Effect of Statin Therapy on Cardiovascular Outcome in Stroke Patients with Low Baseline Low-Density Lipoprotein Cholesterol.

OBJECTIVES: To investigate whether post-stroke statin therapy reduces subsequent major vascular events in statin-naïve patients with pretreatment low-density lipoprotein cholesterol (LDL-C) below the recommended target (≤70 mg/dL for atherosclerotic stroke and ≤100 mg/dL for non-atherosclerotic stroke) at stroke onset. METHODS: Patients from an ongoing stroke registry who had an ischemic stroke between 2011 and 2020 were screened. Statin naïve patients with baseline LDL-C below the target were assessed. The effect of post-stroke statin therapy on major vascular events (composite of recurrent stroke, myocardial infarction, and death) was investigated using weighted Cox regression analyses using stabilized inverse probability treatment weighting. RESULTS: The baseline LDL-C level of the 1,858 patients (mean age 67.9 ± 15.3 years, 61.4% men, 13.2% atherosclerotic stroke) included in the study was 75.7 ± 17.0 mg/dL. Statins were prescribed to 1,256 (67.7%) patients (low-to-moderate intensity, 23.5%; high intensity, 44.1%). Post-stroke statin therapy was associated with a lower risk of major vascular events during 1-year follow-up (weighted hazard ratio 0.55, 95% confidence interval 0.42-0.71). In a subgroup of patients who were at very high risk of atherosclerotic cardiovascular disease with LDL-C <55 mg/dL or patients who were not at very high risk of atherosclerotic cardiovascular disease with LDL-C <70 mg/dL, post-stroke statin therapy was also associated with a reduction in major vascular events (weighted hazard ratio 0.45, 95% confidence interval 0.29-0.70). The intensity of the most beneficial statin varied by subtype of stroke. INTERPRETATION: Statin therapy may improve vascular outcomes after ischemic stroke, even in cases of LDL-C below the target without pre-stroke lipid-lowering therapy. ANN NEUROL 2024;95:876-885.

Tricuspid regurgitation: a hidden risk factor for atrial fibrillation related stroke?

Background and purpose: Tricuspid regurgitation (TR) is a common but overlooked valvular disease, and its association with the etiologic subtypes of ischemic stroke is unclear. We explored the relationship between TR and atrial fibrillation (AF) in patients with acute ischemic stroke. Methods: This retrospective analysis of ongoing stroke registry assessed 6,886 consecutive acute ischemic stroke patients who underwent transthoracic echocardiography during their in-hospital care. Multivariable logistic regression models adjusted for age, sex, stroke characteristics, and echocardiographic indices were used to investigate the association between TR and total AF, and newly diagnosed AF during hospitalization and a 1-year follow-up period, respectively. Results: TR was present in 877 (12.7%) patients (mild, 9.9%; moderate, 2.4%; severe, 0.5%). AF was identified in 24.1% (medical history, 11.1%; first detected in the emergency room, 6.6%; newly diagnosed after admission, 6.4%). TR was associated with AF [adjusted odds ratio (aOR) 4.87 (95% confidence interval (CI), 2.63-9.03)], compared with no/trivial TR. The association between TR and AF was consistent regardless of severity (aOR [95% CI], 4.57 [2.63-7.94] for mild and 7.05 [2.57-19.31] for moderate-to-severe TR) or subtype of TR (5.44 [2.91-10.14] for isolated and 3.81 [2.00-7.28] for non-isolated TR). Among the AF-naïve patients at admission, TR was associated with newly diagnosed AF during hospitalization and a 1-year follow-up period (aOR [95% CI], 2.68 [1.81-3.97]). Conclusions: TR is associated with AF in acute ischemic stroke patients regardless of severity and subtypes of TR. TR is also associated with newly diagnosed AF after stroke.

Long-term outcomes and quantitative radiologic analysis of extracranial-intracranial bypass for hemodynamically compromised chronic large artery occlusive disease.

