Citations

Abstract Objective To evaluate whether postoperative intracranial pressure–lowering interventions reduce recurrence of lateral skull base spontaneous cerebrospinal fluid leaks after surgical repair. Data Sources Ovid MEDLINE, Embase, Cochrane CENTRAL, Web of Science Core Collection, and Scopus were searched from inception through April 7, 2025. Review Methods The review followed the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines and included studies of adults with lateral skull base spontaneous cerebrospinal fluid leaks undergoing primary repair where postoperative intracranial pressure–lowering interventions were specified. Pooled recurrence rates were calculated by intervention type and compared with cohorts that did not receive intracranial pressure–lowering therapy. Results Thirty‐six studies including 687 patients (737 cases) met inclusion criteria. Six comparative cohorts (34 treated, 155 untreated) showed no significant difference in recurrence (risk ratio 1.16, 95% confidence interval [CI]: 0.14–9.76; I 2 = 0%). Across 82 intervention‐only cases, recurrence occurred in three (pooled proportion 0.10, 95% CI: 0.04–0.26; I 2 = 0%). Among 650 control‐only cases, 45 recurred (pooled proportion 0.09, 95% CI: 0.07–0.12; I 2 = 0%). No heterogeneity was observed across pooled analyses. Conclusion Available evidence does not support universal postoperative intracranial pressure–lowering therapy after lateral skull base spontaneous cerebrospinal fluid leak repair. Limited observational data suggest acetazolamide may be associated with lower recurrence, whereas lumbar drainage has not demonstrated sustained benefit. Given the heterogeneity in patient selection and intervention practices, postoperative management should be individualized. Future prospective studies are needed to better define the role of targeted therapy in high‐risk patients.

Abstract Prostate adenocarcinomas (PRAD) can acquire resistance to androgen receptor signaling inhibitors through lineage transition to a cell state known as neuroendocrine prostate cancer (NEPC). Using a panel of isogenic PRAD and NEPC mouse tumoroids, we show that NEPC cells acquire new transcription factor (TF) dependencies that function in a previously undefined network. Through selective perturbation of each TF, we identify ASCL1 as a key regulator of NE lineage fate whereas MYCL functions downstream to drive NEPC growth/survival by recruitment of the TIP60/KAT5 acetyltransferase. Interestingly, while dependencies on specific TF family paralogs can vary across NEPC models, all show markedly enhanced dependency on TIP60. Moreover, the H2A.Z-acetyltransferase activity of the TIP60 complex (TIP60-C) is required for NEPC as well as the acetyl-reader BRD8, which is newly incorporated as a TIP60-C subunit with the NEPC transition. Targeted degradation studies in isogenic tumoroids reveal increased dependence on MYCL in NEPC relative to its paralog MYC in PRAD. In addition to a paralog switch (MYC to MYCL), the MYC pathway-addicted NE state is accompanied by a chaperone switch (from TIP60-C to SRCAP) for H2A.Z histone exchange and a coactivator switch (to TIP60) for MYC target gene expression. The NE-specific coupling of MYCL with TIP60 reveals a previously unappreciated opportunity to target MYC-driven NE diseases through pharmacological inhibition of TIP60.

Abstract Background Community-acquired pneumonia (CAP) is a significant public health concern associated with increased rates of hospital admissions and mortality. The purpose of the current study was to compare beta-lactam plus azithromycin or doxycycline vs fluoroquinolones in adults hospitalized with CAP. Methods We searched several databases from inception to 10 June 2024. Two reviewers independently screened, selected, and extracted data. Disagreements were resolved by consensus or a third reviewer. Study quality was assessed in duplicate with the Cochrane Risk of Bias 2 for randomized studies and the Newcastle-Ottawa Scale for nonrandomized studies. Meta-analysis was conducted by a random effect model when feasible. Network meta-analysis was performed to compare direct and indirect evidence. The GRADE approach (Grading of Recommendations Assessment, Development, and Evaluation) was followed to rate the certainty of evidence. Results Twenty studies were included (20 beta-lactam + azithromycin, 7 beta-lactam + doxycycline, 12 levofloxacin, and 4 moxifloxacin): 6 were randomized, enrolling 861 participants (mean age, 61.8 years; 43.2% women), and 14 were nonrandomized, enrolling 185 928 participants (mean age, 65.2 years; 42.9% women). Direct comparison and network meta-analysis results showed no statistically significant differences and likely equal effectiveness between beta-lactam + azithromycin and beta-lactam + doxycycline in terms of in-hospital mortality, need for intensive care unit admission or invasive mechanical ventilation, or safety outcomes including Clostridioides difficile infection and QT prolongation. Conclusions The current evidence demonstrated no likely difference in outcomes between azithromycin and doxycycline in adults hospitalized with CAP receiving beta-lactam. Head-to-head randomized clinical trials are needed to validate these results.

