Citations

BACKGROUND: Hypertension, the leading cause of cardiovascular mortality, remains suboptimally controlled. Large language models (LLMs) could improve hypertension control by augmenting clinical decision-making, but their reliability for guideline-driven tasks is unverified. This study evaluated the accuracy and safety of hypertension management recommendations generated by 3 LLMs. METHODS: Fifty-one vignettes were constructed and submitted to the LLMs (GPT-4, Gemini, Medical Large Language Model [by Google; MedLM]) and a hypertension expert to generate the responses. Three blinded reviewers rated each response on a 4-point accuracy scale, a binary safety (safe/unsafe) scale, and attempted to identify the source (LLM versus expert) providing the response. RESULTS: GPT-4 had the highest accuracy (83%) and safety (86%) scores among LLMs but remained inferior to expert responses (92% accuracy, 93% safety). Gemini and MedLM performed significantly worse (accuracy: 64% and 35%; safety: 73% and 39%, respectively). GPT-4 generated the most guideline-concordant responses (46%) among the 3 LLMs (Gemini 35%, MedLM 14%) but was lower than expert responses (68%). Interrater reliability for accuracy ratings was higher for LLM-generated responses (GPT-4 [intraclass correlation coefficient, 0.30], Gemini [intraclass correlation coefficient, 0.61], and MedLM [intraclass correlation coefficient, 0.58]), with lower agreement for expert responses (intraclass correlation coefficient, 0.23). A similar pattern was observed for safety and source discrimination ratings. The agreement was strongest for safety assessments and weakest for source discrimination. CONCLUSIONS: Among the 3 tested LLMs, GPT-4 demonstrated closer agreement to expert decisions, thereby showing greater potential for supporting hypertension management. Despite their potential, current LLM versions are inferior to expert recommendations. Human-in-the-loop supervision remains essential when deploying LLMs for clinical decision-making.

Background: Mechanical support strategies used for patients in shock prior to heart transplantation (OHT) correlates with post-transplant outcomes. OHT timing in patients supported by Impella 5.5 remains an evolving decision point, influenced by patient stability and donor logistics. With the emergence of organ transport technologies, we sought to determine whether the choice of system was associated with variations in pre-transplant support duration. Method: A retrospective review between Jan 2020 and May 2025 was completed. Patients were categorized by donor transport technology: traditional cooler, Organ Care System (OCS), or SherpaPak. The primary outcome was duration of Impella support (days) prior to transplant. Group-wise comparisons were performed using Kruskal-Wallis testing. Result: A total of 129 patients were assessed across three transport methods: traditional cooler:12, OCS: 35, SherpaPak: 84. Median age at transplant was 58 (51-65) years with EF of 17% (15-22). Cold ischemic time differed significantly across transport technologies (Kruskal-Wallis H = 77.4, p < 0.001) and was longest in OCS cases (417 min), compared to SherpaPak (236 min) and traditional cooler (234 min). The median duration of Impella support prior to OHT was 27 days for organs transported via traditional cooler, 25 days for those using OCS, and 28.5 days for SherpaPak with near-significant difference by transport method (H=5.77, p=0.056). There was no significant difference of rejection within the first year after OHT. Conclusion: Donor transport technology choice may serve as a surrogate for broader transplant logistics and selection strategies as evidenced by significant differences in ischemic time and near-significant trends in Impella support duration.

