number/total (%) | 197/408 (48.3%) |
| Age in years, median (IQR) | 75 (63-81) |
| Baseline clinical and radiological features | |
| NIHSS score, median (IQR) | 19 (15-22) |
| Atrial fibrillation, number/total (%) | 155/404 (38.4%) |
| Diabetes, number/total (%) | 80/401 (19.9%) |
| Coronary artery disease, number/total (%) | 79/401 (19.7%) |
| Antiplatelet therapy, number/total (%) | 131/400 (32.7%) |
| Anticoagulant therapy, number/total (%) | 81/402 (10.1%) |
| ASPECTS, median (IQR) | 5 (4-5) |
| ASPECTS = 3-5, number/total (%) | 374/408 (91.7%) |
| ASPECTS region involvement | |
| Insula | 94.5% |
| Caudate nucleus | 62.1% |
| Internal Capsule | 35.9% |
| Lentiform Nucleus | 80.8% |
| M1 | 64.6% |
| M2 | 77.0% |
| M3 | 42.7% |
| M4 | 38.5% |
| M5 | 50.2% |
| M6 | 26.2% |
| Site of occlusion, number/total (%) | |
| ICA | 82/407 (20.1%) |
| MCA-M1 | 198/407 (48.6%) |
| MCA-M2 | 30/407 (7.4%) |
| ICA + MCA-M1 | 82/407 (20.1%) |
| ICA + MCA-M2 | 15/407 (3.7%) |
| Therapeutical and procedural data | |
| IVT, number/total (%) | 143/405 (35.3%) |
| Onset-to-groin time in minutes, median (IQR) | 267 (190-346) |
| Onset-to-reperfusion time in minutes, median (IQR) | 320 (245-399) |
| mTICI score 2b-3, number/total (%) | 331/408 (81.1%) |
| Procedure under general anesthesia, number/total (%) | 215/371 (57.9%) |
| Clinical outcome, number/total (%) | |
| 90-day mRS score 0-2 | 100/408 (24.5%) |
| 90-day mRS score 0-2 in patients aged > 75 years | 38/209 (18.2%) |
| 90-day mRS score 0-3 | 151/408 (37.0%) |
| 90-day mRS score 0-3 in patients aged > 75 years | 51/209 (24.4%) |
| Mortality of any cause | 155/408 (38.0%) |
| Parenchymal hemorrhage, number/total (%) | |
| No parenchymal hemorrhage | 286/408 (70.1%) |
| Parenchymal hemorrhage type 1 | 90/408 (22.0%) |
| Parenchymal hemorrhage type 2 | 32/408 (7.8%) |
IQR: interquartile range; NIHSS: National Institute of Health Stroke Scale; ASPECTS: Alberta Stroke Program Early CT score; ICA: internal carotid artery; MCA: middle cerebral artery; IVT: intravenous thrombolysis; mTICI: modified treatment in cerebral infarction; mRS: modified Rankin Scale.
The ICC value confirmed good inter-rater agreement in ASPECTS evaluation (ICC = 0.698, 95% confidence interval (CI) = 0.566–0.796).
In univariate analysis of baseline features, lower age and NIHSS score were associated with the favorable outcome. Rates of patients with ASPECTS = 3–5 was 98.0% in patients with mRS score 0–2 versus 89.6% in patients with unfavorable outcome (p = 0.008). There was no significant difference regarding stroke side and onset-to-groin time. Concerning ASPECTS topography, involvement of the M2, M4, or M5 regions was associated with the unfavorable outcome, whereas involvement of insula was more frequent in patients with the favorable outcome (Table 2). In multivariate model, lower age (odd ratio (OR) = 0.962, 95% CI = 0.943–0.982, p < 0.001) and lower baseline NIHSS score (OR = 0.911, 95% CI = 0.862–0.963, p = 0.001) were associated with a favorable outcome. Involvement of the M2 (OR = 0.398, 95% CI = 0.206–0.770, p = 0.006) or the M4 (OR = 0.496, 95% CI = 0.260–0.945, p = 0.033) regions were instead predictors of unfavorable outcome (Table 3). The prognostic value of M2 and M4 involvement remained significant also after adjusting for successful recanalization and occurrence of PH of any type (Supplemental Table S1).
