Background and Purpose—: The prediction of long-term outcomes in ischemic stroke patients may be useful in treatment decisions. Machine learning techniques are being increasingly adapted for use in the medical field because of their high accuracy. This study investigated the applicability of machine learning techniques to predict long-term outcomes in ischemic stroke patients. Methods—: This was a retrospective study using a prospective cohort that enrolled patients with acute ischemic stroke. Favorable outcome was defined as modified Rankin Scale score 0, 1, or 2 at 3 months. We developed 3 machine learning models (deep neural network, random forest, and logistic regression) and compared their predictability. To evaluate the accuracy of the machine learning models, we also compared them to the Acute Stroke Registry and Analysis of Lausanne (ASTRAL) score. Results—: A total of 2604 patients were included in this study, and 2043 (78%) of them had favorable outcomes. The area under the curve for the deep neural network model was significantly higher than that of the ASTRAL score (0.888 versus 0.839; P <0.001), while the areas under the curves of the random forest (0.857; P =0.136) and logistic regression (0.849; P =0.413) models were not significantly higher than that of the ASTRAL score. Using only the 6 variables that are used for the ASTRAL score, the performance of the machine learning models did not significantly differ from that of the ASTRAL score. Conclusions—: Machine learning algorithms, particularly the deep neural network, can improve the prediction of long-term outcomes in ischemic stroke patients.

Background and Purpose—: Endovascular treatment for acute intracranial atherosclerosis–related large vessel occlusion (ICAS [+]-LVO) is one of the challenging issues in modern mechanical thrombectomy era. We evaluated procedural and clinical outcomes of endovascular treatment for the ICAS (+)-LVO. We also compared their outcomes with those of large vessel occlusion not associated with intracranial atherosclerosis (ICAS [−]-LVO). Methods—: We retrospectively reviewed consecutive patients with acute stroke who underwent endovascular treatment for LVO. Patients were assigned to the ICAS (+)-LVO group or the ICAS (−)-LVO group primarily based on catheter angiogram. Procedural and clinical outcomes were compared between the ICAS (+)-LVO and ICAS (−)-LVO groups. Results—: The present study included 318 patients. Fifty-six patients (17.6%) had an ICAS (+)-LVO. Recanalization was achieved in 45 patients in the ICAS (+)-LVO group (80.4%), which was comparable with the ICAS (−)-LVO group (88.5%; P =0.097). However, recanalization using a stent retriever was less successful in the ICAS (+)-LVO (28.9%) than the ICAS (−)-LVO group (93.5%). Of the remaining patients in the ICAS (+)-LVO group, 84.3% of patients required specific rescue treatments appropriate for ICAS, including balloon angioplasty, stenting, and intra-arterial glycoprotein IIb/IIIa inhibitor infusion. The rates of favorable outcomes (46.4% versus 46.9%), death, and symptomatic intracranial hemorrhage were not significantly different between the 2 groups. Glycoprotein IIb/IIIa inhibitor use was not significantly associated with symptomatic intracranial hemorrhage. Conclusions—: ICAS (+)-LVO was often refractory to mechanical thrombectomy. With specific rescue treatments appropriate for ICAS, patients in the ICAS (+)-LVO group had a recanalization rate comparable with patients in the ICAS (−)-LVO. With comparable recanalization rate, the clinical outcomes did not differ between patients with ICAS (+)-LVO and ICAS (−)-LVO.

