Citations

Abstract Objective The objective of this study is to characterize contemporary surgical management practices and factors influencing surgical decision‐making in the management of Lennox–Gastaut Syndrome (LGS) across pediatric epilepsy centers in the United States. Methods A 45‐item cross‐sectional survey was developed and distributed to Pediatric Epilepsy Research Consortium centers. Domains included institutional demographics, surgical evaluation timing and goals, presurgical workup, procedural utilization and outcomes, and barriers to surgery. Descriptive statistics and thematic analysis were performed. Results Thirty‐two pediatric epilepsy centers participated (38% response), the majority comprising pediatric epileptologists at Level 4 National Association of Epilepsy Centers. Most centers (66%) considered surgery after failure of 3–4 antiseizure medications. Presurgical workup typically included brain magnetic resonance imaging (MRI), long‐term electroencephalography (EEG) monitoring, and neuropsychological assessment, while use of functional imaging, magnetoencephalography, and intracranial EEG varied considerably. All centers offered vagus nerve stimulation and corpus callosotomy; 80%–90% offered resection, hemispherectomy, responsive neurostimulation (RNS); 60%–65% offered deep brain stimulation (DBS) and laser ablation. Among centers offering RNS, half targeted thalamic structures exclusively, and the remainder targeted cortical or combined cortical‐thalamic structures; the centromedian nucleus was the preferred thalamic target for RNS and DBS. Surgical decision‐making was influenced by patient characteristics, procedural considerations, and systemic factors. Seizure frequency, family preference, and multidisciplinary input were highly rated across all interventions. Etiology was critical for resection and hemispherectomy, while financial constraints, insurance barriers, and the need for future neuroimaging impacted neuromodulation selection. Scenario‐specific trends included less frequent consideration of DBS and RNS in children under five years, less use of hemispherectomy in older children, preference for corpus callosotomy in children with predominantly drop seizures, and procedural selection based on MRI findings. Significance Substantial variability exists in the surgical management of LGS. These findings highlight critical opportunities to harmonize practices, reduce disparities, and guide comparative effectiveness research to optimize outcomes. Plain Language Summary Lennox–Gastaut Syndrome (LGS) is a severe form of epilepsy that may require surgery when medications are not effective. This study found that surgical care for LGS varies widely across pediatric epilepsy centers in the United States, including when surgery is recommended, how patients are evaluated, and which procedures are available. Newer approaches such as brain stimulation are being used more often, although access remains inconsistent. These findings highlight the need for clearer treatment pathways to ensure that children with LGS have equitable access to the best possible care.

Abstract Radiological protocol selection is a critical but time-consuming step in clinical workflow, requiring radiologists to match patient indications with the appropriate MRI or CT protocol. Manual selection can be prone to delays or potential errors, and automated approaches must contend with substantial data imbalance, site-specific variation, and evolving nomenclature. We investigated whether a large language model (LLM) can support reliable protocol selection at scale and whether retrieval-augmented generation (RAG) offers operational advantages over direct fine-tuning. Using 498,228 patient reports collected across three Mayo Clinic sites (Arizona, Florida, and Rochester) spanning six radiological divisions, we trained site-specific Llama 3.2 3B models for use with and without retrieval augmentation. Division-scoped Facebook AI Similarity Search (FAISS) indexes constructed from procedure and diagnosis text were used to supply contextual evidence in the RAG framework. Both fine-tuned and RAG-augmented models achieved strong performance across sites, with F1 scores of 0.88–0.90. RAG matched or modestly trailed direct fine-tuning overall but delivered consistent gains in specific divisions (e.g., musculoskeletal imaging). Importantly, the RAG model introduced abstention behavior (Not Enough Information), which concentrated in linguistically diverse divisions and provided an interpretable signal of uncertainty. These findings suggest that RAG-based models are viable for division-scoped protocol selection and offer practical advantages. Retrieval indexes can be refreshed far more easily and with fewer resources than retraining LLMs, enabling continual adaptation to evolving clinical workflows. Future prospective deployment will evaluate real-time accuracy, agreement with practitioners, and the role of abstention as a safety mechanism in clinical decision support.