Table 2.
Univariate analysis of baseline predictors of mRS score 0–2.
| Variable | mRS 0-2 (n = 100) | mRS 3-6 (n = 308) | p-value * |
|---|---|---|---|
| Demographic clinical and therapeutical data | |||
| Female, number (%) | 43 (43.0%) | 154 (50.0%) | 0.224 |
| Age in years, median (IQR) | 69 (59-77) | 76 (66-82) | 0.011 |
| Atrial Fibrillation, number (%) | 35 (35.0%) | 120 (39.5%) | 0.425 |
| Diabetes, number (%) | 14 (14.0%) | 66 (21.9%) | 0.086 |
| Coronary artery disease, number (%) | 14 (15.4%) | 65 (21.6%) | 0.098 |
| Antiplatelet therapy, number (%) | 29 (29.0%) | 102 (34.0%) | 0.356 |
| Anticoagulant therapy, number (%) | 22 (22.0%) | 59 (19.5%) | 0.594 |
| NIHSS score, median (IQR) | 17 (13-19) | 20 (16-22) | <0.001 |
| Left-side stroke, number (%) | 49 (49.0%) | 174 (56.5%) | 0.180 |
| IVT, number (%) | 37 (37.0%) | 106 (34.8%) | 0.683 |
| Onset-to-groin time in minutes, median (IQR) | 252 (180-330) | 270 (196-348) | 0.594 |
| Baseline radiological features | |||
| ASPECTS = 3-5, number (%) | 98 (98.0%) | 276 (89.6%) | 0.008 |
| ASPECTS region, number (%) | |||
| Insula | 97 (99.0%) | 282 (93.0%) | 0.026 |
| Caudate Nucleus | 69 (70.4%) | 180 (59.4%) | 0.051 |
| Internal Capsule | 42 (42.8%) | 102 (33.7%) | 0.099 |
| Lentiform Nucleus | 85 (86.7%) | 239 (78.9%) | 0.086 |
| M1 | 64 (65.3%) | 195 (64.3%) | 0.864 |
| M2 | 63 (64.3%) | 246 (81.2%) | <0.001 |
| M3 | 39 (39.8%) | 132 (43.7%) | 0.496 |
| M4 | 23 (23.5%) | 131 (43.4%) | <0.001 |
| M5 | 35 (35.7%) | 166 (54.7%) | <0.001 |
| M6 | 19 (19.4%) | 86 (28.5%) | 0.076 |
| Occlusion site, number (%) | 0.173 | ||
| ICA | 14 (14.0%) | 68 (22.1%) | |
| MCA-M1 | 56 (56.0%) | 142 (46.2%) | |
| MCA-M2 | 10 (10.0%) | 20 (6.5%) | |
| ICA + MCA-M1 | 16 (16.0%) | 66 (21.5%) | |
| ICA + MCA-M2 | 4 (4.0%) | 11 (3.6%) | |
mRS: modified Rankin Scale; IQR: interquartile range; NIHSS: National Institute of Health Stroke scale; IVT: intravenous thrombolysis; ASPECTS: Alberta Stroke Program Early CT score; ICA: internal carotid artery; MCA: middle cerebral artery.
Statistical significance was considered at p < 0.05.
Table 3.
Multivariate analysis of baseline predictors of mRS score 0–2.