This study aimed to demonstrate the effectiveness of nonemergent extracranial-to-intracranial bypass (EIB) in symptomatic chronic large artery atherosclerotic stenosis or occlusive disease (LAA) through quantitative analysis of computed tomography perfusion (CTP) parameters using RAPID software. We retrospectively analyzed 86 patients who underwent nonemergent EIB due to symptomatic chronic LAA. CTP data obtained preoperatively, immediately postoperatively (PostOp0), and 6 months postoperatively (PostOp6M) after EIB were quantitatively analyzed through RAPID software, and their association with intraoperative bypass flow (BF) was assessed. The clinical outcomes, including neurologic state, incidence of recurrent infarction and complications, were also analyzed. The time-to-maximum (Tmax) > 8 s, > 6 s and > 4 s volumes decreased significantly at PostOp0 and up through PostOp6M (preoperative, 5, 51, and 223 ml (median), respectively; PostOp0, 0, 20.25, and 143 ml, respectively; PostOp6M, 0, 7.5, and 148.5 ml, respectively; p < 0.001, p < 0.001, and p < 0.001, respectively). The postoperative improvement in the Tmax > 6 s and > 4 s volumes was significantly correlated with the BF at PostOp0 and PostOp6M (PostOp0, r = 0.367 (p = 0.001) and r = 0.275 (p = 0.015), respectively; PostOp6M r = 0.511 (p < 0.001) and r = 0.391 (p = 0.001), respectively). The incidence of recurrent cerebral infarction was 4.7%, and there were no major complications that produced permanent neurological impairment. Nonemergent EIB under strict operation indications can be a feasible treatment for symptomatic, hemodynamically compromised LAA patients.

Automated deep neural network analysis of lateral cephalogram data can aid in detecting obstructive sleep apnea.

STUDY OBJECTIVES: Information on obstructive sleep apnea (OSA) is often latently detected in diagnostic tests conducted for other purposes, providing opportunities for maximizing value. This study aimed to develop a convolutional neural network (CNN) to identify the risk of OSA using lateral cephalograms. METHODS: The lateral cephalograms of 5,648 individuals (mean age, 49.0 ± 15.8 years; men, 62.3%) with or without OSA were collected and divided into training, validation, and internal test datasets in a 5:2:3 ratio. A separate external test dataset (n = 378) was used. A densely connected CNN was trained to diagnose OSA using a cephalogram. Model performance was evaluated using the area under the receiver operating characteristic curve (AUROC). Gradient-weighted class activation mapping (Grad-CAM) was used to evaluate the region of focus, and the relationships between the model outputs, anthropometric characteristics, and OSA severity were evaluated. RESULTS: The AUROC of the model for the presence of OSA was 0.82 (95% confidence interval, 0.80-0.84) and 0.73 (95% confidence interval, 0.65-0.81) in the internal and external test datasets, respectively. Grad-CAM demonstrated that the model focused on the area of the tongue base and oropharynx in the cephalogram. Sigmoid output values were positively correlated with OSA severity, body mass index, and neck and waist circumference. CONCLUSIONS: Deep learning may help develop a model that classifies OSA using a cephalogram, which may be clinically useful in the appropriate context. The definition of ground truth was the main limitation of this study. CITATION: Jeong H-G, Kim T, Hong JE, et al. Automated deep neural network analysis of lateral cephalogram data can aid in detecting obstructive sleep apnea. J Clin Sleep Med. 2023;19(2):327-337.

Covert Brain Infarction as a Risk Factor for Stroke Recurrence in Patients With Atrial Fibrillation.

BACKGROUND: We aimed to evaluate covert brain infarction (CBI), frequently encountered during the diagnostic work-up of acute ischemic stroke, as a risk factor for stroke recurrence in patients with atrial fibrillation (AF). METHODS: For this prospective cohort study, from patients with acute ischemic stroke hospitalized at 14 centers between 2017 and 2019, we enrolled AF patients without history of stroke or transient ischemic attack and divided them into the CBI (+) and CBI (-) groups. The 2 groups were compared regarding the 1-year cumulative incidence of recurrent ischemic stroke and all-cause mortality using the Fine and Gray subdistribution hazard model with nonstroke death as a competing risk and the Cox frailty model, respectively. Each CBI lesion was also categorized into either embolic-appearing (EA) or non-EA pattern CBI. Adjusted hazard ratios and 95% CIs of any CBI, EA pattern CBI only, non-EA pattern CBI only, and both CBIs were estimated. RESULTS: Among 1383 first-ever stroke patients with AF, 578 patients (41.8%) had CBI. Of these 578 with CBI, EA pattern CBI only, non-EA pattern CBI only, and both CBIs were 61.8% (n=357), 21.8% (n=126), and 16.4% (n=95), respectively. The estimated 1-year cumulative incidence of recurrent ischemic stroke was 5.2% and 1.9% in the CBI (+) and CBI (-) groups, respectively (P=0.001 by Gray test). CBI increased the risk of recurrent ischemic stroke (adjusted hazard ratio [95% CI], 2.91 [1.44-5.88]) but did not the risk of all-cause mortality (1.32 [0.97-1.80]). The EA pattern CBI only and both CBIs elevated the risk of recurrent ischemic stroke (2.76 [1.32-5.77] and 5.39 [2.25-12.91], respectively), while the non-EA pattern only did not (1.44 [0.40-5.16]). CONCLUSIONS: Our study suggests that AF patients with CBI might have increased risk of recurrent stroke. CBI could be considered when estimating the stroke risk in patients with AF.

