Background: Large language models (LLMs) have demonstrated impressive performances in various medical domains, prompting an exploration of their potential utility within the high-demand setting of emergency department (ED) triage. This study evaluated the triage proficiency of different LLMs and ChatGPT, an LLM-based chatbot, compared to professionally trained ED staff and untrained personnel. We further explored whether LLM responses could guide untrained staff in effective triage. Objective: This study aimed to assess the efficacy of LLMs and the associated product ChatGPT in ED triage compared to personnel of varying training status and to investigate if the models’ responses can enhance the triage proficiency of untrained personnel. Methods: A total of 124 anonymized case vignettes were triaged by untrained doctors; different versions of currently available LLMs; ChatGPT; and professionally trained raters, who subsequently agreed on a consensus set according to the Manchester Triage System (MTS). The prototypical vignettes were adapted from cases at a tertiary ED in Germany. The main outcome was the level of agreement between raters’ MTS level assignments, measured via quadratic-weighted Cohen κ. The extent of over- and undertriage was also determined. Notably, instances of ChatGPT were prompted using zero-shot approaches without extensive background information on the MTS. The tested LLMs included raw GPT-4, Llama 3 70B, Gemini 1.5, and Mixtral 8x7b. Results: GPT-4–based ChatGPT and untrained doctors showed substantial agreement with the consensus triage of professional raters (κ=mean 0.67, SD 0.037 and κ=mean 0.68, SD 0.056, respectively), significantly exceeding the performance of GPT-3.5–based ChatGPT (κ=mean 0.54, SD 0.024; P<.001). When untrained doctors used this LLM for second-opinion triage, there was a slight but statistically insignificant performance increase (κ=mean 0.70, SD 0.047; P=.97). Other tested LLMs performed similar to or worse than GPT-4–based ChatGPT or showed odd triaging behavior with the used parameters. LLMs and ChatGPT models tended toward overtriage, whereas untrained doctors undertriaged. Conclusions: While LLMs and the LLM-based product ChatGPT do not yet match professionally trained raters, their best models’ triage proficiency equals that of untrained ED doctors. In its current form, LLMs or ChatGPT thus did not demonstrate gold-standard performance in ED triage and, in the setting of this study, failed to significantly improve untrained doctors’ triage when used as decision support. Notable performance enhancements in newer LLM versions over older ones hint at future improvements with further technological development and specific training.

Background Myasthenia gravis (MG) is an autoimmune disorder affecting neuromuscular junctions, leading to fluctuating muscle weakness. While many patients respond well to standard immunosuppression, a substantial subgroup faces ongoing disease activity. Emerging treatments such as complement factor C5 inhibition (C5IT) and neonatal Fc receptor (FcRn) antagonism hold promise for these patients. However, the current landscape is hindered by a paucity of comparative data that is crucial for treatment decisions. Objective This study aims to compare the effectiveness and safety of C5IT and FcRn antagonists in a real-world setting. Methods A retrospective analysis of 153 MG patients from 8 German specialised MG centres receiving either C5IT (26 eculizumab, 80 ravulizumab) or efgartigimod (47 patients) was conducted. Propensity score matching (PSM) was employed to compare changes in MG-specific outcome parameters within the first 6 months after treatment initiation, along with safety profiles and concomitant MG therapy. Results Both treatment strategies led to rapid clinical improvements and substantial reductions in prednisolone doses. However, insufficient response was noted in 20%–49.1% of patients based on Quantitative MG and MG Activities of Daily Living (MG-ADL) scores. We did not identify any new safety concerns. After PSM, 40 patients remained in each group. In both cohorts, reductions in MG-ADL as prespecified primary study endpoint were comparable. Moreover, analyses of secondary outcome parameters demonstrated similar results for C5IT versus FcRn. Conclusion In contrast to current meta-analyses and indirect comparisons of clinical trial data, our real-world study demonstrates comparable efficacy and safety of C5IT and FcRn antagonism in MG.