| Variable | OR | 95% CI | p-value * |
|---|---|---|---|
| Age | 0.962 | 0.943-0.982 | <0.001 |
| Diabetes | 0.715 | 0.339-1.507 | 0.378 |
| Coronary artery disease | 0.620 | 0.297-1.293 | 0.202 |
| Baseline NIHSS score | 0.911 | 0.862-0.963 | 0.001 |
| Baseline ASPECTS = 3-5 | 1.889 | 0.346-10.308 | 0.463 |
| ASPECTS region | |||
| Insula | 5.883 | 0.730-47.430 | 0.096 |
| Caudate Nucleus | 0.832 | 0.431-1.608 | 0.584 |
| Internal Capsule | 0.929 | 0.506-1.705 | 0.812 |
| Lentiform Nucleus | 1.032 | 0.458-2.326 | 0.939 |
| M2 | 0.398 | 0.206-0.770 | 0.006 |
| M4 | 0.496 | 0.260-0.945 | 0.033 |
| M5 | 0.606 | 0.335-1.097 | 0.098 |
| M6 | 0.814 | 0.388-1.709 | 0.587 |
mRS: modified Rankin Scale; OR: odds ratio; CI: confidence interval; NIHSS: National Institute of Health Stroke scale; ASPECTS: Alberta Stroke Program Early CT score.
Statistical significance was considered at p < 0.05.
Concerning the 90-day mRS score 0–3, lower age (OR = 0.939, 95% CI = 0.920–0.959, p < 0.001) and lower NIHSS score (OR = 0.903, 95% CI = 0.857–0.951, p < 0.001) remained as the only predictors of favorable outcome. There was a trend toward M4 region involvement being associated with an unfavorable outcome (Supplemental Table S2).
Machine learning analysis confirmed that age and baseline NIHSS score are the most important variables for clinical outcome (MeanDecGini of 34.67 and 32.10 for mRS score 0–2, respectively). Concerning ASPECTS regions, involvement of M5, M4, M2, and M1 can negatively impact functional independence, whereas M4, M1, M5, and M6 areas are relevant for independent ambulation. The OOB estimate of the error rate for our model was 26% for functional independence and 28% for independent ambulation. Graphic representation of the relative importance of each ASPECTS region is reported in Figure 1 and Supplemental Figure S1.
Figure 1.
Color scale representation of the relative importance of the different Alberta Stroke Program Early CT score (ASPECTS) regions after random forest analysis for the primary clinical outcome measure. The M5, M4, M2, and M1 ASPECTS regions show a higher Mean Decrease Gini index, suggesting their relative importance for clinical outcome, in comparison with other regions. Mean Decrease Gini index for these ASPECTS regions was 6.49, 6.23, 5.83, and 5.58, respectively. Mean Decrease Gini index for age and National Institute of Health Stroke Scale score was 34.67 and 32.10, respectively.
SIRUS analysis identified age 77 years and baseline NIHSS score of 20 as values informative for discriminating patients based on the probability of achieving functional independence. The association of either age or baseline NIHSS score values with different ASPECTS region involvement resulted in combination rules for outcome prediction. As an example, the combination of M4 involvement and a baseline NIHSS score ⩾20 was associated with a 1.4% probability of a 90-day mRS score 0–2, and with a 30.4% probability of the same outcome if these criteria were not fulfilled. Combination rules associated with the higher shift in the probability of achieving good outcome for both measures are reported in Table 4 and Supplemental Table S3.
Table 4.
SIRUS combination rules associated with the higher shift in the probability of achieving a 90-day mRS score 0-2.
| SIRUS rule | Probability of mRS 0–2 if respected | Probability of mRS 0–2 if not respected |
|---|---|---|
| Age <77 years and baseline NIHSS <20 | 44.6% | 15.4% |
| M4 involved and baseline NIHSS score ⩾20 | 1.4% | 30.4% |
| M5 involved and baseline NIHSS score ⩾20 | 5.1% | 31.7% |
| Age <77 years and M5 spared | 46.7% | 17.2% |
SIRUS: Stable and Interpretable RUle Set; mRS: modified Rankin Scale; NIHSS: National Institute of Health Stroke Scale.
Discussion
Our retrospective multicenter analysis has shown that, along with age and baseline clinical impairment, presence of early ischemic changes involving cortical areas may impact clinical outcome in patients with large ischemic core undergoing MT.
In our cohort the rates of efficient recanalization and brain bleeding events after MT were in line with those of clinical trials,10–13 thus warranting the generalizability of results concerning the effect of baseline variables on clinical outcome. Stroke side did not appear to influence long-term outcome. Although the mRS score is mainly weighted toward motor disability, our data suggest that lateralization should not be considered when triaging these patients for treatment.