Introduction: The ankle-brachial index (ABI) is the most common method for detecting peripheral arterial disease (PAD), but its sensitivity is limited, especially in patients with poorly compressible vessels. This study proposes a novel ultrasound-based approach for PAD diagnosis by analyzing calf muscle perfusion responses to pressure-cuff occlusion and exercise, interpreted using machine learning (ML) techniques. Methods: Under an IRB approved protocol and signed IRB-approved written consent, 48 participants, 43 symptomatic with abnormal ABI and 5 healthy volunteers were recruited for this study. High-frame-rate ultrasound B-mode image data (500 frames/s) were collected from 64 legs and categorized into two classes based on clinical diagnosis: Abnormal ABI (54 legs) and Healthy (10 legs). Legs with normal ABI from symptomatic patients were excluded to avoid diagnostic ambiguity, as these cases may lie within the diagnostic gray zone for PAD. Each leg underwent a 9-minute scanning protocol comprising 1 minute of baseline, 3 minutes of pressure cuff occlusion (using an automatic rapid inflation/deflation device), 2 minutes post-occlusion, 1 minute of plantar flexion exercise, and 2 minutes post-exercise. The ultrasound data were processed using singular-value decomposition (SVD) clutter filtering and noise equalization to generate Power-Doppler perfusion images at 2 frames per second. The PAD prediction pipeline involved data augmentation to address class imbalance and limited sample size, followed by histogram-based thresholding to extract blood pixels—intensity thresholding was avoided due to variability in pixel intensities. Blood pixel counts were calculated over time and used to train nine machine learning classification models (Figure 1). Results: The best-performing ML classification model achieved an accuracy of 77%, with the median accuracy across all models ranging between 70% and 72%. Notably, an ensemble model that combined Logistic Regression, Gradient Boosting, and Support Vector Machines using a majority voting strategy yielded the highest accuracy of 81% (Table 1). Conclusion: We proposed a novel ultrasound-based perfusion estimation method for diagnosing peripheral arterial disease (PAD), with data analyzed using ML approach. Although this is a feasibility study, the observed diagnostic accuracy exceeding 80% is promising. With further research and larger datasets, this approach has the potential to become a valuable tool for PAD.

Introduction: Heart failure (HF) affects interstitial lung structure and function. AI-assisted quantitative computed tomography (qCT) offers possibility of objective assesment of lung structure, allowing to detect the presence and the severity of HF. Hypothesis: Can automated qCT analysis discriminate HF from controls and identify tissue structures most linked to degree of congestion ? Methods: Subjects underwent right heart catheterization (RHC), native CT chest imaging (full inspiration/expiration) by Naeotom-α photon-counting CT scanner (Siemens), lab testing, echocardiography. Automated AI-based analysis (VIDA Diagnostics) was employed to evaluate lung volumes, densities and tissue texture. Image series were used in mediastinal and lung window reconstruction with slice thickness 0,8 mm and increments 0,5 mm. Results: 186 HF patients (85% HFrEF, NYHA 2.8±0.5, 58 y, 80% males, NTproBNP 4491 pg/ml, LVEF 26%) and 43 controls of similar sex and BMI were analyzed; both groups free of primary lung disease. In HF, 85 % had PH (mPAP>20mmHg). HF patients had comparable inspiratory lung volume, reduced air volume and increased lung tissue volume, compared to controls. Mean lung density (-782±50 vs -832±25 HU, p<0.0001), density variability and estimated lung fluid content (21.3 vs 16.7%, p<0.0001) were all higher in HF compared to controls. Across various density cut-offs, the most notable difference was at lung tissue density < -856 HU at inspiration. Lung tissue density corelated with mean PA pressure (r=0.49, p<0.0001), and less with pulmonary vascular resistance (r=0.21, p=0.006). Insp/exp scan analysis showed no difference in air trapping, but reduced lung tissue compliance (Jacobian strain: 1.8±0.4 vs 1.6±0.3, p=0.0005) in HF. Lung texture analysis showed increased ground-glass opacity (GGO) mass (48.8 g vs 9.1 g; p < 0.001), reticulation (42.2 g vs. 8.4 g; p < 0.001) and consolidation (0.5 g vs 0.2 g; p=0.0011) in HF compared to controls, with no difference in honeycombing. GGO mass correlated strongly with pulmonary capillary wedge pressure (PCWP) (r=0.48, p<0.001) and mitral regurgitation grade (r=0.28,<0.0001), but not with inflammation (hsCRP, IL6, WBC), smoking exposure or oncotic pressure. Conclusions: qCT distinguished HF patients from controls by increased lung tissue mass, reduced air volume, increased lung density and increased mass of GGO and reticulation. GGO mass in HF strongly reflects increased hydrostatic pressure, but not inflammation.