Objective To examine the potential errors of a general large language model (LLM) (ie, Claude 3.5 Sonnet) on data extraction from randomised controlled trials (RCTs). Design and setting An empirical study comparing Claude 3.5 Sonnet extractions against a human-performed verification dataset. The extraction tasks for Claude 3.5 Sonnet were based solely on original RCT portable document format (PDF) files. For PDFs that could not be directly extracted by Claude 3.5 Sonnet, optical character recognition was employed to convert them into text format before extraction. Participants A random sample of 664 trials was selected from a well-established trial bank and a final data pool was established based on rigorous manual cross-checking as a reference standard. Data sources PubMed, EMBASE, Scopus, Web of Science (all databases) and the Cochrane Central Register of Controlled Trials (CENTRAL) up to February 2023. Eligibility criteria for selecting studies RCTs on children involving medication and adverse events. Main outcome measures Claude 3.5 Sonnet was applied to extract the basic information (eg, trial design, population information and source of funding) and adverse outcomes (ie, name of adverse events, number of events). Claude 3.5 Sonnet outputs were compared against the final data pool and all errors were recorded. Results are presented as error rates and with 95% CI, estimated using a generalised linear mixed model. Results For the 664 trials, a total of 23 069 data cells were extracted via Claude 3.5 Sonnet, with 10 624 for basic information and 12 445 for adverse outcomes. The overall error rate for data extraction was 6.6% (95% CI 5.4% to 8.2%), with 5.7% (95% CI 5.2% to 6.1%) in basic information and 7.6% (95% CI 4.9% to 11.8%) in adverse outcomes. When stratified the 1542 total errors by error types, misallocation (assigning data to incorrect fields; 57.1%, 881/1542) and missed or omitted data (incomplete extraction of available data; 23.2%, 357/1542) accounted for the two most frequent errors, with misallocation occurring more in basic information (53.3%, 470/881), while missed or omitted data occurred more in adverse outcomes (96.1%, 343/357). Post hoc analysis examining the association between trial reporting quality (assessed using Consolidated Standards of Reporting Trials (CONSORT) 2025 and LLM data extraction error rates indicated that higher CONSORT adherence was associated with lower extraction error rates. Conclusions The data extraction error of Claude was relatively low, but it alerts LLM applications in evidence synthesis. Detailed checking for LLM outputs should be the primary consideration for evidence synthesisers.

Objectives This study aims to compare the reliability and accuracy of three large language models (LLMs) (Claude, Gemini and GPT) in assessing the risk of bias of nonrandomised studies using the ROBINS-I tool. Methods and analysis We conducted a secondary analysis of 171 nonrandomised studies previously assessed with Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) tool by two independent human review teams. Only studies with concordant human domain-level ratings were included. Each study was independently assessed twice by Claude, Gemini and Generative Pre-trained Transformer (GPT) using agent-based structured implementations of the ROBINS-I tool. Reliability (agreement between two runs of the same LLM) was evaluated using percent agreement and Gwet’s AC1. Accuracy (agreement with human reviewers) was assessed only for studies with consistent LLM ratings, using the same metrics. Results Claude demonstrated high reliability across all domains (79.5–98.0% agreement, AC1=0.729–0.975). Gemini showed moderate-to-high reliability (agreement 76.7–100%, AC1=0.680–1.0). GPT exhibited lower reliability overall, though domain-level agreement ranged from 70.9–95.6% (AC1=0.596–0.944). In terms of accuracy, Claude showed overall poor agreement with human reviewers (14.4–68.5% agreement; low AC1 values). Gemini demonstrated moderate-to-high accuracy in several domains, including deviations from intended interventions (79.6%, AC1=0.848) and measurement of outcomes (73.9%, AC1=0.702), with the highest overall agreement (40.0%, AC1=0.672). GPT showed variable accuracy, with the highest in measurement of outcomes (62.8%, AC1=0.571) and classification of interventions (57.8%, AC1=0.498), but poor performance in selection (14.3%, AC1 = −0.041) and overall agreement (23.0%, AC1=0.267). Conclusions Claude was internally consistent but poorly aligned with human reviewers. Gemini achieved both high reliability and moderate-to-high accuracy, whereas GPT had lower reliability and mixed accuracy. Current off-the-shelf LLMs cannot reliably perform ROBINS-I risk of bias assessments.