SIRUS analysis showed that younger age can discriminate patients who are likely to go well after MT. This is particularly relevant considering that previous studies have shown that the benefit of MT demonstrated in clinical trials for patients with baseline ASPECTS ⩾6 may not apply to the elderly population in real-world practice. 20 This can be even more true in patients with a large infarct core. However, “The Efficacy and Safety of Thrombectomy in Stroke with Extended Lesion and Extended Time Window” (TENSION) trial has shown an advantage of MT also in patients older than 75 years, although this conclusion derives from a limited sample in this age range. 13 Our result, based on a cohort with 51.2% of patients older than 75 years, agrees with that of the meta-analysis by Cagnazzo et al., 21 showing a reduced benefit of MT in these patients.
Concerning the topographic distribution of ischemic areas, the involvement of specific cortical ASPECTS regions (particularly M4, M5, and M2) has a negative impact on clinical outcome more than non-cortical ganglionic regions. Their relative importance is of course smaller than that of age and baseline clinical deterioration, but the combination of specific age or NIHSS score thresholds with the involvement of certain cortical areas may result in a change in the possibility of achieving a favorable clinical outcome. Previous reports have attempted to tackle the impact of different ASPECTS regions on clinical outcome, although with inconsistent results. Seyedsaadat et al. 22 have shown that early ischemic changes in caudate nucleus, insula, and M4 were associated with poor outcome in patients with a baseline ASPECTS ⩾6 receiving MT, whereas involvement of the M1 region was a predictor of favorable outcome. In a series of patients with baseline large ischemic core, a composite higher cortical ASPECTS (that included the 6 cortical regions) was a predictor of favorable outcome. 23 In another study on patients with ASPECT score ⩾6, the presence of infarct in the left M4 or right M6 regions was associated with poor functional outcome. 24 Finally, data from the “Endovascular Treatment in Ischemic Stroke” (ETIS) registry have shown that involvement of the right M6 region or of the left internal capsule were predictors of long-term increased disability. 25 Although we found no overall effect of laterality, it is conceivable that different ASPECTS regions may have a different relevance depending on the affected hemisphere. It has been proposed that the influence of individual ASPECTS regions on clinical outcome mostly resides in their specific function.22,24,25 On the contrary, their involvement may reflect an insufficient deployment of leptomeningeal collateral vessels, that is a known predictor of poor clinical outcome after MT. 26 Indeed, it has been shown that poor collateralization is associated with infarct mainly involving cortical areas.27,28 This is because most of leptomeningeal collaterals originating from the anterior and posterior cerebral arteries supply cortical areas, whereas the basal ganglia are mainly fed by perforating arteries. Preservation of cortical areas may be therefore less probable in patients with poor deployment of collateral vessels following a proximal occlusion. Independently from the mechanism underlying the specific role of cortical areas in clinical outcome, the information regarding topography of affected ASPECTS regions may be valuable when triaging patients with a baseline large infarct core for acute treatment in a real-world clinical setting.
The main limitation of our study derives from its retrospective nature and non-controlled design. Although clinical and procedural records were carefully reviewed, the results could have been affected by the quality of data not collected within the rigid criteria of a randomized trial. The fact that patients were retrospectively selected from prospective databases, rather than following the boundaries of a trial, carries an important limitation deriving from not knowing the reason/s why these patients were considered for MT in the first place. This is because the endovascular treatment was not the standard of care in patients with an established large ischemic core during the period of time considered for our analysis. As the case-by-case clinical reasoning that have led to offer MT could not be reconstructed, we cannot exclude that features that were relevant for decision-making, and possibly for clinical outcome, were not among the collected variables. Several other information that may be predictive of clinical outcome, such as the admission to an intensive care unit or a stroke unit, was not available in our analysis, as well as a group of patients undergoing best medical management to be used for comparison. Data of perfusion CT imaging that could be instrumental for size assessment of the ischemic core in addition to the visual ASPECTS evaluation were not available. This can be relevant because our study population had a median age that was higher than that of patients enrolled in recent trials10–12 and possibly carried a greater degree of leukoaraiosis that could have impacted the proper evaluation of the ischemic core.