In the recent past, a plethora of drugs have been approved for the treatment of multiple sclerosis (MS). These therapeutics are mainly confined to immunomodulatory or immunosuppressive strategies but do not sufficiently address remyelination and neuroprotection. However, several neuroregenerative agents have shown potential in pre-clinical research and entered Phase I to III clinical trials. Although none of these compounds have yet proceeded to approval, understanding the causes of failure can broaden our knowledge about neuroprotection and neuroregeneration in MS. Moreover, most of the investigated approaches are characterised by consistent mechanisms of action and proved convincing efficacy in animal studies. Therefore, learning from their failure will help us to enforce the translation of findings acquired in pre-clinical studies into clinical application. Here, we summarise trials on MS treatment published since 2015 that have either failed or were interrupted due to a lack of efficacy, adverse events, or for other reasons. We further outline the rationale underlying these drugs and analyse the background of failure to gather new insights into MS pathophysiology and optimise future study designs. For conciseness, this review focuses on agents promoting remyelination and medications with primarily neuroprotective properties or unconventional approaches. Failed clinical trials that pursue immunomodulation are presented in a separate article.

In this case series, we report the first off-label use of the CD19xCD3 T-cell engager blinatumomab in two patients with generalized myasthenia gravis (MG). Refractory MG remains a major therapeutic challenge, with patients experiencing severe disability and potentially life-threatening crises despite intensive immunotherapy. This study evaluates the clinical efficacy and safety of short-term blinatumomab treatment in two patients with severe, refractory generalized MG. Both individuals had been experiencing a persistent disease burden with myasthenic crises leading to severe disability, despite multimodal immunotherapy. Following treatment with blinatumomab, both patients showed rapid and sustained clinical improvements, reflected in significant reductions in MG-specific scores (MG-ADL, QMG, and revised MG Quality of Life-15), further patient-reported outcomes, digital activity markers, and gait analyses. Laboratory findings revealed persistent B-cell depletion in patient #1, whereas patient #2 demonstrated clinical improvement and autoantibody reduction despite B-cell repopulation by day 106. Both patients experienced grade 1 cytokine release syndrome during initial treatment phases, but no neurotoxicity or severe adverse events were observed. This report underscores the potential of CD19xCD3 T-cell engagers as a promising therapeutic approach in severe autoimmune neuroimmunological disorders, warranting further investigation in clinical trials and mechanistic studies.

Introduction: The efficacy of anti-CD20 antibodies has significantly contributed to advancing our understanding of disease pathogenesis and improved treatment outcomes in relapsing-remitting multiple sclerosis (RRMS). A comprehensive analysis of the peripheral immune cell profile, combined with prospective clinical characterization, of RRMS patients treated with ocrelizumab (OCR) or ofatumumab (OFA) was performed to further understand immune reconstitution following B-cell depletion. Methods: REBELLION-MS is a longitudinal analysis of RRMS patients treated with either OCR (n = 34) or OFA (n = 25). Analysis of B, T, natural killer (NK) and natural killer T (NKT) cells at baseline, month 1, and 12 was performed by multidimensional flow cytometry. Data were analyzed by conventional gating and unsupervised computational approaches. In parallel, different clinical parameters were longitudinally assessed. Twenty treatment-naïve age/sex-matched RRMS patients were included as the control cohort. Results: B-cell depletion by OCR and OFA resulted in significant reductions in CD20⁺ T and B cells as well as B-cell subsets, alongside an expansion of CD5⁺CD19⁺CD20⁻ B cells, while also elevating exhaustion markers (CTLA-4, PD-1, TIGIT, TIM-3) across T, B, NK, and NKT cells. Additionally, regulatory T-cell (T_REG) numbers increased, especially in OCR-treated patients, and reductions in double-negative (CD3⁺CD4⁻CD8⁻) T cells (DN T cells) were observed, with these DN T cells having higher CD20 expression compared to CD4 or CD8 positive T cells. These immune profile changes correlated with clinical parameters, suggesting pathophysiological relevance in RRMS. Conclusions: Our interim data add weight to the argumentation that the exhaustion/activation markers, notably TIGIT, may be relevant to the pathogenesis of MS. In addition, we identify a potentially interesting increase in the expression of CD5+ on B cells. Finally, we identified a population of double-negative T cells (KLRG1+HLADR+, in particular) that is associated with MS activity and decreased with CD20 depletion.