Junpei Koge

Background: Predicting outcomes after endovascular thrombectomy (EVT) for acute ischemic stroke is challenging. This study aimed to develop a machine learning model to predict outcomes after EVT using only clinical characteristics available shortly after hospital arrival, excluding imaging, laboratory, and angiographic data. Methods: We analyzed data from the JSDB (Japan Stroke Data Bank), a multicenter, prospective registry of patients with acute stroke. A total of 4353 patients who underwent EVT between 2016 and 2020 were included, with those from 2016 to 2019 assigned to the derivation cohort and those from 2020 to the validation cohort. The model was developed using 132 routinely collected clinical variables that can be rapidly obtained on patient arrival. Feature selection was performed using the Boruta algorithm, and the model was built using CatBoost. The primary outcome was a modified Rankin Scale (mRS) score of 0 to 2 at discharge. Results: Among 4353 patients treated with EVT, the derivation cohort included 2764 patients and the validation cohort, 1589 patients. In these cohorts, 973 (35.2%) and 547 (34.4%) achieved mRS 0 to 2, respectively. Boruta selected 47 predictors. CatBoost achieved an area under the receiver operating characteristic curve of 0.744 (95% CI, 0.710–0.776) in the derivation test set and 0.781 (95% CI, 0.755–0.803) in the validation cohort. Validation areas under the receiver operating characteristic curves were 0.792 (95% CI, 0.765–0.815) for mRS 0 to 1, 0.766 (95% CI, 0.744–0.787) for mRS 0 to 3, and 0.680 (95% CI, 0.631–0.729) for mRS 6. Conclusions: Our machine learning model predicted outcomes following EVT using only admission clinical data. This imaging‐independent machine learning approach may support early prognostication and facilitate discussions of expected clinical courses.

BACKGROUND: The ABCD ² score is widely used to stratify stroke risk after transient neurological events. We aimed to evaluate the association between the ABCD ² score and the risk of subsequent ischemic stroke among patients with diffusion-weighted imaging (DWI)–positive (DWI+) or DWI−negative (DWI−) findings for the index event. METHODS: We pooled individual participant data from 3 multicenter cohorts including patients presenting with transient or minor neurological events. Patients presenting with transient neurological symptoms, a normal neurological examination, and available DWI for the index event were included. Associations between the ABCD ² score and 90-day ischemic stroke risk were analyzed separately in the DWI+ and DWI− groups using logistic regression models adjusted for sex, atrial fibrillation, and study source. RESULTS: Among 2014 patients (1143 [56.8%] men; median age 67 years), the median ABCD ² score was 4 (interquartile range, 3–5), and 504 (25.0%) were DWI+. Ischemic stroke occurred within 90 days in 20 patients (4.0%) in the DWI+ group and 33 (2.2%) in the DWI− group. The ABCD ² score was associated with stroke risk in the DWI− group (per 1-point increase, odds ratio, 1.36 [95% CI, 1.04–1.79]), but this relationship was not significant in the DWI+ group (per 1-point increase, odds ratio, 1.08 [95% CI, 0.76–1.53]). When stratified ABCD ² score into 0 to 3, 4 to 5, and 6 to 7, the predicted 90-day stroke risks were 3.2%, 3.9%, and 4.5% in the DWI+ group (per 1-category increase, odds ratio, 1.23 [95% CI, 0.63–2.40]) and 1.0%, 2.3%, and 4.1% in the DWI− group (per 1-category increase, odds ratio, 1.88 [95% CI, 1.11–3.20]), respectively. CONCLUSIONS: The ABCD ² score was associated with 90-day stroke risk among patients with DWI− findings. Patients with both a low ABCD ² score and DWI− findings had a particularly low risk of subsequent stroke, which may inform risk stratification and clinical decision-making after transient neurological events.

Introduction: Distal occlusion tracker (DOT) signs on immediate post-interventional flat-panel detector CT (FDCT) can identify residual vessel occlusions underestimated by conventional angiography after endovascular thrombectomy (EVT). However, whether these FDCT-detected occlusions are associated with tissue-level outcomes remains unclear. We investigated whether FDCT-augmented reperfusion grading better reflects tissue imaging outcomes than operator-assessed reperfusion grading in patients with ischaemic stroke. Patients and methods: We retrospectively analysed EVT-treated patients with expanded thrombolysis in cerebral infarction (TICI) ≥ 2b who underwent post-interventional FDCT and follow-up MRI. DOT signs were defined as punctiform or tubular hyperdensities on FDCT. Final infarct volume (FIV) was segmented using follow-up DWI. Reperfusion status was categorised as incomplete (TICI 2b) or near-complete/complete (TICI 2c/3) based on FDCT-augmented TICI assessed by a core laboratory or operator-graded TICI. Results: Among 200 patients, the median FIV was 20.0 mL (IQR, 5.9–74.6), and DOT signs were identified in 86 (43.0%). Operator-graded TICI classified 61 patients (30.5%) as having incomplete reperfusion, compared with 72 patients (36.0%) by FDCT-augmented TICI; 17 patients were reclassified. Compared with near-complete/complete reperfusion, incomplete reperfusion by FDCT-augmented TICI was associated with larger FIV (β = 24.0 mL; 95% CI, 3.6–44.4), whereas operator-defined incomplete reperfusion was not (β = 12.2 mL; 95% CI, −9.5 to 33.9). The model incorporating FDCT-augmented TICI explained slightly more variance in FIV than the model incorporating operator-graded TICI (adjusted R2 0.256 vs 0.239). Conclusion: FDCT-augmented reperfusion assessment may provide additional tissue-relevant information after EVT by identifying residual vessel occlusions associated with larger FIV.