Another limitation may derive from the inter-rater variability of ASPECTS. Although agreement may be good when the composite ASPECTS is dichotomized into 0–5 and 6–10, variability exists when scoring specific regions. 29 Indeed, inter-rater agreement appears high for the insula (96% of agreement) and low for the M3 region (68%), with all other cortical regions scoring between 77% and 84%. 30 In our study, all diagnostic images were carefully reviewed by experts neuroradiologists with good inter-rater reliability. However, because all evaluations were performed locally and without the use of a core lab, we cannot exclude inconsistencies in the detection (or exclusion) of early ischemic changes in specific regions during visual evaluation of baseline scans. We used SIRUS analysis to define combination rules that can be predictive of outcome in a retrospective cohort but that need validation in independent prospective studies before being incorporated in workable prediction models.
On the contrary, our results derive from a relatively large cohort of patients with large ischemic core, with a considerable proportion of patients aged 75 years and older, in a real-world clinical setting.
Conclusion
Our retrospective multicenter analysis has shown that, along with age and baseline clinical impairment, presence of early ischemic changes involving cortical areas may impact clinical outcome in patients with baseline large ischemic core undergoing MT. Nonetheless, these results should be considered with caution and hypothesis generating and need to be confirmed by forthcoming dedicated studies.
Supplemental Material
Supplemental material, sj-pdf-1-wso-10.1177_17474930241245828 for Baseline clinical and neuroradiological predictors of outcome in patients with large ischemic core undergoing mechanical thrombectomy: A retrospective multicenter study by Andrea M Alexandre, Mauro Monforte, Valerio Brunetti, Luca Scarcia, Luigi Cirillo, Andrea Zini, Irene Scala, Vincenzo Nardelli, Francesco Arbia, Giuseppe Arbia, Giovanni Frisullo, Erwah Kalsoum, Arianna Camilli, Davide De Leoni, Francesca Colò, Serena Abruzzese, Mariangela Piano, Claudia Rollo, Antonio Macera, Maria Ruggiero, Elvis Lafe, Joseph D Gabrieli, Giacomo Cester, Nicola Limbucci, Francesco Arba, Simone Ferretti, Valerio Da Ros, Luigi Bellini, Giancarlo Salsano, Nicola Mavilio, Riccardo Russo, Mauro Bergui, Antonio A Caragliano, Sergio L Vinci, Daniele G Romano, Giulia Frauenfelder, Vittorio Semeraro, Maria P Ganimede, Emilio Lozupone, Andrea Romi, Anna Cavallini, Luca Milonia, Massimo Muto, Paolo Candelaresi, Paolo Calabresi, Alessandro Pedicelli and Aldobrando Broccolini in International Journal of Stroke
Footnotes
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by Ricerca Corrente Reti IRCCS 2022, RCR-2022-23682294, Rete IRCCS delle Neuroscienze e della Neuroriabilitazione—RIN, and Istituto Virtuale Nazionale Malattie Cerebrovascolari.
ORCID iDs: Andrea M Alexandre 
https://orcid.org/0000-0002-8080-3916
Andrea Zini
https://orcid.org/0000-0003-1486-4507
Irene Scala
https://orcid.org/0000-0003-2370-840X
Giovanni Frisullo
https://orcid.org/0000-0002-1604-6594
Francesca Colò
https://orcid.org/0000-0002-7164-6584
Serena Abruzzese
https://orcid.org/0009-0002-3793-5566
Francesco Arba
https://orcid.org/0000-0003-3941-7383
Giancarlo Salsano
https://orcid.org/0000-0003-0602-7952
Alessandro Pedicelli
https://orcid.org/0000-0002-2558-8838
Aldobrando Broccolini
https://orcid.org/0000-0001-8295-9271
Supplemental material: Supplemental material for this article is available online.
