Faculty Bibliography

OBJECTIVE:Encephalitis is brain parenchyma inflammation, frequently resulting in seizures which worsens outcomes. Early anti-seizure medication could improve outcomes but requires identifying patients at greatest risk of acute seizures. The SEIZURE (SEIZUre Risk in Encephalitis) score was developed in UK cohorts to stratify patients by acute seizure risk. A 'basic score' used Glasgow Coma Scale (GCS), fever and age; the 'advanced score' added aetiology. This study aimed to evaluate the score internationally to determine its global applicability. DESIGN/METHODS:Patients were retrospectively analysed regionally, and by country, in this international evaluation study. Univariate analysis was conducted between patients who did and did not have inpatient seizures, followed by multivariable logistic regression, hierarchical clustering and analysis of the area under the receiver operating curves (AUROC) with 95% CIs. PARTICIPANTS AND SETTING/METHODS:2032 patients across 13 countries were identified, among whom 1324 were included in SEIZURE score calculations and 970 were included in regression modelling. The involved countries comprised 19 organisations spanning all WHO regions. OUTCOME MEASURES/METHODS:The primary outcome was measuring inpatient seizure rates. RESULTS:Autoantibody-associated encephalitis, low GCS and presenting with a seizure were frequently associated with inpatient seizures; fever showed no association. Globally, the score had limited discriminatory ability (basic AUROC 0.58 (95% CI 0.55 to 0.62), advanced AUROC 0.63 (95% CI 0.60 to 0.66)). The scoring system performed acceptably in western Europe, excluding Spain, with the best performance in Portugal (basic AUROC 0.82 (95% CI 0.69 to 0.94), advanced AUROC 0.83 (95% CI 0.72 to 0.95)). CONCLUSIONS:The SEIZURE score performed best in several countries in Western Europe but performed poorly elsewhere, partly due to differing and unknown aetiologies. In most regions, the score did not reach a threshold to be clinically useful. The Western European results could aid in designing clinical trials assessing primary anti-seizure prophylaxis in encephalitis following further prospective trials. Beyond Western Europe, there is a need for tailored, localised scoring systems and future large-scale prospective studies with optimised aetiological testing to accurately identify high-risk patients.

BACKGROUND AND PURPOSE/OBJECTIVE:Patients with intracranial stenosis-related large-vessel occlusion (ICAS-LVO) may experience better outcomes with stent placement compared with stand-alone mechanical thrombectomy (MT). This study evaluates the safety and clinical outcomes of self-expanding stents (SES) versus balloon-mounted stents (BMS) in patients with ICAS-LVO treated with MT and stent placement. MATERIALS AND METHODS/METHODS:This secondary analysis of the Registry of Emergent Large-Vessel Occlusion Due to Intracranial Stenosis, a multicenter observational study, included patients with ICAS-LVO from 25 stroke centers who underwent stent placement. Patients were stratified by stent type (SES or BMS). The primary end point was 90-day mRS = 0-2. Secondary outcomes included successful reperfusion, recurrent stroke, and infarct volume. Symptomatic intracranial hemorrhage was the primary safety outcome. Inverse probability-weighting was adjusted for confounders. RESULTS:= .001), particularly in patients without prestenting angioplasty (14% versus 1%). CONCLUSIONS:SES and BMS demonstrated comparable safety and clinical outcomes in patients with ICAS-LVO. However, SES were linked to higher rates of restenosis and recurrent strokes, potentially influenced by the absence of prestenting angioplasty. Further research is needed to refine stent-placement strategies in this population.

In eukaryotes, each ribosomal subunit includes a ribosomal protein (RP) that is encoded as a fusion protein with ubiquitin (Ub). In yeast, each Ub-RP fusion requires processing by deubiquitylating enzymes (DUBs) to generate ribosome assembly-competent RPs and contribute to the cellular Ub pool. However, how Ub-RP fusions are processed by DUBs in human cells remains unclear. Here, we discovered that Ub-RPs are substrates of the Ub-fusion degradation (UFD) pathway in human cells via lysine 29 and 48 (K29/K48)-specific ubiquitylation and proteasomal degradation. We identified a pool of DUBs that catalytically process Ub-RPs, as well as DUBs that physically occlude Ub-RP interaction with UFD pathway Ub E3 ligases to prevent their degradation in a non-catalytic manner. Our results suggest that DUBs both process and stabilize Ub-RPs, whereas the UFD pathway regulates levels of Ub-RPs that cannot be fully processed by DUBs to fine-tune protein homeostasis.