Background and Purpose—: Effective rescue treatment has not yet been suggested in patients with mechanical thrombectomy (MT) failure. This study aimed to test whether rescue stenting (RS) improved clinical outcomes in MT-failed patients. Methods—: This is a retrospective analysis of the cohorts of the 16 comprehensive stroke centers between September 2010 and December 2015. We identified the patients who underwent MT but failed to recanalize intracranial internal carotid artery or middle cerebral artery M1 occlusion. Patients were dichotomized into 2 groups: patients with RS and without RS after MT failure. Clinical and laboratory findings and outcomes were compared between the 2 groups. It was tested whether RS is associated with functional outcome. Results—: MT failed in 148 (25.0%) of the 591 patients with internal carotid artery or middle cerebral artery M1 occlusion. Of these 148 patients, 48 received RS (RS group) and 100 were left without further treatment (no stenting group). Recanalization was successful in 64.6% (31 of 48 patients) of RS group. Compared with no stenting group, RS group showed a significantly higher rate of good outcome (modified Rankin Scale score, 0–2; 39.6% versus 22.0%; P =0.031) without increasing symptomatic intracranial hemorrhage (16.7% versus 20.0%; P =0.823) or mortality (12.5% versus 19.0%; P =0.360). Of the RS group, patients who had recanalization success had 54.8% of good outcome, which is comparable to that (55.4%) of recanalization success group with MT. RS remained independently associated with good outcome after adjustment of other factors (odds ratio, 3.393; 95% confidence interval, 1.192–9.655; P =0.022). Follow-up vascular imaging was available in the 23 (74.2%) of 31 patients with recanalization success with RS. The stent was patent in 20 (87.0%) of the 23 patients. Glycoprotein IIb/IIIa inhibitor was significantly associated with stent patency but not with symptomatic intracranial hemorrhage. Conclusions—: RS was independently associated with good outcomes without increasing symptomatic intracranial hemorrhage or mortality. RS seemed considered in MT-failed internal carotid artery or middle cerebral artery M1 occlusion.

Summary: Increased red blood cell distribution width (RDW), which is a marker of anisocytosis, is associated with mortality and cardiovascular events in the general population and in patients with heart failure or coronary heart disease. We investigated whether RDW in acute cerebral infarction is predictive of functional outcome and mortality. A total of 847 consecutive patients with first-ever acute cerebral infarction who presented to the emergency department within seven days of symptom onset were enrolled in this study. We investigated the association of RDW with poor functional outcome (modified Rankin Scale >2) and all-cause mortality at three months, as well as survival time for one year after stroke onset. Multivariate logistic regression revealed that higher RDW was independently associated with poor functional outcome (adjusted odds ratio [OR], 1.222 per 1% increment in RDW, 95% confidence interval [CI] 1.059–1.409, p=0.006) and all-cause death (adjusted OR, 1.395 per 1% increment in RDW, 95% CI 1.168–1.665, p<0.001) at three months after stroke onset. RDW was an independent predictor of survival in multivariate Cox-proportional regression model (adjusted hazard ratio, 1.328 per 1% increment in RDW, 95%CI 1.178–1.498, p<0.001). The addition of RDW to a survival model significantly increased predictability for survival across the entire follow-up period (weighted average of the area-under the curves, 0.858 vs. 0.841, p<0.05). In conclusion, higher RDW measured in cases of acute stage cerebral infarction was associated with poor functional outcome and mortality. RDW may be used as a biomarker for the prediction of long-term outcomes in patients with acute cerebral infarction.

Objective: To determine associations between cerebral microbleed (CMB) burden with recurrent ischemic stroke (IS) and intracerebral hemorrhage (ICH) risk after IS or TIA. Methods: We identified prospective studies of patients with IS or TIA that investigated CMBs and stroke (ICH and IS) risk during ≥3 months follow-up. Authors provided aggregate summary-level data on stroke outcomes, with CMBs categorized according to burden (single, 2–4, and ≥5 CMBs) and distribution. We calculated absolute event rates and pooled risk ratios (RR) using random-effects meta-analysis. Results: We included 5,068 patients from 15 studies. There were 115/1,284 (9.6%) recurrent IS events in patients with CMBs vs 212/3,781 (5.6%) in patients without CMBs (pooled RR 1.8 for CMBs vs no CMBs; 95% confidence interval [CI] 1.4–2.5). There were 49/1,142 (4.3%) ICH events in those with CMBs vs 17/2,912 (0.58%) in those without CMBs (pooled RR 6.3 for CMBs vs no CMBs; 95% CI 3.5–11.4). Increasing CMB burden increased the risk of IS (pooled RR [95% CI] 1.8 [1.0–3.1], 2.4 [1.3–4.4], and 2.7 [1.5–4.9] for 1 CMB, 2–4 CMBs, and ≥5 CMBs, respectively) and ICH (pooled RR [95% CI] 4.6 [1.9–10.7], 5.6 [2.4–13.3], and 14.1 [6.9–29.0] for 1 CMB, 2–4 CMBs, and ≥5 CMBs, respectively). Conclusions: CMBs are associated with increased stroke risk after IS or TIA. With increasing CMB burden (compared to no CMBs), the risk of ICH increases more steeply than that of IS. However, IS absolute event rates remain higher than ICH absolute event rates in all CMB burden categories.