Background: Patients with end-stage heart failure-related cardiogenic shock (HF-CS) are conclusively associated with a poor health-related quality of life (HRQL). Axillary mechanical circulatory support (aMCS), such as the Impella 5.5, is increasingly used in this population and may improve HRQL during hospitalization by providing enhanced left ventricular unloading. We aimed to assess changes in HRQL between admission and two weeks after Impella 5.5 placement in patients with HF-CS, using the Kansas City Cardiomyopathy Questionnaire (KCCQ). Methods: We conducted a prospective longitudinal analysis on patients with the Impella 5.5 between May 2023 and July 2023. Participants completed the condensed KCCQ-12 at admission and again two weeks post-implantation. Changes in the scores were evaluated using the Wilcoxon signed-rank test. Results: Fifteen patients were enrolled. The median age was 59 years (50–63), and the median ejection fraction at implantation was 20% (15–30). On admission, most patients reported an overall HRQL of poor-to-fair (46.7%) according to the summary KCCQ-12 score. The median overall summary score increased significantly after Impella 5.5 support (50.52 vs. 28.13, p = 0.005). Symptom frequency (70.83 vs. 43.75, p = 0.009) and quality-of-life (50.00 vs. 12.50, p = 0.023) domains improved significantly, while physical limitation showed a positive trend and social limitation remained unchanged. These HRQL improvements occurred alongside a significant shift toward lower SCAI shock stages, marked increases in cardiac output and cardiac index, and no escalation in vasoactive-inotropic requirements. Conclusions: Impella 5.5 support in HF-CS patients was associated with early and clinically meaningful improvements in HRQL, particularly in symptom frequency and quality of life, during the critical pre-transplant or recovery period. These findings suggest that the Impella 5.5 may provide both physiological and patient-perceived benefits in this high-risk population.

We describe the first use of phage therapy in Canada for the treatment of a life-threatening periprosthetic joint infection (PJI), with successful outcome. PJI is a devastating complication of joint replacement surgery, with high morbidity and mortality. Our patient presented with early sepsis from a chronic recalcitrant multidrug-resistant (MDR) Staphylococcus epidermidis hip PJI which had repeatedly failed standard therapy. She had previously undergone 10 operations of the right hip, and only three weeks after completing a prolonged course of daptomycin following her most recent hip revision, she developed a draining sinus tract. Given the high burden of disease, inability to achieve surgical source control, and lack of antibiotic treatment options for long-term suppressive therapy, bacteriophage (phage) therapy was pursued. The patient underwent irrigation and debridement with complex flap reconstruction: intraoperative tissue cultures again yielded MDR S. epidermidis. We developed a novel phage therapy protocol for this patient, with twice daily, intra-articular and intravenous (7 × 109 PFU/dose) phage delivery over a planned 14-day course. Complete healing of the wound with cessation of drainage occurred within one month after treatment. A marked improvement in right hip pain and mobility occurred within three months after treatment. Twelve months following phage treatment, there is normalization of serum inflammatory markers with diminished pain, increased mobility, and no recurrent surgery. Our patient continues to improve and is currently living independently at home, with sustained clinical control of infection.

ABSTRACT Objectives Sarcomas are rare, mesenchymal tumors that require multimodality treatment provided by multidisciplinary teams. This study seeks to better understand sarcoma patient narratives and quality of life in relation to the social determinants of health (SDoH). Methods Eligible patients who were diagnosed with sarcoma, treated by our institution's orthopedic oncology team, and identified with at least one social risk factor were contacted about participating in a semi‐structured interview. SDoH information collected included (1) Economic Stability, (2) Access to Health Care, (3) Neighborhood & Physical Environment, (4) Social & Community Context, and (5) Education Access and Quality. Interviews were then recorded, de‐identified, transcribed, and coded by reviewers. The validated Framework Method for qualitative research was used to summarize themes. Descriptive statistics were used to summarize the percentage of participants that identified with each SDoH concern. Results 100% of patients ( N = 17) discussed how quality and access to healthcare affected their experience. 88.2% of participants expressed the importance of transportation. 29.4% of participants discussed how they thought their educational background affected their cancer care. Other common themes centered around social support, employment, financial security, and insurance. Conclusion Our research elicited ways in which SDoH affected the care experience of patients with sarcoma. The identified domains offer clinicians valuable information into non‐clinical factors, such as employment concerns or transportation issues, that may affect care for patients with sarcoma. Interdisciplinary care providers (social workers, psychologists, financial navigators) may also help promote open dialogues with patients about their social circumstances and actively facilitate connections to institutional resources or community support systems.