Abstract Background Atypical hemolytic uremic syndrome (aHUS) is a rare, complement-mediated disorder. Disease-associated variants have been found in complement factor H-related genes and complement regulatory genes. Particularly, homozygous deletions of CFHR1, CFHR3, or CFHR4 have been associated with aHUS. Generally, multiplex ligation-dependent probe amplification (MLPA) is used to detect copy number changes in the CFH and CFHR regions. We have developed a cost-effective droplet digital PCR (ddPCR) method for measuring copy number variations (CNVs) in these genes. Methods A total of 41 DNA samples were analyzed using both ddPCR and MLPA to assess large deletions in the CFHR1, CFHR3, CFHR4 regions. The ddPCR primers were designed to avoid rare single nucleotide polymorphisms to enhance hybridization efficiency and accurate identification of deletions within the CFHR regions. The MLPA probe mix (Probemix P236 CFH Region) targeting CFHR regions was purchased from MRC Holland. Results Multiple heterozygous deletions were observed in the CFHR1, CFHR3, and CFHR4 genes. Homozygous deletions of CFHR1 and CFHR3 were seen in 2 cases, while 2 other cases showed homozygous CFHR1 and heterozygous CFHR3 deletions. All 41 samples demonstrated 100% concordance in copy number results between ddPCR and MLPA, as well as consistent reproducibility in both inter- and intratechnologist evaluations. Conclusions Our findings demonstrate that ddPCR is a robust platform to identify large deletions in the CFHR regions. Future studies are needed to evaluate its ability to distinguish full vs partial deletions and to further establish its clinical diagnostic value.

ABSTRACT Patients with chronic thromboembolic pulmonary disease (CTEPD) without resting pulmonary hypertension (PH) frequently experience dyspnea which is not explained by stationary assessment and therefore requires cardiopulmonary exercise testing for further diagnostic insight. Ventilatory inefficiency in CTEPD is commonly identified; however, consensus definitions are lacking. Upright iCPET and careful definition of ventilatory inefficiency (i.e., lack of reduction in VD/VT with exercise) may allow optimal characterization of patients with chronic thromboembolic disease and specifically helps correctly identify patients with preload insufficiency who may otherwise be misclassified or undiagnosed.

Abstract One of the most significant unmet clinical challenges in paediatric oncology is the development of novel therapeutic strategies for recurrent medulloblastoma (R-MB). MYC-driven MBs are defined as classically cold tumours with a low incidence of infiltrating immune cells, resulting in a therapeutic challenge. The identification of cell-cell interactions between tumour and immune cells may provide insight into critical intercellular communications and manipulations occurring within the tumour immune microenvironment (TME). We hypothesised that the key cell-cell interactions in MYC-driven primary and recurrent MB may reveal dominant immune-suppressive mechanisms and uncover targetable therapeutic vulnerabilities. Paired primary-recurrent bulk RNA-sequencing data, confirmed myeloid cells as the most infiltrating immune cell type in group3-MB and group4-MB. Comprehensive spatial phenotypic and cell-cell communication analyses corroborated this discovery, validating an increased incidence of macrophages in the matched-recurrent tumours. Subsequently, we used innovative algorithms for 10X MB single-cell data to predict interactions between tumour-cell ligands and immune-cell receptors within the TME; macrophages emerged as the core immune-cells involved in interactions throughout the TMEs, with the most significant ligand-receptor interaction and inflammatory response between MIF and CD74. In-depth immunohistochemistry analyses of primary and recurrent group3 and group4 tumours, and exhaustive tissue microarrays demonstrated expression of both CD74 and MIF, with limited expression of CD74 within the brain. To investigate the therapeutic potential of CD74, we developed recurrent, immune competent MYC-driven medulloblastoma mouse models. Comprehensive deconvolution analysis confirmed the TME integrity of our models to mirror that of the human disease. Locoregional delivery and repeat dosing of a bioactive-CD74 peptide demonstrated complete tumour clearance in our immune-competent mouse models of primary and recurrent MB, demonstrating the significant therapeutic potential of targeting the CD74-MIF axis in MYC-driven primary and recurrent MB. Key cellular interactions and therapeutic vulnerabilities within the tumour microenvironment of MYC-driven medulloblastoma (MB) have been identified, highlighting the CD74-MIF axis as a target for next-generation immunotherapies.