Strengthening deep-learning models for intracranial hemorrhage detection: strongly annotated computed tomography images and model ensembles.

Background and purpose: Multiple attempts at intracranial hemorrhage (ICH) detection using deep-learning techniques have been plagued by clinical failures. We aimed to compare the performance of a deep-learning algorithm for ICH detection trained on strongly and weakly annotated datasets, and to assess whether a weighted ensemble model that integrates separate models trained using datasets with different ICH improves performance. Methods: We used brain CT scans from the Radiological Society of North America (27,861 CT scans, 3,528 ICHs) and AI-Hub (53,045 CT scans, 7,013 ICHs) for training. DenseNet121, InceptionResNetV2, MobileNetV2, and VGG19 were trained on strongly and weakly annotated datasets and compared using independent external test datasets. We then developed a weighted ensemble model combining separate models trained on all ICH, subdural hemorrhage (SDH), subarachnoid hemorrhage (SAH), and small-lesion ICH cases. The final weighted ensemble model was compared to four well-known deep-learning models. After external testing, six neurologists reviewed 91 ICH cases difficult for AI and humans. Results: InceptionResNetV2, MobileNetV2, and VGG19 models outperformed when trained on strongly annotated datasets. A weighted ensemble model combining models trained on SDH, SAH, and small-lesion ICH had a higher AUC, compared with a model trained on all ICH cases only. This model outperformed four deep-learning models (AUC [95% C.I.]: Ensemble model, 0.953[0.938-0.965]; InceptionResNetV2, 0.852[0.828-0.873]; DenseNet121, 0.875[0.852-0.895]; VGG19, 0.796[0.770-0.821]; MobileNetV2, 0.650[0.620-0.680]; p < 0.0001). In addition, the case review showed that a better understanding and management of difficult cases may facilitate clinical use of ICH detection algorithms. Conclusion: We propose a weighted ensemble model for ICH detection, trained on large-scale, strongly annotated CT scans, as no model can capture all aspects of complex tasks.

Predicting intraoperative hypotension using deep learning with waveforms of arterial blood pressure, electroencephalogram, and electrocardiogram: Retrospective study.

To develop deep learning models for predicting Interoperative hypotension (IOH) using waveforms from arterial blood pressure (ABP), electrocardiogram (ECG), and electroencephalogram (EEG), and to determine whether combination ABP with EEG or CG improves model performance. Data were retrieved from VitalDB, a public data repository of vital signs taken during surgeries in 10 operating rooms at Seoul National University Hospital from January 6, 2005, to March 1, 2014. Retrospective data from 14,140 adult patients undergoing non-cardiac surgery with general anaesthesia were used. The predictive performances of models trained with different combinations of waveforms were evaluated and compared at time points at 3, 5, 10, 15 minutes before the event. The performance was calculated by area under the receiver operating characteristic (AUROC), area under the precision-recall curve (AUPRC), sensitivity and specificity. The model performance was better in the model using both ABP and EEG waveforms than in all other models at all time points (3, 5, 10, and 15 minutes before an event) Using high-fidelity ABP and EEG waveforms, the model predicted IOH with a AUROC and AUPRC of 0.935 [0.932 to 0.938] and 0.882 [0.876 to 0.887] at 5 minutes before an IOH event. The output of both ABP and EEG was more calibrated than that using other combinations or ABP alone. The results demonstrate that a predictive deep neural network can be trained using ABP, ECG, and EEG waveforms, and the combination of ABP and EEG improves model performance and calibration.

Pharmacological inhibition of STriatal-Enriched protein tyrosine Phosphatase by TC-2153 reduces hippocampal excitability and seizure propensity.