Background: Heart failure with preserved ejection fraction (HFpEF) involves myocardial aging, inflammation, and diastolic dysfunction. Monoamine oxidase-A (MAO-A), a mitochondrial enzyme generating reactive oxygen species (ROS), may contribute to cardiac injury, but its role in HFpEF-associated cardiomyocyte senescence is unclear. Hypothesis: Cardiac MAO-A promotes cardiomyocyte senescence via ROS-induced mitochondrial DNA (mtDNA) release and cGAS-STING signaling, exacerbating HFpEF. Methods: Plasma MAO-A activity, circulating mtDNA, and inflammatory cytokines were measured in HFpEF patients and controls. In mice, HFpEF was induced (high-fat diet+l-NAME, 15 wk) in cardiac-specific MAO-A knockout (Maoa^fl/fl; αMHC-Cre) and control littermates, with additional groups receiving MAO-A inhibitor (clorgyline) or STING inhibitor (H-151). Cardiac function, histology, ROS, mtDNA leakage, cGAS-STING activation, and senescence markers were evaluated. In vitro, H9c2 cardiomyoblasts were treated with angiotensin II+high glucose ± MAO-A siRNA. Mitochondrial ROS, cGAS-STING, senescence markers (SA-β-gal, Cdkn2a/Cdkn1a), and senescence-associated secretory phenotype (SASP) secretion were assessed. Conditioned media were used to evaluate fibroblast activation, macrophage polarization, and endothelial inflammation. Results: HFpEF patients had increased MAO-A activity and circulating mtDNA correlating with worse diastolic dysfunction. In mice, cardiomyocyte-specific MAO-A deletion or MAO-A/STING inhibition reduced ROS, mtDNA leakage, cGAS-STING activation, senescence markers, and improved diastolic function . In vitro, dual stimulation activated cGAS-STING, induced senescence and SASP secretion, effects attenuated by MAO-A silencing. Conditioned media from senescent H9c2 cells significantly induced cardiac fibroblast activation, as evidenced by increased α-smooth muscle actin; promoted macrophage M1 polarization, indicated by elevated inducible nitric oxide synthase; and triggered an endothelial inflammatory phenotype, characterized by higher vascular cell adhesion molecule-1 and intercellular adhesion molecule-1 expression together with enhanced monocyte adhesion. These effects were partially reversed when MAO-A was silenced in the donor H9c2 cells. Conclusions: MAO-A promotes ROS-dependent cardiomyocyte senescence and diastolic dysfunction in HFpEF via cGAS-STING signaling. Targeting this axis may offer therapeutic benefit in HFpEF.

Introduction: Complex congenital heart disease (CHD) patients with atrial arrhythmias have higher rates of morbidity and mortality compared to less complex CHD patients. This study aims to identify echocardiographic predictors of thromboembolism (TE) and major bleeding in patients with Tetralogy of Fallot (ToF) or Pulmonary Atresia with Intact Ventricular Septum (PA-IVS) and atrial fibrillation or flutter (AF). Methods: This cohort study included patients with ToF or PA-IVS who had a diagnosis of AF who underwent a TTE across three sites. Outcomes were incidence rate of major bleeding or TE, calculated using univariable and multivariable cox regression with interaction for anticoagulant choice and fractional area change (FAC). The HAS-BLED score was augmented with major predictors of bleeding, and receiver operator curves compared. Results: We included 300 patients (287 ToF [95.7%], 13 PA-IVS [4.3%]), mean age 45.6 years. Most received warfarin (227 [79.4%]), 40 (13%) used DOACs, and 5 (1.6%) used enoxaparin. TTE predictors of TE included large LVOT diameter ≥3cm (HR 8.92, p=0.002), small LVOT VTI (p=0.014), FAC <30% (HR 7.08, p=0.026) and moderate RV dysfunction (HR 10.75, p=0.026). Predictors of major bleeding included moderate tricuspid regurgitation (HR 2.40, p=0.022), right atrial pressure (1.12, p<0.001), severe RA enlargement (HR 2.30, p=0.048), RV mid diameter (HR 1.06, p=0.005), RV end diastolic area (HR 1.04, p=0.021), RV end systolic area (HR 1.05, p=0.016) elevated RVSP and FAC ≤45% (HR 0.37, p=0.016). Among warfarin users, impaired RV function (FAC ≤45%) was linked to reduced major bleeding events (HR 0.27, p=0.025). There was a nonsignificant trend towards higher bleeding in DOAC users with FAC >45% (HR 3.87, p=0.284), but significantly higher bleeding in DOAC users with FAC ≤45% (HR 12.41, p=0.008) vs. warfarin users with normal RV function (table 1). We developed an augmented HAS-BLED score: the standard score plus one point for moderate-to-severe pulmonic or tricuspid regurgitation and one for FAC ≤45%. This outperformed the standard HAS-BLED in predicting major bleeding (AUC 0.67 augmented vs. 0.57 un-augmented, p=0.008) (figure 1). Conclusion: Echocardiographic evidence of right heart disease is a strong predictor of major bleeding in patients with ToF or PA-IVS. FAC is a useful tool to determine excess major bleeding risk, particularly in DOAC users. Augmenting HAS-BLED with right heart parameters improves diagnostic accuracy.