Abstract Objectives Large language models (LLMs) offer potential in natural language processing tasks in healthcare. Due to the need for high accuracy, understanding their limitations is essential. The purpose of this study was to evaluate the performance of LLMs in classifying radiology reports for the presence of pulmonary embolism (PE) under various conditions, including different prompt designs and data perturbations. Materials and Methods In this retrospective, institutional review board approved study, we evaluated 3 Google’s LLMs including Gemini-1.5-Pro, Gemini-1.5-Flash-001, and Gemini-1.5-Flash-002, in classifying 11 999 pulmonary CT angiography radiology reports for PE. Ground truth labels were determined by concordance between a computer vision-based PE detection (CVPED) algorithm and multiple LLM runs under various configurations. Discrepancies between algorithms’ classifications were aggregated and manually reviewed. We evaluated the effects of prompt design, data perturbations, and repeated analyses across geographic cloud regions. Performance metrics were calculated. Results Of 11 999 reports, 1296 (10.8%) were PE-positive. Accuracy across LLMs ranged between 0.953 and 0.996. The highest recall rate for a prompt modified after a review of the misclassified cases (up to 0.997). Few-shot prompting improved recall (up to 0.99), while chain-of-thought generally degraded performance. Gemini-1.5-Flash-002 demonstrated the highest robustness against data perturbations. Geographic cloud region variability was minimal for Gemini-1.5+-Pro, while the Flash models showed stable performance. Discussion and Conclusion LLMs demonstrated high performance in classifying radiology reports, though results varied with prompt design and data quality. These findings underscore the need for systematic evaluation and validation of LLMs for clinical applications, particularly in high-stakes scenarios.

ABSTRACT Rationale Longer duration of epilepsy before surgery is a predictor of poor outcome. While referral delays of surgical candidates are well documented, factors causing delay during the presurgical evaluation remain unclear and may vary depending on institutional characteristics. By benchmarking the duration of presurgical evaluation across multiple centers and identifying patient and evaluation characteristics contributing to duration, we can ascertain best practices and address modifiable contributors to reduce delays. Methods We queried the Pediatric Epilepsy Research Consortium Surgery Database, a prospective, observational multicenter study enrolling children 0–18 years at 27 US pediatric epilepsy centers, for all patients undergoing initial presurgical evaluation for drug‐resistant epilepsy (DRE). We included patients with completed evaluations and data on duration from initiation of presurgical evaluation to final surgical decision. We compared patient characteristics and evaluation components between those with long duration evaluations (> 75% quartile) and those with short evaluations (< 25% quartile). Akaike information criteria selection identified variables associated with longer duration. From these, we developed a logistic prediction model for evaluation duration, using a random 80/20 training/testing split of the entire cohort. The model was tested among institutions with ≥ 10 patients in the cohort to assess its accuracy in predicting long durations. Linear models for each site assessed each variable's impact on duration. Variables with < 10% of the patient population at each site were excluded. Beta values were compared to identify intra‐ and inter‐institution variability and to delineate institutions with the shortest added duration for each variable. Results Of 2318 patients undergoing surgical evaluation, 1655 (71%) from 23 sites had complete data. Median evaluation duration was 8 weeks (interquartile range 3–22); 453 (27%) were short‐duration evaluations and 414 (25%) were long‐duration evaluations. Multiple patient and evaluation characteristics were associated with duration (Table 1). Table 6 provides the average duration each variable contributes to evaluation by site, highlighting the shortest durations compared with other groups. Conclusions Duration of presurgical evaluation for DRE can be accurately modeled using multiple patient characteristics and testing strategies commonly employed in epilepsy surgery evaluations. This predictive model can not only estimate evaluation duration but also identify opportunities to improve systemic efficiency. Institution‐level modeling identifies specific program strengths, providing an opportunity to learn from successful processes. Subsequent research will focus on institutional process mapping to better understand systemic practices that lead to improved efficiencies, then sharing these processes across the consortium to shorten evaluation durations.

The NCCN Clinical Practice Guidelines (NCCN Guidelines) for Wilms Tumor (WT; nephroblastoma) cover strategies for the screening, diagnosis, and treatment of WT, which is the most frequent primary kidney tumor in children. WT can generally be separated into 2 histology types: favorable histology WT and anaplastic WT. Five-year survival is high for children with favorable histology WT who receive appropriate treatment; however, survival rates are much lower for patients who present with higher stage diffuse anaplastic WT. Treatment of WT can range from surgery alone to surgery plus intensive chemotherapy and radiation depending on whether the tumor is unilateral or bilateral, histology, and local stage. The goal of therapy is to maximize cure while minimizing long-term toxicities. The content featured in this issue covers the NCCN panel’s recommendations for overall management of both favorable histology WT and anaplastic WT.