Francisco José Romeo

Elevated low-density lipoprotein cholesterol (LDL-C) and subclinical inflammation - measured with high-sensitivity C-reactive protein (hsCRP) - contribute to atherosclerosis progression. Despite medical therapy, high-risk patients may still have residual cholesterol risk, residual inflammatory risk (RIR), both, or neither. We aimed to study the impact of RIR in patients undergoing contemporary percutaneous coronary intervention (PCI). A comprehensive search of Pubmed, EMBASE, Cochrane and MEDLINE was performed up to December 2024. Only studies including patients with coronary artery disease undergoing PCI were considered. Inflammatory burden was evaluated with two hs-CRP measurements at baseline and follow-up (> 4 weeks apart). High RIR was defined as hsCRP ≥ 2 mg/L at 1-month follow-up. The risk ratio (RR) with a 95% confidence interval (CI) was computed using a random-effect model. We identified five studies enrolling 13,604 patients, including 5,833 with high RIR at 30 days post–PCI. At 12-month follow-up, persistent high RIR was associated with an increased risk of major adverse cardiovascular events (MACE) (random-effects RR = 1.64, 95% CI 1.33–2.03; I² = 80.2%). High RIR was also associated with increased all-cause mortality (random-effects RR = 3.25, 95% CI 2.49–4.25; I² = 67%), non-fatal myocardial infarction (random-effects RR = 1.46, 95% CI 1.00–2.12; I² = 73%), and non-fatal stroke (random-effects RR = 1.64, 95% CI 1.14–2.37; I² = 0%). Sensitivity analyses, including Baujat plots and leave-one-out analyses, identified a single study as a major contributor to heterogeneity; exclusion of this study substantially reduced heterogeneity without materially altering the direction or magnitude of the associations. In patients undergoing contemporary PCI, a high RIR was associated with a significantly higher risk of adverse cardiovascular outcomes. This data could help tailor lipid-lowering and anti-inflammatory therapies in this high-risk population.

Resumo Fundamento Modelos de linguagem baseados em inteligência artificial estão sendo cada vez mais utilizados para gerar materiais de educação em saúde para pacientes. No entanto, sua acurácia, completude e aderência às diretrizes clínicas permanecem incertas. Objetivos Comparar o ChatGPT-Mini e o ChatGPT-4.0 na geração de conteúdo educativo sobre hipertensão arterial sistêmica (HAS) quanto à acurácia, completude, qualidade estrutural utilizando o Ensuring Quality Information for Patients (EQIP), consistência das respostas e alinhamento com diretrizes estabelecidas. Métodos Um conjunto padronizado de 31 perguntas relacionadas à HAS foi submetido a ambos os modelos. As respostas foram avaliadas de forma independente por 10 clínicos cegos, utilizando um escore EQIP modificado, uma escala de acurácia de 5 pontos e uma escala de completude de 3 pontos. A consistência das respostas foi avaliada por meio do BERTScore. As comparações entre os modelos foram realizadas utilizando o teste de Wilcoxon para amostras independentes, bicaudal (p < 0,05). Os tamanhos de efeito foram apresentados como diferenças medianas de Hodges–Lehmann (HL) e delta de Cliff (δ), ambos com intervalos de confiança de 95% (IC 95%). A confiabilidade interavaliadores foi estimada por meio do coeficiente de correlação intraclasse (ICC; modelo de efeitos aleatórios em duas vias, concordância absoluta). Resultados As medidas de tendência central favoreceram o ChatGPT-4.0, embora as diferenças tenham sido pequenas. As medianas foram as seguintes: acurácia, 4,10 (3,70-4,20) versus 3,73 (3,60-4,05); completude, 1,26 (1,17-1,41) versus 1,10 (0,96-1,23); e escore EQIP total, 19,5 (18,0-25,0) versus 18,5 (16,0-23,0) para o ChatGPT-4.0 e o ChatGPT-Mini, respectivamente. As diferenças medianas de HL foram pequenas, com ICs de 95% cruzando zero (acurácia: +0,37, −0,25 a +0,50; completude: +0,16, −0,06 a +0,36; EQIP: +1,0, −1,0 a +6,0). Os valores de δ de Cliff foram consistentemente pequenos e positivos nos desfechos primários, indicando apenas modesta dominância estocástica do ChatGPT-4.0. A clareza de identificação tendeu a ser maior com o ChatGPT-4.0, enquanto a consistência das respostas, medida pelo BERTScore F1, foi geralmente maior para o ChatGPT-Mini (> 0,92 versus 0,885-0,932). A confiabilidade interavaliadores foi de boa a excelente em todas as medidas (ICC > 0,80). Conclusões O ChatGPT-4.0 demonstrou melhorias pequenas e não significativas em acurácia, completude e qualidade estrutural em comparação com o ChatGPT-Mini. Os tamanhos de efeito foram modestos, e todos os ICs de 95% incluíram zero. O ChatGPT-Mini produziu respostas mais consistentes. Esses achados ressaltam a importância de relatar rotineiramente tamanhos de efeito com ICs de 95% e apoiam o uso de métodos padronizados de avaliação e de estruturas de validação em tempo real para conteúdo médico educativo gerado por IA.

An abstract is unavailable. This article is available as a PDF only.

Masson W, et al. J Lipid Atheroscler. 2023 Sep;12(3):267-276. https://doi.org/10.12997/jla.2023.12.3.267

Coronary reperfusion therapy has played a pivotal role for reducing mortality and heart failure after acute myocardial infarction. Although several adjunctive approaches have been studied for reducing infarct size further, both ischemia–reperfusion injury and microvascular obstruction are still major contributors to both early and late clinical events after acute myocardial infarction. The progress in the field of cardioprotection has found several promising proof‐of‐concept preclinical studies. However, translation from bench to bedside has not been very successful. This comprehensive review discusses the importance of infarct size as a driver of clinical outcomes post‐acute myocardial infarction and summarizes recent novel device‐based approaches for infarct size reduction. Device‐based interventions including mechanical cardiac unloading, myocardial cooling, coronary sinus interventions, supersaturated oxygen therapy, and vagal stimulation are discussed. Many of these approaches can modify ischemic myocardial biology before reperfusion and offer unique opportunities to target ischemia–reperfusion injury.