Background The heart team (HT) is considered the gold standard for determining appropriate complex coronary artery disease (CAD) treatment. However, the role of artificial intelligence (AI) as a clinical decision support system in this area is unknown. We sought to assess the concordance between large language model (LLM)-generated revascularization recommendations and multidisciplinary HT decisions for patients with complex CAD, and to identify factors influencing agreement. Methods We evaluated 546 patients with complex CAD assessed by an HT in 2019 to 2024. Each case was input into 2 commercially available LLMs with 2 prompt formats: (1) unstructured clinical narrative input (CNI), and (2) structured clinical proforma (SCP). Model recommendations and HT decisions were compared. Primary outcomes included concordance (κ statistic) and association with major adverse cardiac events. Results Concordance depended on the format: treatment recommendations achieved 75% concordance (κ = 0.40) with CNI but only 35% (κ = −0.03) with SCP inputs. Near-identical results were found between models. Discordant HT-AI cases were significantly associated with poorer clinical outcomes, including a >3-fold elevation in myocardial infarction (P < .001). Guideline augmentation did not improve model performance. High frequencies of advanced age, frailty, dialysis, and anemia in discordant cases suggested model limitations in high-risk cohorts. Conclusions LLMs showed moderate agreement with expert clinical decisions, but only when they received detailed, context-rich CNI. Discordance correlated with poorer outcomes and might potentially serve as a risk signal. These findings underscore the need for format-aware AI integration, rigorous validation, and careful oversight before clinical deployment.
Mitral transcatheter edge-to-edge repair (M-TEER) in the setting of prior transcatheter aortic valve replacement (TAVR) and severe mitral annular calcification (MAC) can pose procedural challenges. We describe the case of an 82-year-old woman with prior TAVR (29 mm EVOLUT FX+; Medtronic), P2 flail, and heavy circumferential MAC underwent M-TEER with the PASCAL Ace device (Edwards Lifesciences). This case illustrates that M-TEER can be performed safely and effectively in anatomies complicated by severe MAC and prior TAVR with careful preprocedural planning, intraprocedural imaging, and operator experience.
Background Transcatheter aortic valve replacement (TAVR), a standard therapy for aortic stenosis, is preferred for older or high-risk patients. Various complications, including heart failure (HF), resulting in increased mortality, have been reported in the literature. Sodium-glucose cotransporter-2 inhibitors (SGLT2i) are recommended in HF, irrespective of ejection fraction; however, their role in reducing post-TAVR cardiovascular complications remains underexplored. We conducted a retrospective study to evaluate cardiovascular outcomes associated with the use of SGLT2i in post-TAVR patients. Methods This retrospective cohort was based on a global database of electronic health records of more than 160 million patients. After propensity-score matching, the study population included 9930 post-TAVR diabetic patients divided into 2 cohorts based on their exposure to SGLT2i. Exposure to SGLT2i was defined as exposure after 1 day of TAVR. Results In this study of 9930 post-TAVR diabetic patients, the primary outcome of major adverse cardiovascular events and secondary outcomes, ie, cardiovascular mortality, ischemic stroke, and HF exacerbation, were significantly reduced in SGLT2i-exposed post-TAVR patients at the 1-month and 1-year follow-up. Hemorrhagic stroke was comparable at the 1-month follow-up; however, it was significantly lower in SGLT2i-exposed cohorts at the 1-year follow-up. The risk of new-onset myocardial infarction was significantly reduced at 1 month; however, it was comparable between the 2 cohorts at the 1-year follow-up. Our study also demonstrated a significant reduction in the risk of end-stage renal disease in the SGLT2i-exposed cohort at the 1-year follow-up. The periprocedural risk of acute kidney injury (within 7 days of TAVR) as well as at the 1-month and 1-year follow-up was significantly lower in the SGLT2i-exposed cohort. Conclusions In this study, post-TAVR SGLT2i use in diabetic patients was associated with reduced cardiovascular morbidity and mortality for up to 1 year.
Background Aortic stenosis (AS) remains a leading cause of cardiovascular death, yet temporal mortality trends specifically among older adults remain poorly characterized. This study analyzed trends and disparities in AS-related mortality among older adults in the United States from 1999 to 2024. Methods Using the CDC WONDER database, we analyzed AS-related deaths among adults aged ≥65 years. Age-adjusted mortality rates (AAMR) and average annual percent change (AAPC) were calculated across demographic and geographic subgroups. Results A total of 637,178 deaths were recorded. The overall AAMR for AS-related mortality decreased significantly from 1999 (55.74) to 2024 (52.46) (AAPC, –0.23; P =.004). Although rates rose from 2018 to 2021, a significant decline followed through 2024 (annual percent change, –2.40). Men had consistently higher AAMR than women; mortality remained stable in men (AAPC, –0.06; P = .33), but declined among women (AAPC, –0.35; P =.002). AS-related mortality decreased among patients aged 75 to 84 years, but increased in those ≥85 years (AAPC, 0.28), who had the highest AAMR throughout. Mortality declined in all census regions except for the Midwest, reaching the highest AAMR in 2024 (60.91). Nonmetropolitan areas saw increasing trends, whereas metropolitan areas saw declines. AAMR remained highest among non-Hispanic (NH) White individuals; however, significant decreases were observed in Hispanic/Latino and NH Asian/Pacific Islander populations. Conclusions Aortic stenosis–related mortality declined modestly among older adults from 1999 to 2024. However, notable demographic and regional disparities persisted, including higher mortality in men, NH White individuals, older adults ≥85 years, and residents of nonmetropolitan regions and the Midwest. Further investigation and targeted interventions are needed to improve outcomes.
