Kardiologia polska最新文献

The widespread adoption of leadless pacemakers has introduced new clinical and technical challenges related to device management at end of service and in the setting of device-related complications. While transvenous lead extraction is an established treatment strategy for conventional pacing systems, there is currently no standardized framework for transvenous leadless pacemaker extraction (TLLE). With increasing numbers of leadless pacemakers approaching end of service and expanding indications for leadless pacing, a structured approach to TLLE has become clinically relevant. This review proposes a systematic concept of TLLE, including definitions, clinical indications, procedural strategies, risk assessment, and institutional requirements. Indications for TLLE including device end-of-life, dislocation, interactions, infection, or the need for upgrade to a different device were summarized. Procedural strategies, primarily based on the extraction most frequently from the femoral approach, with the use of snare-based techniques, with the choice of other tools and access tailored to device type, dwell time, fixation mechanism, and degree of encapsulation, have been summarized. Literature findings reporting high success rates were presented, with potential complications, such as cardiac perforation, tamponade, tricuspid valve injury, embolization, and vascular access complications, being summarized as well. Patient selection, pre-procedural imaging have also been evaluated, and factors influencing decision-making, including balance of periprocedural risk against long-term implications of device abandonment have been presented.

Artificial intelligence (AI) can detect subtle patterns in data and therefore has the potential to improve patient care. This promise has inspired the development of numerous AI algorithms intended for use in cardiovascular medicine. The rapid evolution of AI technology has made it difficult for clinicians to critically evaluate AI-based clinical decision support tools. This review provides an overview of modern AI models and how they compare with traditional machine learning approaches. Also discussed are metrics for evaluating AI model performance and kay AI limitations, notably bias, generalizability, and lack of explainability.

Background: Left ventricular global longitudinal strain (LVGLS) detects myocardial injury. Reduced LVGLS is associated with worse outcomes in several cardiovascular conditions. Currently, there is a paucity of evidence regarding LVGLS in patients with coronary microvascular dysfunction (CMD).

Aims: To assess whether LVGLS is reduced in patients with CMD.

Methods: We included patients who underwent clinically indicated invasive assessment of coronary microvascular function with ejection fraction >50%. LVGLS measurement was performed using semi-automated speckle-tracking software and is presented in absolute value. Associations between LVGLS, echocardiographic parameters, and coronary microvascular function were assessed using univariate and multivariable analyses.

Results: 93 patients were included in the analysis (58% female; median age 66, interquartile range [IQR] 58-74) of which 49 (52.6%) had evidence of CMD. Median ejection fraction and LVGLS for the study cohort were 60% and 16.2%, respectively. Patients with CMD had worse LVGLS compared to those without CMD (15.0% [IQR 13-17] vs. 17.9% [IQR 15-21]; P = 0.04). Patients with structural CMD (14.4%; IQR 12.3-16.6) vs. functional CMD (16.5%; IQR 14.3-19.4), and no CMD (17.9%; IQR 15.2-21.3) had lower LVGLS (P for trend = 0.001). Reduced LVGLS (<17.5%) was associated with CMD (odds ratio, 3.08; 95% confidence interval, 1.28-7.43) and remained significant after multivariable adjustment (adjusted odds ratio, 3.61; 95% confidence interval, 1.35-9.68; P = 0.01).

Conclusions: Patients with CMD and preserved ejection function have lower LVGLS compared with patients without CMD, with worse LVGLS among patients with structural vs. functional CMD. Reduced LVGLS might emerge as an early non-invasive marker for CMD.