OBJECTIVE: STriatal-Enriched protein tyrosine Phosphatase (STEP) is a brain-specific tyrosine phosphatase. Membrane-bound STEP61 is the only isoform expressed in hippocampus and cortex. Genetic deletion of STEP enhances excitatory synaptic currents and long-term potentiation in the hippocampus. However, whether STEP61 affects seizure susceptibility is unclear. Here we investigated the effects of STEP inhibitor TC-2153 on seizure propensity in a murine model displaying kainic acid (KA)-induced status epilepticus and its effect on hippocampal excitability. METHODS: Adult male and female C57BL/6J mice received intraperitoneal injection of either vehicle (2.8% dimethylsulfoxide [DMSO] in saline) or TC-2153 (10 mg/kg) and then either saline or KA (30 mg/kg) 3 h later before being monitored for behavioral seizures. A subset of female mice was ovariectomized (OVX). Acute hippocampal slices from Thy1-GCaMP6s mice were treated with either DMSO or TC-2153 (10 µM) for 1 h, and then incubated in artificial cerebrospinal fluid (ACSF) and potassium chloride (15 mM) for 2 min prior to live calcium imaging. Pyramidal neurons in dissociated rat hippocampal culture (DIV 8-10) were pre-treated with DMSO or TC-2153 (10 µM) for 1 h before whole-cell patch-clamp recording. RESULTS: TC-2153 treatment significantly reduced KA-induced seizure severity, with greater trend seen in female mice. OVX abolished this TC-2153-induced decrease in seizure severity in female mice. TC-2153 application significantly decreased overall excitability of acute hippocampal slices from both sexes. Surprisingly, TC-2153 treatment hyperpolarized resting membrane potential and decreased firing rate, sag voltage, and hyperpolarization-induced current (Ih ) of cultured hippocampal pyramidal neurons. SIGNIFICANCE: This study is the first to demonstrate that pharmacological inhibition of STEP with TC-2153 decreases seizure severity and hippocampal activity in both sexes, and dampens hippocampal neuronal excitability and Ih . We propose that the antiseizure effects of TC-2153 are mediated by its unexpected action on suppressing neuronal intrinsic excitability.

Size-Related Differences in Computed Tomography Markers of Hematoma Expansion in Acute Intracerebral Hemorrhage.

BACKGROUND: Noncontrast computed tomography (NCCT) markers for hematoma expansion (HE) in intracerebral hemorrhage (ICH) are difficult to be found in small ICHs, of which can also expand. We aimed to investigate whether there were size-related differences in the prevalence of NCCT markers and their association with HE. METHODS: This retrospective analysis of prospectively collected stroke registry included 267 consecutive patients with ICH who underwent baseline NCCT within 12 h of onset. Qualitative NCCT markers, including heterogeneous density and irregular shape, were assessed. Hematoma density, defined as mean Hounsfield unit of hematoma, and hematoma volume were measured by semiautomated planimetry. Hematoma volume was categorized as small (≤ 10 ml) and large (> 10 ml). Associations of NCCT markers with HE were analyzed using multivariable logistic regression analyses. The model performances of NCCT markers and hematoma density were compared using receiver operating characteristic curves. RESULTS: Hematoma expansion occurred in 29.9% of small ICHs and 35.5% of large ICHs. Qualitative NCCT markers were less frequently observed in small ICHs. Heterogeneous density, irregular shape, and hematoma density were associated with HE in small ICH (adjusted odds ratios [95% confidence interval] 3.94 [1.50-10.81], 4.23 [1.73-10.81], and 0.72 [0.60-0.84], respectively), and hematoma density was also related to HE in large ICH (0.84 [0.73-0.97). The model performance was significantly improved in small ICHs when hematoma density was added to the baseline model (DeLong's test, p = 0.02). CONCLUSIONS: The prevalence of NCCT markers and their association with HE differed according to hematoma volume. Quantitative hematoma density was associated with HE, regardless of hematoma size.

Risk Factors for Intravenous Propacetamol-Induced Blood Pressure Drop in the Neurointensive Care Unit: A Retrospective Observational Study.

BACKGROUND: Intravenous propacetamol is commonly used to control fever and pain in neurocritically ill patients in whom oral administration is often difficult. However, several studies reported that intravenous propacetamol may cause blood pressure drop. Thus, we aimed to investigate the occurrence and risk factors for intravenous propacetamol-induced blood pressure drop in neurocritically ill patients. METHODS: This retrospective study included consecutive patients who were administered intravenous propacetamol in a neurointensive care unit at a single tertiary academic hospital between April 2013 and June 2020. The exact timing of intravenous propacetamol administration was collected from a database of the electronic barcode medication administration system. Blood pressure drop was defined as a systolic blood pressure below 90 mm Hg or a decrease by 30 mm Hg or more. Blood pressure, pulse rate, and body temperature were collected at baseline and within 2 h after intravenous propacetamol administration. The incidence of blood pressure drop was evaluated, and multivariable logistic regression analysis was performed to identify risk factors for blood pressure drop events. RESULTS: A total of 16,586 instances of intravenous propacetamol administration in 4916 patients were eligible for this study. Intravenous propacetamol resulted in a significant decrease in systolic blood pressure (baseline 131.1 ± 17.8 mm Hg; within 1 h 124.6 ± 17.3 mm Hg; between 1 and 2 h 123.4 ± 17.4 mm Hg; P < 0.01). The incidence of blood pressure drop events was 13.5% within 2 h after intravenous propacetamol. Older age, lower or higher baseline systolic blood pressure, fever, higher Acute Physiology and Chronic Health Evaluation II score, and concomitant administration of vasopressors/inotropes or analgesics/sedatives were significant factors associated with the occurrence of blood pressure drop events after intravenous propacetamol administration. CONCLUSIONS: Intravenous propacetamol can induce hemodynamic changes and blood pressure drop events in neurocritically ill patients. This study identified the risk factors for blood pressure drop events. On the basis of our results, judicious use of intravenous propacetamol is warranted for neurocritically ill patients with risk factors that make them more susceptible to hemodynamic changes.