Background and Aim: Iron deficiency (ID) is a common condition worldwide, affecting about one-third of the population, and is particularly prevalent in heart failure (HF) patients. In HF, ID worsens symptoms such as fatigue and shortness of breath, contributing to higher hospitalization rates and mortality. Intravenous (IV) iron therapies like Ferric Carboxymaltose (FCM) and Ferric Derisomaltose (FDI) have effectively enhanced symptoms and improved quality of life. This meta-analysis examines the safety and efficacy of these iron treatments. Methods: A comprehensive literature search was conducted using PubMed, Clinical Trial.gov, and Cochrane Library databases from inception to May 2024 to identify relevant studies. All statistical analyses were conducted using R statistical software version 4.1.2. Our protocol was registered with PROSPERO (CRD#CRD42023469246). Our study included Randomized Controlled Trials (RCTs). The primary outcomes of our study includes major adverse events such as Cardiovascular Mortality, All-Cause Death, heart failure hospitalization rate, KCCQ, LVEF, MLHFQ3, and Peak Oxygen Consumption (pVO2). Results: Twenty-four RCTs with 11,333 participants were analyzed. The pooled analysis indicated that All-cause mortality risk reduction was observed with FCM [RR=0.89; 95% CI: 0.75–1.04] and FDI [RR=0.95; 95% CI: 0.81–1.12]. FCM significantly reduced composite outcomes of hospitalization for HF or cardiovascular mortality [HR=0.74; 95% CI: 0.59–0.93], while FDI showed modest effects [HR=0.86; 95% CI: 0.56–1.26]. Kansas City Cardiomyopathy Questionnaire scores improved notably with FCM [MD=5.21; 95% CI: 2.22–8.19]. Left Ventricular Ejection Fraction improvement was highest with Iron Sucrose [MD=2.92; 95% CI: 1.61–4.21]. For the Minnesota Living with Heart Failure Questionnaire (MLHFQ), Iron Sucrose also showed the most improvement [MD = -18.00; 95% CI: -38.52 to 2.52], followed by FCM and FDI. In peak oxygen consumption (pVO2), FCM exhibited modest benefits [MD = 1.25; 95% CI: -2.34 to 4.84]. The six-minute walk test (6MWT) results identified Ferrous Sulphate as the most effective [MD = 53; 95% CI: -18.85 to 124.85], followed by FCM. Conclusion: This highlights Ferric Carboxymaltose as a promising iron therapy for heart failure patients with iron deficiency. However, clinical decisions should consider patient factors, tolerability, and cost-effectiveness. Further long-term studies are needed to optimize treatment and prevent hospitalization.