Background Myocardial infarction with non-obstructive coronary arteries (MINOCA) requires multimodality evaluation, including cardiac magnetic resonance (CMR), intracoronary (IC) imaging, and coronary function testing (CFT). However, real-world uptake of these guideline-recommended modalities varies widely. We conducted a global survey to characterize practice variation in MINOCA evaluation and management across demographic characteristics, continents, and economic settings. Methods A 16-item online survey targeting cardiologists and cardiology trainees was distributed globally between August and December 2024. Respondents provided information on demographic characteristics, institutional context, and routine diagnostic and therapeutic strategies for MINOCA. Descriptive statistics, χ2 tests, and multiple regression analysis were used to assess associations. Results A total of 329 cardiology professionals from 6 continents participated (69.6% interventional cardiologists; 45.9% practicing in high-income countries [HIC]). Overall, 64.1% of respondents reported routine statin use and 55.6% reported beta-blocker use in MINOCA management. Advanced diagnostic modalities demonstrated substantial geographic variation. Europe reported the highest use of CMR (80.6%) and CFT (38.9%), whereas IC imaging was most frequently reported in North America (44.1%). Respondents from HIC reported greater routine use of CMR (56.3% vs 31.6%), IC imaging (41.7% vs 28.9%), and CFT (31.1% vs 13.2%) compared with those from lower-income settings (all P < .01). In multivariable analyses, HIC status remained the strongest independent predictor of routine CMR use (odds ratio [OR], 3.95; 95% CI, 2.34-6.68), IC imaging use (OR, 3.01; 95% CI, 1.74-5.20), and CFT use (OR, 2.29; 95% CI, 1.26-4.14). Academic or teaching hospital affiliation was independently associated with higher CMR use (OR, 2.19; 95% CI, 1.07-4.51), while female respondents reported higher odds of CFT use (OR, 1.89; 95% CI, 1.01-3.54). Conclusions Significant global disparities exist in the diagnostic evaluation of MINOCA. After adjustment, economic context emerged as the most consistent determinant of advanced diagnostic modality use. These findings highlight persistent structural barriers to the implementation of guideline-recommended MINOCA evaluation and underscore the need for strategies that expand access to advanced diagnostics and promote standardized evaluation pathways across diverse health care settings.
Background High-risk pulmonary embolism (PE) is associated with high mortality. Optimal reperfusion strategy remains debated. Systemic thrombolysis (ST) is traditionally used but carries a high risk of intracranial bleeding. Catheter-directed or mechanical thrombectomy (MT) may be safer, but real-world data are limited. We compared thrombectomy vs thrombolysis using a multicenter data network. Methods We conducted a retrospective cohort study on the TriNetX Research Network across 107 healthcare organizations. Adults with hemodynamically unstable PE on vasopressors from January 2016 to September 2025 who received MT (n = 1064) or ST (n = 3325) within 1 day of diagnosis were included. We excluded patients with COVID-19, malignancy, obstetric conditions, trauma, and burns. Propensity matching (1:1) for Pulmonary Embolism Severity Index (PESI) variables and other baseline characteristics yielded 1052 per group. Results Mortality was lower with MT (RR, 0.53; 95% CI, 0.44-0.63). Kaplan-Meier analysis showed better survival in the matched MT group (79.4% vs 61.3%; P < .001). Hemorrhagic stroke (RR, 0.71; 95% CI, 0.41-1.22) and ischemic stroke (RR, 0.80; 95% CI, 0.57-1.13) were also reduced in the thrombectomy group. Conclusions Thrombectomy was associated with lower mortality and fewer strokes vs thrombolysis. Thrombectomy appears safer and more effective for high-risk PE. This finding is hypothesis generating. Large randomized controlled trials, although difficult to conduct, are needed to validate these findings.
Background Atrial fibrillation (AF) is prevalent among patients undergoing transcatheter aortic valve replacement (TAVR) and increases risks of thromboembolism and bleeding with chronic anticoagulation. Concomitant left atrial appendage occlusion (LAAO) during TAVR may reduce long-term stroke and bleeding risks. Methods PubMed, Embase, and Cochrane databases were searched through May 2025 for studies comparing TAVR + LAAO versus TAVR + standard medical therapy in patients with atrial fibrillation. Random-effects models were used to pool risk ratios (RRs) or mean differences with 95% CIs. Results Four studies (2 randomized controlled trials, 2 observational; n = 563, mean age 82 years) were included; 279 patients (49.5%) underwent TAVR + LAAO. Thirty-day outcomes showed no differences in mortality (RR 0.91; 95% CI, 0.28-3.01), stroke/transient ischemic attack (RR 0.27; 95% CI, 0.04-1.66), or major bleeding (RR 0.27; 95% CI, 0.04-1.66). Over a 9.8-month to 2-year follow-up, all-cause mortality (RR 0.95; 95% CI, 0.68-1.32), cardiac death (RR 0.95; 95% CI, 0.60-1.50), stroke/transient ischemic attack (RR 1.33; 95% CI, 0.30-5.83), and major bleeding (RR 1.06; 95% CI, 0.71-1.59) remained similar. Contrast use and hospital length of stay were comparable. Conclusions Concomitant TAVR plus LAAO appears feasible, with no statistically significant differences in short-term outcomes versus TAVR plus standard medical therapy; however, certainty is low, and findings are hypothesis-generating, warranting larger randomized trials and prospective registries.