Simultaneous Craniotomies for Multiple Intracranial Aneurysm Clippings-One-Stage Surgery with Multiple Craniotomies.

OBJECTIVE: The treatment of multiple intracranial aneurysms (MIAs) involves various modalities and sometimes requires staged operations. This study aimed to prove the efficacy and safety of one-stage multiple craniotomies (OSMC) for multiple cerebral aneurysms. METHODS: We retrospectively reviewed the medical records of the patients who underwent treatment for intracranial aneurysms between May 2003 and April 2020. The surgical results, complications, and lengths of hospital stay were compared between the patients who underwent OSMC and those who underwent multistage multiple craniotomies. RESULTS: The demographic characteristics of the OSMC and multistage multiple craniotomies groups (n = 82 and 43, respectively) were similar. There were no statistically significant differences between the 2 groups when the amount of blood transfused, complications, and surgical results were compared (P = n.s. for all); however, the operation time and hospitalization period (353.9 minutes vs. 490.3 minutes and 12.3 days vs. 21.8 days, respectively; P = 0.001 for both) were shorter in the OSMC group. The treatment cost (17,000 USD vs. 22,000 USD, P = 0.001) was lower in the OSMC group. CONCLUSIONS: OSMC for aneurysm clipping in patients with MIAs is a relatively safe and economical method. Furthermore, it has good clinical outcomes. This new surgical method is worthwhile in that it can be applied to patients who are afraid to undergo multiple surgeries, and we suggest that it is an efficient, low-cost option for the treatment of MIAs.

Heavily T2-Weighted Magnetic Resonance Myelography as a Safe Cerebrospinal Fluid Leakage Detection Modality for Nontraumatic Subdural Hematoma.

OBJECTIVE: Nontraumatic subdural hematoma (SDH) is a common disease, and spinal cerebrospinal fluid (CSF) leakage is a possible etiology of unknown significance, which is commonly investigated by several invasive studies. This study demonstrates that heavily T2-weighted magnetic resonance myelography (HT2W-MRM) is a safe and clinically effective imaging modality for detecting CSF leakage in patients with nontraumatic SDH. METHODS: All patients who underwent HT2W-MRM for nontraumatic SDH workup at our institution were searched and enrolled in this study. Several parameters were measured and analyzed, including patient demographic data, initial modified Rankin Scale (mRS) score upon presentation, SDH bilaterality, hematoma thickness upon presentation, CSF leakage sites, treatment modalities, follow-up hematoma thickness, and follow-up mRS score. RESULTS: Forty patients were identified, of which 22 (55.0%) had CSF leakage at various spinal locations. Five patients (12.5%) showed no change in mRS score, whereas the remaining (87.5%) showed decreases in follow-up mRS scores. In terms of the overall hematoma thickness, four patients (10.0%) showed increased thickness, two (5.0%) showed no change, 32 (80.0%) showed decreased thickness, and two (5.0%) did not undergo follow-up imaging for hematoma thickness measurement. CONCLUSION: HT2W-MRM is not only safe but also clinically effective as a primary diagnostic imaging modality to investigate CSF leakage in patients with nontraumatic SDH. Moreover, this study suggests that CSF leakage is a common etiology for nontraumatic SDH, which warrants changes in the diagnosis and treatment strategies.

Deep learning for anatomical interpretation of video bronchoscopy images.