IMPORTANCE/UNASSIGNED:Open-label trials of antiseizure medications (ASMs) and devices suggest seizure reduction in focal treatment-resistant epilepsy (FTRE) may demonstrate treatment-related disease-modifying effects. Understanding FTRE trends can provide insight into treatment responses. OBJECTIVE/UNASSIGNED:To determine whether seizure frequency in FTRE improves over time. DESIGN, SETTING, AND PARTICIPANTS/UNASSIGNED:The Human Epilepsy Project 2 was a prospective, observational, multicenter study of patients with FTRE from May 2018 to September 2021 who were followed up for 18 to 36 months at 10 US-based comprehensive epilepsy centers. Analysis was performed from 2021 to 2024. Study data included seizure frequency, medication use, device use, surgeries tracked using daily electronic diaries, monthly check-ins, medical record review, and case report forms. Eligibility criteria included focal epilepsy diagnosis, age between 16 and 65 years, and failure of 4 or more ASMs (≥2 due to seizure control failure). Participants were recruited as a volunteer sample. EXPOSURES/UNASSIGNED:Participants were treated with multiple interventions at their physicians' discretion. MAIN OUTCOMES AND MEASURES/UNASSIGNED:The primary outcome was seizure frequency trends, evaluated by quantifying seizure freedom rates and frequency reductions. Medication and device treatment responses were assessed by tracking ASM and device changes. RESULTS/UNASSIGNED:Of 196 approached participants, 146 met eligibility criteria and were included in the study. Mean (SD) participant age was 40 (12) years, and epilepsy was diagnosed at a mean (SD) age of 19.8 (13.6) years. The cohort had 84 (57.5%) female participants. A total of 35 participants had implantable devices; 1 had epilepsy surgery during the study. Of 146 participants, 128 provided sufficient seizure data for analysis, and 2 were excluded as outliers. Seizure frequency was reduced in 86 participants (68.3%) during the second half of study participation compared to the first half. In the overall cohort, mean modeled monthly seizure frequency percentage reduction was 68.73% (95% CI, 52.92%-84.54%). From 0 to 12 months (cohort 1), mean modeled percentage reduction was 67.76% (95% CI, 19.42%-116.09%); for 12 to 24 months (cohort 2), 36.00% (95% CI, 9.27%-53.46%); and for longer than 24 months (cohort 3), 66.03% (95% CI, 48.25%-83.80%) (all P < .001). An ASM was added in 69 participants (54.7%), of whom 46 (66.7%) experienced seizure frequency reduction, including seizure freedom. Seizure trajectories in participants with devices did not significantly differ from those without devices. CONCLUSIONS AND RELEVANCE/UNASSIGNED:Findings from the HEP2 study imply that FTRE improves over time, ASM additions had low probability of achieving seizure freedom but contributed to seizure reduction, and device-treated participants exhibited similar seizure trajectories to those without devices. Whether improvements reflected the natural history of FTRE or active management remains unclear, but our findings suggest cautious interpretation of open-label studies positing disease-modifying effects and further research into FTRE treatment response.

Worldwide, over half of all individuals with dementia are undiagnosed. In the United States, racial, ethnic, and economic inequities mirror global findings, with higher rates of missed and delayed diagnosis and poorer diagnostic quality among minoritized and disadvantaged groups. For example, delayed diagnosis is more prevalent among people identifying as non-Hispanic Black or Latino than non-Hispanic White. Systematic efforts to improve detection can increase diagnosis rates; there is broad consensus that earlier detection and initiation of focused care and support services benefit both affected individuals and their loved ones. Systemic under-detection and its contributions to persistent population-level suffering underscore the importance of early detection of dementia as a key public health issue. Improving early detection calls for comprehensive, coordinated responses from local, regional, and national public health systems in partnership with health care delivery systems and community-based organizations. The Public Health Center of Excellence on Early Detection of Dementia (PHCOE on EDD), funded by the Centers for Disease Control and Prevention (CDC), is a national resource to promote understanding and implementation of evidence-based and evidence-informed public health strategy for early detection of dementia. We, together with the PHCOEs on Dementia Risk Reduction and Dementia Caregiving, and nearly four dozen state and local initiatives, seek to operationalize the priorities of the Building Our Largest Dementia Infrastructure for Alzheimer's Act and National Healthy Brain Initiative, established by federal legislation in 2018 and 2024. Our efforts support the CDC's mandate to build a national public health infrastructure for brain health and dementia.

Cardiovascular collapse and arrest in the periprocedural setting and intensive care unit differ from arrests in other contexts (such as out-of-hospital or hospital ward) because clinicians almost always witness the event, and the most likely precipitating cause may be known. In comparison to other settings, the response can be timelier and more focused on treating the underlying cause(s). Since many patients deteriorate over minutes to hours, clinicians can evaluate the patient expeditiously, generate a diagnosis, and initiate appropriate treatment more rapidly than in other arrest circumstances. This iteration of Perioperative Resuscitation and Life Support (PeRLS) employs Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) methodology to review the most recent evidence on preventing and managing cardiac arrest during the perioperative period. Furthermore, many of the recommendations and algorithms may also be applicable to areas outside the operating room, such as the intensive care unit and emergency room.