ABSTRACT Objectives Systematic evidence reviews (SERs) produced by the U.S. Agency for Healthcare Research and Quality (AHRQ) Evidence‐based Practice Center (EPC) Program use contextual questions to provide context and background information on the topic. There is currently no standardized approach to address contextual questions in systematic reviews. This study explored the use of publicly available large language models (LLMs) in addressing contextual questions. Study Design Using a set of 20 published and 5 yet to be published SERs, we selected one contextual question per report and used it as a prompt to elicit answers from an LLM (ChatGPT, Bard, Claude, or Perplexity). Two independent reviewers rated the results using a priori established evaluation criteria ( https://osf.io/4k3cu/ ), comparing the response in the SER to LLM‐generated responses. The study was guided by six research questions addressing feasibility, validity of content, validity of structure, mistakes, congruence between responses, and incremental validity of using LLMs to address contextual questions. Results Using minimal prompt engineering produced relevant responses and documented the feasibility of LLM‐generated answers to contextual questions. Responses differed in content and format and are not reproducible (e.g., LLMs update regularly), but LLMs were able to produce articulate, clinically plausible, and well‐structured responses. We detected few factual errors, contradictions, and no instance of suspected bias, but citations supporting LLM‐generated responses could often not be produced or could not be verified (‘confabulations’). Congruence with human generated responses varied, with LLM‐generated responses providing more background on the topic and SERs providing more nuanced answers in response to the contextual question. Results regarding incremental validity were mixed and may depend on the tool. Conclusion LLMs are potentially helpful in addressing contextual questions in systematic reviews but human expertise remains essential for using the generated information in a meaningful way.

Abstract Background The ability of large language models (LLMs) to work collaboratively and screen studies in a systematic review (SR) is under-explored. Hence, we aimed to evaluate the effectiveness of LLMs in automating the process of screening in systematic reviews. Methods This is an observational study which included labeled data (title and abstracts) for five SRs. Originally, two reviewers screened the citations independently for eligibility. A third reviewer cross-checked each citation for quality assurance. GPT-4, Claude-3-Sonnet, and Gemini-Pro-1.0 were used using zero-shot chain-of-thought prompting. Collaborative approaches included (i): conflict resolution using benefit of the doubt, (ii) majority voting using an independent third LLM and (iii) conflict resolution using an informed third LLM. Performance was assessed using accuracy, precision for exclusion, and recall for inclusion. Work saved over samples (WSS) was computed to estimate the reduction in manual human effort. Results A total of 11300 articles were included in this study. The individual models, GPT-4, Claude-3-Sonnet, and Gemini-Pro-1.0 exhibited a high precision for exclusion, achieving 99.7%, 99.7%, and 99.2% and high recall for inclusion achieving 95.5%, 96.6% and 85.7%, respectively. However, the collaborative approach utilizing the two best-performing models (GPT-4 and Claude-3S) achieved an average precision of 99.9% and a recall of 98.5% (across all collaborative approaches). Furthermore, the proposed collaborative approach resulted in an average WSS of 63.5%, compared to the average WSS of 45.2% for individual models. Conversational LLM interactions showed a consistent pattern of results. Limitations This study was limited due to reliance on proprietary models, and evaluation on oncology datasets. Conclusion Evidence shows that collaborative LLMs enable efficient, high-performing screening in systematic reviews, supporting continuous evidence updates. Primary funding source NIH (U24CA265879-01-1) and Carolyn-Ann-Kennedy-Bacon Fund.