Anesthesiologists commonly use video bronchoscopy to facilitate intubation or confirm the location of the endotracheal tube; however, depth and orientation in the bronchial tree can often be confused because anesthesiologists cannot trace the airway from the oropharynx when it is performed using an endotracheal tube. Moreover, the decubitus position is often used in certain surgeries. Although it occurs rarely, the misinterpretation of tube location can cause accidental extubation or endobronchial intubation, which can lead to hyperinflation. Thus, video bronchoscopy with a decision supporting system using artificial intelligence would be useful in the anesthesiologic process. In this study, we aimed to develop an artificial intelligence model robust to rotation and covering using video bronchoscopy images. We collected video bronchoscopic images from an institutional database. Collected images were automatically labeled by an optical character recognition engine as the carina and left/right main bronchus. Except 180 images for the evaluation dataset, 80% were randomly allocated to the training dataset. The remaining images were assigned to the validation and test datasets in a 7:3 ratio. Random image rotation and circular cropping were applied. Ten kinds of pretrained models with < 25 million parameters were trained on the training and validation datasets. The model showing the best prediction accuracy for the test dataset was selected as the final model. Six human experts reviewed the evaluation dataset for the inference of anatomical locations to compare its performance with that of the final model. In the experiments, 8688 images were prepared and assigned to the evaluation (180), training (6806), validation (1191), and test (511) datasets. The EfficientNetB1 model showed the highest accuracy (0.86) and was selected as the final model. For the evaluation dataset, the final model showed better performance (accuracy, 0.84) than almost all human experts (0.38, 0.44, 0.51, 0.68, and 0.63), and only the most-experienced pulmonologist showed performance comparable (0.82) with that of the final model. The performance of human experts was generally proportional to their experiences. The performance difference between anesthesiologists and pulmonologists was marked in discrimination of the right main bronchus. Using bronchoscopic images, our model could distinguish anatomical locations among the carina and both main bronchi under random rotation and covering. The performance was comparable with that of the most-experienced human expert. This model can be a basis for designing a clinical decision support system with video bronchoscopy.

Mediation effects of mean Hounsfield unit on relationship between hemoglobin and expansion of intracerebral hemorrhage.

Low hemoglobin levels are known to be associated with hematoma expansion (HE) and poor functional outcome in patients with intracerebral hemorrhage (ICH). However, it is not yet known whether low hemoglobin itself causes HE directly or is merely a confounder. Thus, we investigated the mediation effect of the mean Hounsfield unit (HU) of hematoma on the relationship between low hemoglobin and expansion of ICH. Overall, 232 consecutive patients with ICH who underwent non-contrast computed tomography (NCCT) within 12 h since onset were included. The mean HU and hematoma volume on NCCT were investigated using semi-automated planimetry. HE was defined as an increase in hematoma volume > 33% or 6 mL. The respective associations among the hemoglobin level, mean HU, and HE were analyzed using multivariable regression analysis, adjusting for age, sex, and known HE predictors. Mediation analysis was performed to examine the potential causal association among the three. HE occurred in 34.5% of patients; hemoglobin levels were inversely associated with HE occurrence (adjusted odds ratio, 0.90; p = 0.03). The mean HU of the hematoma was lower in patients with HE than in patients without HE (58.5 ± 3.3 vs. 56.8 ± 3.0; p < 0.01). Hemoglobin levels on admission were linearly related to the mean HU (adjusted ß, 0.33; p < 0.01) after adjusting for known HE predictors (time from onset to CT, antithrombotic use, hematoma volume). Causal mediation analysis showed a significant mediation effect of the mean HU on the association between hemoglobin levels and HE (p = 0.04). The proportion of indirect effect through the mean HU among the total effect was 19% (p = 0.05). The mediation effect became nonsignificant in the when the multivariable model was adjusted with additional covariates (baseline systolic blood pressure and hematoma location). The mean HU of the hematoma mediated the association between hemoglobin levels and HE occurrence. Therefore, the mean HU of the hematoma may be a potential marker of impaired hemostasis in patients with ICH.

Classification of cardioembolic stroke based on a deep neural network using chest radiographs.

BACKGROUND: Although chest radiographs have not been utilised well for classifying stroke subtypes, they could provide a plethora of information on cardioembolic stroke. This study aimed to develop a deep convolutional neural network that could diagnose cardioembolic stroke based on chest radiographs. METHODS: Overall, 4,064 chest radiographs of consecutive patients with acute ischaemic stroke were collected from a prospectively maintained stroke registry. Chest radiographs were randomly partitioned into training/validation (n = 3,255) and internal test (n = 809) datasets in an 8:2 ratio. A densely connected convolutional network (ASTRO-X) was trained to diagnose cardioembolic stroke based on chest radiographs. The performance of ASTRO-X was evaluated using the area under the receiver operating characteristic curve. Gradient-weighted class activation mapping was used to evaluate the region of focus of ASTRO-X. External testing was performed with 750 chest radiographs of patients with acute ischaemic stroke from 7 hospitals. FINDINGS: The areas under the receiver operating characteristic curve of ASTRO-X were 0.86 (95% confidence interval [CI], 0.83-0.89) and 0.82 (95% CI, 0.79-0.85) during the internal and multicentre external testing, respectively. The gradient-weighted class activation map demonstrated that ASTRO-X was focused on the area where the left atrium was located. Compared with cases predicted as non-cardioembolism by ASTRO-X, cases predicted as cardioembolism by ASTRO-X had higher left atrial volume index and lower left ventricular ejection fraction in echocardiography. INTERPRETATION: ASTRO-X, a deep neural network developed to diagnose cardioembolic stroke based on chest radiographs, demonstrated good classification performance and biological plausibility. FUNDING: Grant No. 14-2020-046 and 08-2016-051 from the Seoul National University Bundang Research Fund and NRF-2020M3E5D9079768 from the National Research Foundation of Korea.