BACKGROUND:In many parts of the world including India, the prevalence of autoimmune inflammatory diseases such as neuromyelitis optica spectrum disorders (NMOSD), myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD), and multiple sclerosis (MS) is rising. A diagnosis is often delayed due to insufficient diagnostic tools. Machine learning (ML) models have accurately differentiated eyes of patients with MS from those of healthy controls (HCs) using optical coherence tomography (OCT)-based retinal images. Examining OCT characteristics may allow for early differentiation of these conditions. The objective of this study was to determine feasibility of ML analyses to distinguish between patients with different autoimmune inflammatory diseases, other ocular diseases, and HCs based on OCT measurements of the peripapillary retinal nerve fiber layer (pRNFL), ganglion cell-inner plexiform layer (GCIPL), and inner nuclear layers (INLs). METHODS:Eyes of people with MS (n = 99 patients), NMOSD (n = 40), MOGAD (n = 74), other ocular diseases (OTHER, n = 16), and HCs (n = 54) from the Mangalore Demyelinating Disease Registry were included. Support vector machine (SVM) classification models incorporating age, pRNFL, GCIPL, and INL were performed. Data were split into training (70%) and testing (30%) data and accounted for within-patient correlations. Cross-validation was used in training to choose the best parameters for the SVM model. Accuracy and area under receiver operating characteristic curves (AUROCs) were used to assess model performance. RESULTS:The SVM models distinguished between eyes of patients with each condition (i.e., MOGAD vs NMOSD, NMOSD vs HC, MS vs OTHER, etc) with strong discriminatory power demonstrated from the AUROCs for these comparisons ranging from 0.81 to 1.00. These models also performed with moderate to high accuracy, ranging from 0.66 to 0.81, with the exception of the MS vs NMOSD comparison, which had an accuracy of 0.53. CONCLUSIONS:ML models are useful for distinguishing between autoimmune inflammatory diseases and for distinguishing these from HCs and other ocular diseases based on OCT measures. This study lays the groundwork for future deep learning studies that use analyses of raw OCT images for identifying eyes of patients with such disorders and other etiologies of optic neuropathy.

INTRODUCTION/BACKGROUND:Neurologic complications, including seizures, are common in pediatric patients undergoing heart surgery, especially those requiring postoperative extracorporeal membrane oxygenation (ECMO), requiring prompt, vigilant postoperative monitoring. Prolonged EEG monitoring in critically ill children presents a risk of scalp/pressure injuries. The skin's sensitivity to microcirculatory changes can also provide valuable insights into the patient's overall tissue perfusion, making it a critical component in the management of these vulnerable patients. METHODS:We initiated a quality improvement (QI) project to assess and reduce scalp injuries related to prolonged EEG monitoring in critically ill neonates and infants. The project involved reviewing baseline data, which included 2336 inpatient video EEGs performed from January 2022 to December 2024, and implementing interventions to improve skin safety during electrode placement, while incorporating best practices from ACNS and ASET guidelines. RESULTS:Five critically ill infants developed deep tissue injuries (DTIs) related to EEG electrodes, with most injuries occurring over the occipital region. The frequency of scalp injuries decreased from 0.30% in 2022 to 0% in 2024 after implementing the QI protocol, and was observed in conditions with known hypoperfusion. DISCUSSION/CONCLUSIONS:Electrode-related skin injuries are a common complication of prolonged EEG monitoring, particularly in critically ill pediatric patients. Our findings suggest that adherence to expert guidelines and tailored skin care protocols focused on skin preparation, electrode application, and monitoring parameters can reduce the risk of electrode-related skin injuries. Further research is needed to refine safety protocols and address the unique skin care challenges faced by this high-risk population.

RATIONALE AND OBJECTIVES/OBJECTIVE:To evaluate the performance, stability, and decision-making behavior of large language models (LLMs) for title and abstract screening for radiology systematic reviews, with attention to prompt framing, confidence calibration, and model robustness under disagreement. MATERIALS AND METHODS/METHODS:We compared five LLMs (GPT-4o, GPT-4o mini, Gemini 1.5 Pro, Gemini 2.0 Flash, Llama 3.3 70B) on two imaging-focused systematic reviews (n = 5438 and n = 267 abstracts) using binary and ternary classification tasks, confidence scoring, and reclassification of true and synthetic disagreements. Disagreements were framed as either "LLM vs human" or "human vs human." We also piloted autonomous PubMed retrieval using OpenAI and Gemini Deep Research tools. RESULTS:LLMs achieved high specificity and variable sensitivity across reviews and tasks, with F1 scores ranging from 0.389 to 0.854. Ternary classification showed low abstention rates (<5%) and modest sensitivity gains. Confidence scores were significantly higher for correct predictions. In disagreement tasks, models more often selected the human label when disagreements were framed as "LLM vs human," consistent with authority bias. GPT-4o showed greater resistance to this effect, while others were more prone to defer to perceived human input. In the autonomous search task, OpenAI achieved moderate recall and high precision; Gemini's recall was poor but precision remained high. CONCLUSION/CONCLUSIONS:LLMs hold promise for systematic review screening tasks but require careful prompt design and circumspect human-in-the-loop oversight to ensure robust performance.