Background: Ultrasound-guided percutaneous fasciotomy and debridement (USPFD) is a minimally invasive treatment for patients with plantar fasciopathy refractory to conservative management. Previous studies have shown clinical improvement with USPFD, but pain relief and patient satisfaction have not been demonstrated beyond 2 years. Methods: This retrospective case series included patients who underwent USPFD at a quaternary academic medical center between April 2013 and January 2018, with refractory plantar fasciopathy and at least 4 years’ follow-up. Patient outcomes were assessed through electronic surveys and telephone interviews. Primary outcome measures included mean changes in visual analog scale pain (VAS-Pain) scores and patient satisfaction via Likert scale. Secondary measures included VAS-Pain with the first 5 steps, VAS-Pain during daily weightbearing, and procedural complications. Results: Seventy unique procedures were identified among 62 patients. Follow-up data were obtained from 51 of 62 patients (82%) who responded to surveys or phone calls. This corresponded to 58 of 70 procedures (83%) at a mean postprocedure follow-up of 6.6 ± 1.4 years. Patients had a mean duration of symptoms of 2.1 ± 3.0 years prior to the procedure. Baseline VAS-Pain was 6.8 ± 2.2 and final follow-up VAS-Pain was 0.7 ± 1.1. At the last follow-up, patient-reported VAS-Pain with first 5 steps was 0.9 ± 1.6 and VAS-Pain with weightbearing through the day was 1.3 ± 1.8. Overall, 91% (53/58) of the procedures were considered successful, where patients responded they were “satisfied” or “very satisfied” and 52 of 58 (90%) procedural responses indicated that based on the outcome, they would recommend the procedure to others. Final follow-up VAS-Pain, VAS-first 5 steps, and VAS-weightbearing, were all improved in the success vs failure groups ( P ≤ .002). No complications were reported. Conclusion: USPFD was associated with high patient satisfaction and durable pain relief for chronic plantar fasciopathy at an average follow-up of 6.6 ± 1.4 years.

Introduction: Knee injuries constitute a significant proportion of sports-related injuries. Osseous oedema can present in specific patterns with certain injuries Objective: This study aims to analyse the patterns of bone marrow contusion in the lateral tibial condyle (LTC) post knee trauma, categorising the condyle into anterior, middle, and posterior thirds, and to evaluate the prevalence of associated ligamentous and meniscal injuries. Materials and Methods: A cross-sectional study was conducted on 150 patients (92 males, 58 females) with an average age of 16.9 years, who underwent MRI within one week following knee trauma. The LTC was divided into anterior, middle, and posterior thirds for analysis. The bone marrow contusion patterns, along with associated ACL, MCL, PCL, LCL, medial meniscus, and lateral meniscus injuries, were documented. Results and Analysis: Bone marrow oedema in the LTC was observed as follows: anterior one third in 6, middle third in 13, and posterior third in 31 cases. The analysis revealed the following injury associations: Anterior one-third osseous oedema was associated with ACL injuries in two cases, MCL injuries in 3, PCL in 1, and medial meniscus injuries in 1. Middle one-third osseous oedema was associated with ACL injuries in 11 cases, MCL injuries in 3, medial meniscus injuries in two cases, and lateral meniscus injuries in 2. Posterior third osseous oedema was associated with ACL injuries in 28 cases, MCL injuries in 9, and medial meniscus injuries in four cases. Conclusion: The study highlights a clear association between the location of LTC bone marrow oedema and specific patterns of ligamentous and meniscal injuries. The posterior third, with the highest incidence of oedema, showed a strong correlation with ACL and MCL injuries, while anterior and middle oedema also demonstrated significant associations with ACL and MCL, and varied involvement of meniscal injuries. These insights contribute to a more comprehensive understanding of knee injury mechanisms, advocating for precise diagnostic strategies and potentially guiding more effective treatment protocols in radiological and orthopaedic practice.

Summary Chronic myelomonocytic leukaemia (CMML) is a morphological hybrid between myelodysplastic and myeloproliferative neoplasms and is characterized by clonal monocytosis. Myeloid neoplasm‐associated somatic mutations are present in >90% of patients with CMML but are not diagnostically specific but a subset are prognostically relevant. Current drug therapy in CMML is mostly palliative and not disease‐modifying. Allogeneic haematopoietic stem cell transplantation is a potentially curative treatment modality in CMML and its timing is determined by formal prognostic models including the recently developed BLAST and BLAST‐molecular risk models.