Citations

Background REM sleep behaviour disorder (RBD) is a prodrome of α -synucleinopathy, yet mechanistic pathways are unresolved. Narcolepsy type 1 with RBD (NT1-RBD) provides a human model of orexin deficiency. We tested whether REM motor semiology differs categorically between isolated RBD (iRBD) and NT1-RBD. Methods We retrospectively analyzed blinded video-polysomnographic scorings from 57 patients (iRBD n = 34; NT1-RBD n = 23). Across 857 REM events (iRBD 717; NT1-RBD 140), we classified topography (head/neck, trunk, upper, lower limbs), complexity (elementary vs. complex), content (scenic, violent, self-referential), vocal/orofacial features, spatial distribution and laterality using a pre-specified codebook. The patient was the primary unit of inference. Binary features used Fisher’s exact tests with Cohen’s h ; the per-patient complex-event proportion used Mann–Whitney and Cliff’s δ (bootstrap 95% CI). Robustness checks comprised Beta-Binomial posteriors (Beta [1,1]) and patient-label permutation tests (10,000 permutations). Results REM motor phenotypes diverged categorically. Lower-limb dominance occurred in 18/23 (78.3%) NT1-RBD vs. 7/34 (20.6%) iRBD (Fisher p < 0.0001; Cohen’s h ≈ +1.23), whereas upper-limb dominance occurred in 24/34 (70.6%) iRBD vs. 2/23 (8.7%) NT1-RBD ( p < 0.0001; h ≈ −1.40). Any complex event was present in 27/34 (79.4%) iRBD vs. 3/23 (13.0%) NT1-RBD ( p < 0.0001; h ≈ −1.46); violent enactments in 16/34 (47.1%) vs. 0/23 (0%) ( p < 0.0001; h ≈ −1.51). The per-patient complex-event proportion was higher in iRBD [median 0.21 (0.05–0.33)] than NT1-RBD [0.00 (0.00–0.00)] (Mann–Whitney p < 0.0001; Cliff’s δ = −0.665; 95% CI −0.853 to −0.441). Event-level summaries were concordant; permutation p -values were 0.0001 for upper-limb involvement and 0.0083 for complex behaviour. Conclusion iRBD and NT1-RBD exhibit qualitatively distinct REM motor phenotypes: upper-body-dominant, complex/scenic behaviours in iRBD versus elementary, predominantly bilateral lower-limb behaviours with notable trunk recruitment in NT1-RBD, supported by large effect sizes and convergent robustness checks. These findings motivate mechanistic studies of hypothalamic-brainstem-cortical integration in REM and suggest that semiological profiling may aid stratification in prodromal neurodegeneration.

Background Tumor-derived small extracellular vesicles were shown to contribute to immunosuppression in various cancer types, including head and neck squamous cell carcinoma. Dendritic cells, major regulators of specific anti-tumor immune response, are often impaired in their function in the context of cancer. Methods Monocyte-derived dendritic cells were incubated with isolated small extracellular vesicles during or after maturation. Cell surface markers and antigen processing machinery were assessed by flow cytometry. Vesicle-treated dendritic cells were tested for interleukin 12p70 production by ELISA, and differences in uptake of labeled vesicles were shown by confocal microscopy. Results Small extracellular vesicles derived from HPV(+) tumor cell lines increased maturation and activation of immature dendritic cells, priming them for more rapid phagocytosis. Immature dendritic cells co-incubated with vesicles had lower levels of antigen presenting machinery components and produced less IL12p70 than controls, while similarly co-incubated mature dendritic cells remained unaffected. Vesicles in plasma of patients with head and neck cancer inhibited functions of mature dendritic cells by downregulating expression of antigen presenting machinery components. Conclusion Our results uncover mechanisms through which tumor-derived small extracellular vesicles modulate dendritic cell functions locally and systemically. These mechanisms may also affect dendritic cell-based vaccine efficacy.

Abstract Background Large language models (LLMs) are increasingly used in medication-related tasks despite limited evidence supporting their accuracy and safety. LLMs are vulnerable to adversarial attacks and may confabulate in response to fabricated information in the prompt. Given the substantial risk of harm from medication errors, the purpose of this study was to establish a benchmark task to evaluate LLM performance in distinguishing fabricated from real medications. Methods Two datasets (brand and generic), each consisting of 250 medication lists were developed, each including 4-6 medications and one fabricated medication (specifically a Pokémon character), with complete dosing information provided. Six LLMs were evaluated (GPT-5-Chat, GPT-4o-mini, DrugGPT, Gemma-3-27B-IT, Llama-3.3-70B-Instruct, and Qwen3-32B). For each medication list, models were queried for dosing information and disease indication across three experimental conditions: standard decoding, standard decoding with mitigation prompting, and deterministic decoding (temperature = 0). Each experiment was performed in triplicate. Outputs were deemed confabulations if the LLM failed to recognize and reject the fabricated medication. The primary outcome was the rate of confabulations. Exact paired-permutation tests were used to compare confabulation rates across LLMs and prompting approaches. Results Confabulation rates under standard conditions ranged from 2.7% to 99.6%. DrugGPT demonstrated the lowest baseline confabulation rates (2.7-6.4%), whereas other models demonstrated substantially higher rates (47.2-99.6%). Mitigation prompting significantly reduced confabulation rates compared to standard and deterministic decoding across all tasks and datasets (p < 0.001): GPT-5-Chat achieved confabulation rates of 0% with mitigation prompting from baseline rates of 66.4-88.5%. Deterministic decoding did not substantially impact confabulation rates. Conclusions LLMs performing medication-related tasks are susceptible to confabulation in response to invalid or fabricated inputs, raising important safety concerns for clinical use. While mitigation prompting reduced confabulation rates, this study underscores the importance of robust safeguards when deploying LLMs in medication-related decision-making.