References
- 1. Goyal M, Menon BK, Van Zwam WH, et al. Endovascular thrombectomy after large-vessel ischaemic stroke: a meta-analysis of individual patient data from five randomised trials. Lancet 2016; 387: 1723–1731. [DOI] [PubMed] [Google Scholar]
- 2. Powers WJ, Rabinstein AA, Ackerson T, et al. Guidelines for the early management of patients with acute ischemic stroke: 2019 update to the 2018 guidelines for the early management of acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2019; 50: e344–e418. [DOI] [PubMed] [Google Scholar]
- 3. Barber PA, Demchuk AM, Zhang J, 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. [DOI] [PubMed] [Google Scholar]
- 4. Barber PA, Hill MD, Eliasziw M, et al. Imaging of the brain in acute ischaemic stroke: comparison of computed tomography and magnetic resonance diffusion-weighted imaging. J Neurol Neurosurg Psychiatry 2005; 76: 1528–1533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. McTaggart RA, Jovin TG, Lansberg MG, et al. Alberta stroke program early computed tomographic scoring performance in a series of patients undergoing computed tomography and MRI: reader agreement, modality agreement, and outcome prediction. Stroke 2015; 46: 407–412. [DOI] [PubMed] [Google Scholar]
- 6. Zaidat OO, Liebeskind DS, Jadhav AP, et al. Impact of age and Alberta stroke program early computed tomography score 0 to 5 on mechanical thrombectomy outcomes: analysis from the STRATIS registry. Stroke 2021; 52: 2220–2228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Goyal M, Demchuk AM, Menon BK, et al. Randomized assessment of rapid endovascular treatment of ischemic stroke. N Engl J Med 2015; 372: 1019–1030. [DOI] [PubMed] [Google Scholar]
- 8. Berkhemer OA, Fransen PSS, Beumer D, et al. A randomized trial of intraarterial treatment for acute ischemic stroke. N Engl J Med 2015; 372: 11–20. [DOI] [PubMed] [Google Scholar]
- 9. Jovin TG, Chamorro A, Cobo E, et al. Thrombectomy within 8 hours after symptom onset in ischemic stroke. N Engl J Med 2015; 372: 2296–2306. [DOI] [PubMed] [Google Scholar]
- 10. Yoshimura S, Sakai N, Yamagami H, et al. Endovascular therapy for acute stroke with a large ischemic region. N Engl J Med 2022; 386: 1303–1313. [DOI] [PubMed] [Google Scholar]
- 11. Huo X, Ma G, Tong X, et al. Trial of endovascular therapy for acute ischemic stroke with large infarct. N Engl J Med 2023; 388: 1272–1283. [DOI] [PubMed] [Google Scholar]
- 12. Sarraj A, Hassan AE, Abraham MG, et al. Trial of endovascular thrombectomy for large ischemic strokes. N Engl J Med 2023; 388: 1259–1271. [DOI] [PubMed] [Google Scholar]
- 13. Bendszus M, Fiehler J, Subtil F, et al. Endovascular thrombectomy for acute ischaemic stroke with established large infarct: multicentre, open-label, randomised trial. Lancet 2023; 402: 1753–1763. [DOI] [PubMed] [Google Scholar]
- 14. Gerber JC, Miaux YJ, Von Kummer R. Scoring flow restoration in cerebral angiograms after endovascular revascularization in acute ischemic stroke patients. Neuroradiology 2015; 57: 227–240. [DOI] [PubMed] [Google Scholar]
- 15. Von Kummer R, Broderick JP, Campbell BC, et al. The Heidelberg bleeding classification: classification of bleeding events after ischemic stroke and reperfusion therapy. Stroke 2015; 46: 2981–2986. [DOI] [PubMed] [Google Scholar]
- 16. Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychol Assessment 1994; 6: 284–290. [Google Scholar]
- 17. Breiman L. Random forests. Mach Learn 2001; 45: 5–32. [Google Scholar]
- 18. Bénard C, Biau G, Da Veiga S, et al. SIRUS: stable and Interpretable RUle set for classification. Electron J Stat 2021; 15: 427–505. [Google Scholar]
- 19. R Core Team. R: a language and environment for statistical computing, https://www.R-project.org (accessed 9 November 2023).