High Chloride Burden and Clinical Outcomes in Critically Ill Patients With Large Hemispheric Infarction.

Background: In general, disease severity has been found to be associated with abnormal chloride levels in critically ill patients, but hyperchloremia is associated with mixed results regarding patient-centered clinical outcomes. We aimed to investigate the impact of maximum serum chloride concentration on the clinical outcomes of critically ill patients with large hemispheric infarction (LHI). Methods: We conducted a retrospective observational cohort study using prospective institutional neurocritical care registry data from 2013 to 2018. Patients with LHIs involving over two-thirds of middle cerebral artery territory, with or without infarction of other vascular territories, and a baseline National Institutes of Health Stroke Scale score of ≥13 were assessed. Those with a baseline creatinine clearance of <15 mL/min and required neurocritical care for <72 h were excluded. Primary outcome was in-hospital mortality. Secondary outcomes included 3-month mortality and acute kidney injury (AKI) occurrence. Outcomes were compared to different maximum serum chloride levels (5 mmol/L increases) during the entire hospitalization period using multivariable logistic regression analyses. Results: Of 90 patients, 20 (22.2%) died in-hospital. Patients who died in-hospital had significantly higher maximum serum chloride levels than did those who survived up to hospital discharge (139.7 ± 8.1 vs. 119.1 ± 10.4 mmol/L; p < 0.001). After adjusting for age, sex, and Glasgow coma scale score, each 5-mmol/L increase in maximum serum chloride concentration was independently associated with an increased risk of in-hospital mortality (adjusted odds ratio (aOR), 4.34; 95% confidence interval [CI], 1.98-9.50; p < 0.001). Maximum serum chloride level was also an independent risk factor for 3-month mortality (aOR, 1.99 [per 5 mmol/L increase]; 95% CI, 1.42-2.79; p < 0.001) and AKI occurrence (aOR, 1.57 [per 5 mmol/L increase]; 95% CI, 1.18-2.08; p = 0.002). Conclusions: High maximum serum chloride concentrations were associated with poor clinical outcomes in critically ill patients with LHI. This study highlights the importance of monitoring serum chloride levels and avoiding hyperchloremia in this patient population.

Hematoma Hounsfield units and expansion of intracerebral hemorrhage: A potential marker of hemostatic clot contraction.

BACKGROUND: Clot contraction reinforces hemostasis by providing an impermeable barrier and contractile force. Since computed tomography attenuation of intracerebral hemorrhage is largely determined by the density of red blood cells, clot contraction can be reflected in an increase of Hounsfield unit (HU) of hematoma. AIMS: We hypothesized that hematoma expansion is inversely associated with mean HU of intracerebral hemorrhage at presentation. METHODS: Eighty-nine consecutive spontaneous intracerebral hemorrhage patients with onset to first computed tomography within 24 h were included. Hematomas were segmented using semiautomated planimetry to measure the volume and mean HU. Hematoma expansion was defined as an increase in hematoma volume by over 33% or 6 mL. Multivariable logistic regression was performed for hematoma expansion. The discrimination power of mean HU for hematoma expansion was assessed using C-statistic. RESULTS: The computed tomography attenuation of hematoma at presentation was 57.5 ± 3.3 HU and the volume was 16.9 ± 23.2 mL. Hematoma expansion occurred in 37.1% of patients. The computed tomography attenuation of hematoma was lower in patients with hematoma expansion than with no expansion (55.7 ± 2.9 HU vs. 58.6 ± 3.1 HU, p-value < 0.01). Multivariable logistic regression revealed that the mean HU of hematoma was inversely associated with hematoma expansion (adjusted odds ratio, 0.64; 95% confidence interval, 0.51-0.80). The C-statistic of the model with four known predictors increased from 0.66 to 0.84 after incorporating mean HU (p-value < 0.01). CONCLUSIONS: Intracerebral hemorrhage with lower mean HU of hematoma at presentation is more likely to undergo hematoma expansion. This finding suggests the potential presence of clot contraction process that reinforces hemostasis in intracerebral hemorrhage.

Therapeutic Temperature Modulation for a Critically Ill Patient with COVID-19.