With most states lacking comprehensive legislation addressing overdose preparedness in schools, there is an urgent necessity for prompt and effective solutions that accommodate existing resource constraints. Arizona, the primary entry point for over 55% of US fentanyl trafficking, responded to this critical public health challenge by initiating the nation’s first unfunded grassroots collaboration between community and government sectors to enhance statewide school overdose prevention and preparedness. The School Training, Overdose Preparedness and Intelligence Task force (STOP-IT) included more than 60 volunteers from the health care, legal, education, local government and behavioral health fields and worked collaboratively with the Arizona Department of Education to develop a response. Within 6 months of inception, STOP-IT conducted a comprehensive school survey to identify barriers to naloxone accessibility, created evidence-based school policies, established a sustainable and complimentary naloxone supply chain for schools with an ordering platform, and implemented a statewide naloxone distribution plan. The amassed resources were organized into a Fentanyl and Overdose Response Toolkit, widely disseminated to more than 2000 public, private, and charter schools. Moreover, STOP-IT launched both a local and national public information campaign to effectively raise awareness about the opioid crisis in schools. Using various media channels such as webinars, podcasts, media broadcasts, newsletters, and peer-reviewed publications, the campaign successfully amplified its message and enhanced community engagement. The STOP-IT initiative emerged as an innovative collaboration between government and community leaders, united in their mission to combat the growing opioid crisis in schools.

Purpose: Cutaneous squamous cell carcinoma has a propensity for perineural invasion. Treating perineural spread has involved surgery and adjuvant chemoradiation, and more recently, immune checkpoint inhibitors. This study aims to evaluate motor and sensory functional outcomes in patients with cranial neuropathies from perineural spread of squamous cell carcinoma, particularly those undergoing immunotherapy. Methods: This was a multicenter case series and literature review. Results: Eighteen patients with cranial neuropathies from perineural spread had a mean age of 72.5 ± 9.7 years. Cranial nerves V and VII were most often involved. Treatments included radiotherapy (8/18; 44.4%), chemotherapy (7/18; 38.9%), and/or immunotherapy (11/18; 61.1%), with many receiving a combination of modalities (8/18; 44.4%). Of those receiving immunotherapy, 72.7% (8/11) demonstrated at least partial clinical and/or radiologic tumor response. Among those with functional outcomes documented, half had improvement in sensory/motor nerve function following immunotherapy (4/8; 50%). A literature review identified 55 articles describing 449 patients total with cranial neuropathies from squamous cell carcinoma perineural spread. Combinations of surgery, chemotherapy, and/or radiotherapy were employed, with immunotherapy used in 16 patients. From 5 patients on immunotherapy with documented functional outcomes, all demonstrated at least some improvement. Conclusion: Immunotherapy plays an evolving role in managing advanced squamous cell carcinoma. Functional improvements in cranial neuropathy were demonstrated in 50% of patients receiving treatment with immunotherapy at a mean of 32.6 months of follow-up. This may hold important implications for the timing of surgical intervention, particularly in the case of cranial nerves V and VII palsies.

Background: Vertebral body tethering (VBT) for adolescent idiopathic scoliosis (AIS) is an alternative to posterior fusion. There are limited prospective, multicenter data available on VBT following US Food and Drug Administration approval. We hypothesize that curve correction on first postoperative standing (first erect, FE) imaging is associated with higher rates of successful correction at final follow-up. Methods: All qualifying patients with AIS who underwent thoracic and lumbar VBT between 2019 and 2022 were prospectively enrolled from 9 institutions. Radiographic and clinical data were compared preoperatively, at FE, and at final follow-up with minimum of 2 years. Success was defined as major curve magnitude of ≤35° at final follow-up and no fusion surgery. Results: One hundred twenty-seven patients were enrolled (79.5% female), with mean follow-up 2.4 years. Mean age at surgery was 12.9 ± 1.4 years, most had bone age of Sanders 4 or lower (93/112, 83.0%). In average, 7.6 ± 1.7 levels were tethered. Mean preoperative major curve magnitude was 50 ± 8°, with mean initial correction at FE of 29 ± 8° (% correction, 39 ± 18%). At final follow-up, mean curve magnitude was maintained at 26 ± 11° (% correction, 45 ± 23%) despite 29% of tether breakage. Patients who had mean FE curve magnitude of ≤35° were 88% successful compared with only 60% in those with >35° on FE (p = 0.0021). Patients showed stable sagittal alignment across all timepoints. Scoliosis Research Society-22 scores improved significantly by 2 years (p < 0.0001). Conclusion: This was the first prospective, multicenter study to assess outcomes of VBT for patients with AIS. VBT shows promise, but optimal results may depend on careful patient selection and surgical technique. FE major curve magnitude of ≤35° was associated with 88% success rate compared with only 60% success for those with poor correction. Level of Evidence: Level II . See Instructions for Authors for a complete description of levels of evidence.