- 20. Groot AE, Treurniet KM, Jansen IGH, et al. Endovascular treatment in older adults with acute ischemic stroke in the MR CLEAN Registry. Neurology 2020; 95: e131–e139. [DOI] [PubMed] [Google Scholar]
- 21. Cagnazzo F, Derraz I, Dargazanli C, et al. Mechanical thrombectomy in patients with acute ischemic stroke and ASPECTS ⩽6: a meta-analysis. J Neurointerv Surg 2020; 12: 350–355. [DOI] [PubMed] [Google Scholar]
- 22. Seyedsaadat SM, Neuhaus AA, Nicholson PJ, et al. Differential contribution of ASPECTS regions to clinical outcome after thrombectomy for acute ischemic stroke. AJNR Am J Neuroradiol 2021; 42: 1104–1108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Xing PF, Zhang YW, Zhang L, et al. Higher baseline cortical score predicts good outcome in patients with low Alberta stroke program early computed tomography score treated with endovascular treatment. Neurosurgery 2021; 88: 612–618. [DOI] [PubMed] [Google Scholar]
- 24. Rangaraju S, Streib C, Aghaebrahim A, Jadhav A, Frankel M, Jovin TG. Relationship between lesion topology and clinical outcome in anterior circulation large vessel occlusions. Stroke 2015; 46: 1787–1792. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Panni P, Michelozzi C, Blanc R, et al. The role of infarct location in patients with DWI-ASPECTS 0-5 acute stroke treated with thrombectomy. Neurology 2020; 95: e3344–e3354. [DOI] [PubMed] [Google Scholar]
- 26. Berkhemer OA, Jansen IGH, Beumer D, et al. Collateral status on baseline computed tomographic angiography and intra-arterial treatment effect in patients with proximal anterior circulation stroke. Stroke 2016; 47: 768–776. [DOI] [PubMed] [Google Scholar]
- 27. Verma RK, Gralla J, Klinger-Gratz PP, et al. Infarction distribution pattern in acute stroke may predict the extent of leptomeningeal collaterals. PLoS ONE 2015; 10: e0137292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Seyman E, Shaim H, Shenhar-Tsarfaty S, Jonash-Kimchi T, Bornstein NM, Hallevi H. The collateral circulation determines cortical infarct volume in anterior circulation ischemic stroke. BMC Neurol 2016; 16: 206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Nicholson P, Hilditch CA, Neuhaus A, et al. Per-region interobserver agreement of Alberta Stroke Program Early CT Scores (ASPECTS). J Neurointerv Surg 2020; 12: 1069–1071. [DOI] [PubMed] [Google Scholar]
- 30. Van Horn N, Kniep H, Broocks G, et al. ASPECTS interobserver agreement of 100 investigators from the TENSION study. Clin Neuroradiol 2021; 31: 1093–1100. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental material, sj-pdf-1-wso-10.1177_17474930241245828 for Baseline clinical and neuroradiological predictors of outcome in patients with large ischemic core undergoing mechanical thrombectomy: A retrospective multicenter study by Andrea M Alexandre, Mauro Monforte, Valerio Brunetti, Luca Scarcia, Luigi Cirillo, Andrea Zini, Irene Scala, Vincenzo Nardelli, Francesco Arbia, Giuseppe Arbia, Giovanni Frisullo, Erwah Kalsoum, Arianna Camilli, Davide De Leoni, Francesca Colò, Serena Abruzzese, Mariangela Piano, Claudia Rollo, Antonio Macera, Maria Ruggiero, Elvis Lafe, Joseph D Gabrieli, Giacomo Cester, Nicola Limbucci, Francesco Arba, Simone Ferretti, Valerio Da Ros, Luigi Bellini, Giancarlo Salsano, Nicola Mavilio, Riccardo Russo, Mauro Bergui, Antonio A Caragliano, Sergio L Vinci, Daniele G Romano, Giulia Frauenfelder, Vittorio Semeraro, Maria P Ganimede, Emilio Lozupone, Andrea Romi, Anna Cavallini, Luca Milonia, Massimo Muto, Paolo Candelaresi, Paolo Calabresi, Alessandro Pedicelli and Aldobrando Broccolini in International Journal of Stroke