We report a rapidly deteriorating coronavirus disease 2019 (COVID-19) patient, a-58-year-old woman, with severe acute respiratory distress syndrome and shock with hyperpyrexia up to 41.8°C, probably due to the cytokine storm syndrome. Considering extracorporeal membrane oxygenation (ECMO) as the last resort, we applied therapeutic temperature modulation for management of hyperpyrexia. The patient demonstrated rapid improvement in oxygenation and shock after achieving normothermia, and fully recovered from COVID-19 three weeks later. Therapeutic temperature modulation may have successfully offloaded the failing cardiorespiratory system from metabolic cost and hyperinflammation induced by hyperpyrexia. The therapeutic temperature modulation can safely be applied in a specific group of patients with cytokine storm syndrome and hyperpyrexia, which may reduce the number of patients requiring ECMO in the global medical resource shortage.

Selection of Candidates for Endovascular Treatment: Characteristics According to Three Different Selection Methods.

BACKGROUND AND PURPOSE: To investigate the number and characteristics of patients eligible for endovascular treatment (EVT) determined using three different selection methods clinical-core mismatch, target mismatch, and collateral status. METHODS: Using the data of consecutive patients from two prospectively maintained registries of university medical centers, the number and characteristics of patients according to the three selection methods were investigated and their correlation was analyzed. Patients with anterior circulation stroke due to occlusion of the middle cerebral and/or internal carotid artery and a National Institute of Health Stroke Scale (NIHSS) score of ≥6 points, who arrived within 8 hours or between 6 and 12 hours of symptom onset and underwent magnetic resonance imaging prior to EVT, were included. Collateral status was assessed using magnetic resonance perfusion-derived collateral flow maps. RESULTS: Three hundred thirty-five patients were investigated; the proportions of patients who were eligible and ineligible for EVT in all three selection methods were both small (n=85, 25.4%; n=54, 16.1%, respectively). The intercorrelation among the three selection methods was low (κ=0.235). The baseline NIHSS score and onset-to-selection time interval were associated with the presence of clinical-core mismatch, while the penumbra/core volume ratio and onset-to-selection time interval were related to target mismatch; none of these variables were associated with collateral status. The infarct core volume was associated with favorable profiles in all three selection methods. CONCLUSION: s Although the application of individual selection methods resulted in favorable outcomes after EVT in clinical trials, there is a significant discrepancy in EVT eligibility depending on the selection method used.

Blood Pressure Drop and Penumbral Tissue Loss in Nonrecanalized Emergent Large Vessel Occlusion.

Background and Purpose- For patients with emergent large vessel occlusion who may not benefit from timely recanalization treatment, maintaining adequate cerebral perfusion to prevent penumbral tissue loss is a key therapeutic strategy. Cerebral perfusion should be proportional to systemic blood pressure (BP) due to the loss of autoregulation properties in ischemic brain tissue. We hypothesized that acute fluctuations in BP would lead to aggravated penumbral tissue loss in persistent large vessel occlusion. Methods- A total of 80 patients with persistent large vessel occlusion of internal carotid artery or middle cerebral artery admitted within 24 hours after onset, and with a baseline, National Institutes of Health Stroke Scale score ≥4-point were included. Baseline and follow-up (median 88 hours) magnetic resonance images were analyzed, and penumbra was defined as the Tmax>6 s region excluding baseline infarction. The hypoperfusion intensity ratio (Tmax>10 s/Tmax>6 s) was calculated within the penumbra. Penumbral tissue loss (%) was defined as the proportion of follow-up infarct in the penumbra. With serial BP measurements in the first 24 hours (median 29, interquartile range 26-35), BP and BP variability parameters, including BPdropmax (change from local maxima to minima), were calculated and compared. Generalized linear models were applied to examine the association between BP parameters and the penumbral tissue loss. Results- The median penumbral volume was 79.3 mL (interquartile range, 38.2-129.6) and median penumbral tissue loss was 36.7% (interquartile range, 12.0-56.1). In a multivariable analysis, systolic BP (SBP) SBPdropmax (ß±SE of fourth quartile, 17.82±6.58; P value, 0.01) and diastolic BP (DBP) DBPdropmax (ß±SE of fourth quartile, 14.04±6.38; P value, 0.01) were associated with increasing penumbral tissue loss, independently of age, baseline infarction and hypoperfusion intensity ratio. DBPincmax, SBPmax, DBPmax, SBPmax-min, DBPmax-min, and most of the DBP variability indices were associated with penumbral tissue loss. Conclusions- BP fluctuations, even a brief and drastic BP drop in the first 24 hours, significantly contributed to penumbral tissue loss irrespective of baseline hypoperfusion.