Purpose: The Woven EndoBridge (WEB) was introduced in Japan in January 2021 and approved for all subtypes of wide-neck bifurcation aneurysms (WNBA). This retrospective study evaluated the safety and efficacy of the WEB device for all subtypes of WNBA.Materials and Methods: All patients treated with the WEB at our facility between January 2021 and May 2024 was reviewed. We selected the WEB device according to an oversizing policy, based on cumulative clinical evidence from global experience.Results: We analyzed 120 aneurysms in 117 patients (56 males and 61 females with a mean age of 65.5±12.7 years). There were 45 anterior communicating artery aneurysms, 27 middle cerebral artery aneurysms, 17 internal carotid artery-posterior communicating artery aneurysms, 15 basilar artery top aneurysms, and 16 aneurysms in other locations. Aneurysm characteristics included a maximum diameter of 6.5 [5.3, 7.7] mm, height of 4.9 [3.9, 6.0] mm, width of 4.8 [4.0, 6.2] mm, and dome/neck ratio of 1.2 [1.1, 1.4]. All data are expressed in median [interquartile range]. Angiographic follow-up at 12 months in 96 cases showed complete obliteration in 68.8% and adequate obliteration in 90.6% of cases. Intraoperative ischemic events occurred in 5 cases (4.2%). Hemorrhagic events occurred in 2 cases (1.7%), with symptoms resolving by discharge, except for 1 case of mild paralysis. During follow-up, 1 patient developed a major stroke, resulting in morbidity (0.8%). Retreatment was required in 3 cases (2.5%). On multivariate analysis for complete occlusion at 12 months following WEB treatment, age was statistically associated with the outcome (odds ratio, 0.957 per year; 95% confidence interval, 0.919–0.996; P=0.033).Conclusion: WEB is safe and effective for all subtypes of WNBA, with a low retreatment rate using an oversizing policy. This is the first report in a Japanese population.

BACKGROUND: Insertable cardiac monitoring (ICM) detects atrial fibrillation (AF) in substantial proportions of cryptogenic stroke, noncryptogenic ischemic stroke without known AF, and nonstroke patients who are at risk of underlying AF. Given differences in patient characteristics across studies, there may be differences in AF detection rates on ICM across these subgroups that have not been identified. We investigate whether AF detection rates on ICM are higher in cryptogenic stroke or transient ischemic attack (C-IS/TIA) patients compared with individuals with noncryptogenic stroke or without stroke, when accounting for differences in study populations. METHODS: This is an individual-participant data meta-analysis of prospective studies and randomized controlled trials of ICM in C-IS/TIA, noncryptogenic ischemic stroke, and nonstroke patients. Multilevel multivariable logistic regression models were used to test whether C-IS/TIA is associated with increased AF detection relative to other categories. We performed multiple imputation to derive values for variables with P =0.009), indicating that 47% of AF detected in C-IS/TIA is pathogenic. Limiting the comparator group to ischemic stroke or history of stroke yielded similar results (adjusted odds ratio, 2.83 [95% CI, 1.47–5.44]; P =0.002). Days to AF detection were significantly shorter in C-IS/TIA patients (median 65 versus 169; P <0.001). CONCLUSIONS: In this individual-participant data meta-analysis of patients undergoing ICM, AF detection was higher in C-IS/TIA patients, with shorter time to AF detection compared with noncryptogenic/nonstroke individuals. These findings suggest that some of the AF detected in patients with C-IS/TIA may be pathogenic.

