Citations

Abstract Aims Lipoprotein(a) [Lp(a)] is a causal risk factor for cardiovascular events. However, the effect of Lp(a) on coronary plaque composition and high-risk plaque features has not been fully characterized. This study aimed to investigate the association between Lp(a) and coronary atherosclerotic plaque phenotype at the plaque level. Methods and results This study included 710 patients who underwent coronary computed tomography angiography (CCTA) and had Lp(a) measured between 2008 and 2024. CCTA scans were analysed with a previously validated artificial intelligence-based algorithm (AI-QCT, Cleerly Inc.). The association of Lp(a) with non-calcified and calcified plaque volumes and HRP features was evaluated at the plaque level using generalized estimating equation models adjusted for traditional cardiovascular risk factors. Among the 710 patients, the mean age was 56 ± 10 years and 379 (53%) were male. In total, 3642 plaques were identified. In the adjusted plaque-level analysis, elevated Lp(a) (≥150 nmol/L) was associated with an increase in non-calcified plaque volume per plaque (P = 0.009), but not with an increase in calcified plaque volume (P = 0.81). Furthermore, elevated Lp(a) (≥150 nmol/L) was strongly associated with the presence of high-risk plaque at the plaque level (adjusted odds ratio: 1.78, 95% CI: 1.25–2.54, P = 0.001), whereas low-density lipoprotein-cholesterol and high-sensitivity C-reactive protein were not (P > 0.05). Conclusion In this plaque-level CCTA study of 3642 individual plaques, elevated Lp(a) levels were independently associated with increased non-calcified plaque volume and with the presence of high-risk plaque, but not with calcified plaque volume. These findings elucidate the impact of Lp(a) on unstable plaque phenotypes, even in patients without prior cardiovascular events.

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 Cytomegalovirus (CMV) infection remains one of the most prevalent and consequential post solid organ transplant (SOT) infections. Treatment is often challenging, especially when dealing with refractory and resistant CMV infections. Methods We performed a retrospective multicenter cohort study of SOT recipients with clinically significant CMV infection (csCMVi) during 2010–2016. The primary outcome was early refractory CMV infection and secondary outcomes were drug resistance, CMV recurrence, and mortality. Analysis was done with Kaplan‐Meier, univariable logistic regression analysis, and multivariable Cox regression. Results We included 145 SOT recipients with csCMVi, majority were liver transplant (49%). Most common induction was an anti‐IL‐2 antibody (43.9%). The majority ( n = 82; 56.5%) were CMV D+/R‐ mismatch and presented asymptomatic infection (50.3%). After the initial 3 weeks of antiviral therapy, 13 (8.9%) patients had probable refractory csCMVi; most of them (10 [76.9%]) were CMV D+/R‐ mismatch. Longer time after transplant had lower risk (OR 0.68; CI 0.48–0.95, p = 0.029) while lower absolute lymphocyte count (ALC) had a higher risk of early refractory csCMVi (OR 1.54; CI 1.01–2.44, p = 0.048). Drug‐resistant csCMVi occurred in 7 patients (4.8%); a high initial CMV DNA level was associated with resistant csCMVi (OR 2.00; CI 1.08–3.93, p = 0.031). 25 patients (17.2%) experienced recurrent csCMVi within 6 months. 30 patients (20.7%) died; refractory, resistant or recurrent CMV infections were not associated with mortality. Conclusion Refractory csCMVi is associated with a low ALC and an earlier onset after transplant. Resistant csCMVi is associated with high initial CMV DNA levels. Recurrence of csCMVi is common and potentially associated with low ALC at the end of antiviral treatment.

The persistent shortage of allogenic donor organs has driven the exploration of xenotransplantation as a viable alternative for patients with end-stage liver disease. This review investigates the progress and challenges in liver xenotransplantation, focusing on the use of genetically engineered animals, particularly pigs, as potential organ donors. This narrative review discusses various sources of xenografts, highlighting their advantages and limitations, and provides a historical overview of liver xenotransplantation, encompassing both clinical and experimental experiences. Key obstacles, such as immunological rejection, coagulation dysfunction, and the risk of zoonotic infections, are analyzed, with emphasis on how these challenges have been addressed through genetic modifications and adjunctive therapies. Recent clinical trials are also reviewed, offering insights into the outcomes and lessons learned from these pioneering efforts. We also explored the regulatory, legal, and ethical considerations critical to advancing liver xenotransplantation into clinical practice, along with future perspectives on the field’s evolution. This comprehensive overview aims to provide a clear understanding of the current state of liver xenotransplantation and the strategies being developed to overcome existing barriers, paving the way for future clinical applications.