BACKGROUND AND OBJECTIVES: We aimed to clarify the association between neurological deterioration pre-endovascular therapy (EVT) and outcome in patients with large-vessel occlusion due to intracranial atherosclerotic disease (ICAD-LVO) undergoing EVT. METHODS: Consecutive patients with acute ischemic stroke due to ICAD-LVO within 24 h of onset who underwent EVT were enrolled in the Japanese multicenter registry from 2017 to 2019. Patients were grouped according to neurological severity transition as follows: mild symptoms (baseline National Institutes of Health Stroke Scale [NIHSS] score <6 and NIHSS score pre-EVT <6), symptom deterioration (baseline NIHSS score <6 and NIHSS score pre-EVT ≥6), and severe symptoms (baseline NIHSS score ≥6 and NIHSS score pre-EVT ≥6). Outcomes included favorable outcomes (modified Rankin Scale [mRS] score of 0-2 at 90 days), ordinal mRS shift, and symptomatic intracranial hemorrhage. Multivariable logistic regression assessed the association of outcomes with the transition of neurological severity by calculating odds ratios and 95% CIs, with mild symptoms as reference. RESULTS: In total, 480 patients with acute ICAD-LVO who underwent EVT (150 women [31.2%]; median age, 72 years IQR, 66-80) and had median baseline NIHSS score 12 (IQR, 6-20) were analyzed. Patients with symptom deterioration (n = 34) and severe symptoms (n = 375) had lower favorable outcomes (deterioration 38.2% vs mild 62.9%; adjusted odds ratio 0.30, 95% CI 0.09-0.97, severe 35.3%; 0.47, 0.33-0.65) and a significant mRS shift (deterioration vs mild; 3.63, 1.46-9.03, severe; 2.27, 1.74-2.96) than those with mild symptoms (n = 71). Symptomatic intracranial hemorrhage rates did not differ (mild 0%; deterioration 0%; severe 1.9%). CONCLUSION: Patients with ICAD-LVO who experienced worsening symptoms were less likely to achieve favorable outcomes after EVT than those with mild symptoms. Early identification of neurological deterioration and EVT intervention may improve outcomes in these patients.

Purpose Digital subtraction angiography (DSA) is the gold standard for follow-up evaluation of intracranial aneurysms treated with the Woven EndoBridge (WEB) device. This study aimed to assess the efficacy of high-resolution CT angiography (HR-CTA) as a less invasive alternative by comparing its diagnostic performance with that of DSA. Methods This single-center retrospective study analyzed the angiographic and clinical data of patients treated with the WEB device for cerebral aneurysms between January 2021 and December 2024. Patients who underwent HR-CTA within 2 weeks before or after follow-up DSA were included. Occlusion status was assessed using the Bicêtre Occlusion Scale Score (BOSS) and binary classification. The concordance rate between HR-CTA and DSA was evaluated. Results A total of 54 eligible examinations were identified. Using the BOSS, 46 examinations were concordant, resulting in an agreement rate of 85.2%. The Cohen’s κ coefficient was 0.81 (95% CI 0.69 to 0.93), indicating a very high level of agreement. All discordant cases resulted from HR-CTA overestimating occlusion status; however, HR-CTA accurately identified aneurysm remnants. Univariate analyses identified BOSS 0’ as the only significant factor contributing to discrepancies. In the binary evaluation, all 54 examinations were fully concordant (κ=1.00, 95% CI 1.00 to 1.00). Conclusions HR-CTA demonstrated a high concordance rate with DSA for evaluating occlusion status after WEB placement. Its reliable assessment of aneurysm remnants suggests HR-CTA could serve as a practical and less invasive alternative to DSA in follow-up evaluations.

Purpose Visualizing the culprit perforating artery in subcortical infarction using in vivo imaging is challenging. We aimed to identify the culprit perforating arteries in subcortical infarctions and assess their morphology using an image fusion technique. Methods We retrospectively reviewed consecutive patients who had an ischemic stroke in the anterior circulation perforating area (caudate nucleus, lentiform nucleus, internal capsule, corona radiata, or centrum semiovale) and underwent three-dimensional rotational-angiography (3D-RA) and 3D fluid-attenuated inversion recovery MRI. Images were registered using an original fusion software. The spatial relationship between the infarction and culprit perforating artery and its morphological characteristics were analyzed in the fusion images. Stenosis was defined as > 50% luminal narrowing or a focal intraluminal defect in the perforating artery. Results Of 118 patients, the culprit perforating artery was identified in 52 patients (44%); They tended to have younger age and had a higher baseline NIHSS score and higher prevalence of infarcts in the lentiform nucleus than did those without identified culprit perforating artery. Among the 44 patients with assessable morphology of the culprit perforating artery, 27 (61%) exhibited stenosis in the proximal segment. Atrial fibrillation was more frequent in patients without stenosis in the proximal segment of the culprit perforating artery than in those with stenosis (29% vs. 4%, P = 0.03). Conclusion The 3D-RA and MRI fusion technique enables identification of the culprit perforating arteries in subcortical infarctions, especially in the lentiform nucleus. Morphological features of the culprit perforating artery may be associated with the etiological mechanism of stroke.