Daily Cardiology Research Analysis
Analyzed 197 papers and selected 3 impactful papers.
Summary
Three impactful cardiology studies stood out: a deep-learning echocardiography model accurately detected cardiac amyloidosis across international cohorts; an individual patient data meta-analysis showed genotype-guided de-escalation of P2Y12 therapy after ACS PCI reduces bleeding without increasing ischemic events; and a Bayesian meta-analysis of RCTs found higher 5-year mortality with TAVI compared to SAVR in low- to intermediate-risk severe aortic stenosis, underscoring durability concerns.
Research Themes
- AI-enabled cardiovascular diagnostics
- Pharmacogenomics-guided antiplatelet therapy
- Long-term outcomes and durability of structural heart interventions
Selected Articles
1. Diagnosis of Cardiac Amyloidosis on Echocardiography Using Artificial Intelligence.
A multicenter study developed and externally validated AI tools to detect cardiac amyloidosis from echocardiography. A fully automated deep-learning video model achieved high accuracy across US and Japanese cohorts and outperformed an AI-derived multiparametric echocardiographic score, while robustly discriminating CA from hypertrophic phenocopies.
Impact: This work advances scalable, vendor-agnostic AI for noninvasive CA detection that could enable earlier diagnosis and triage using routine echocardiography. The rigorous external validation across continents increases generalizability and potential clinical adoption.
Clinical Implications: Automated CA screening at the point of echocardiography could shorten diagnostic pathways, prompt confirmatory testing (e.g., nuclear scintigraphy, CMR, biopsy), and accelerate initiation of disease-modifying therapy. Integration into lab workflows may reduce missed CA in hypertrophic phenotypes.
Key Findings
- Deep-learning video model achieved external accuracies of 87.5% (US) and 88.4% (Japan), with sensitivities 86.6% and 92.3%, respectively.
- AI-derived multiparametric echo score reached ~79.5–79.7% accuracy in external cohorts but was outperformed by the deep-learning model (AUC 0.93 vs 0.88).
- Robust discrimination from hypertrophic phenocopies (AUC 0.91–0.93 vs hypertension, HCM, aortic stenosis, and CKD).
Methodological Strengths
- Large, international multicenter cohorts with external validations across different health systems
- Direct head-to-head comparison with an established multiparametric echocardiographic score
Limitations
- Retrospective design without prospective clinical impact assessment
- Potential selection and imaging-quality biases; deployment across vendors and acquisition protocols requires further calibration
Future Directions: Prospective implementation studies assessing diagnostic yield, workflow impact, and outcomes; calibration across vendors; and regulatory evaluation for clinical deployment.
BACKGROUND: Diagnosing cardiac amyloidosis (CA) on echocardiography can be challenging due to the imaging overlap between CA and more prevalent causes of a hypertrophic phenotype. This study sought to (1) evaluate the performance of artificial-intelligence (AI) derived measurements incorporated into the established multiparametric echocardiographic scoring system to detect CA; (2) develop and validate an AI-based deep-learning model for video-based detection of CA on echocardiography. METHODS: The study population comprised 5776 patients (CA, 2756; controls, 3020). The training data set included patients from the UK National Amyloidosis Center and Taiwan MacKay Memorial Hospital (CA, 2241; controls, 2130). External test data sets were obtained from the US Duke University Health System (CA, 334; LVH controls, 668) and Japan National Cerebral and Cardiovascular Center (CA, 181; LVH controls, 222). RESULTS: The multiparametric echocardiographic score computed using AI-derived measurements achieved an accuracy of 79.5% (sensitivity, 75.4%; specificity, 81.5%) in the United States cohort and 79.7% (sensitivity, 81.6%; specificity, 78.1%) in the Japan cohort. The deep-learning model demonstrated accuracies of 96.2% (sensitivity, 96.8%; specificity, 95.7%) and 95.8% (sensitivity, 97.3%; specificity, 94.3%) in the internal validation and internal test sets, respectively. External validation of the deep-learning model showed accuracies of 87.5% (sensitivity, 86.6%; specificity, 87.9%) in the United States and 88.4% (sensitivity, 92.3%; specificity, 85.3%) in the Japanese cohort. Subgroup analysis demonstrated that the deep-learning model showed robust discrimination of CA from other hypertrophic phenocopies: CA versus hypertension (area under the curve [AUC], 0.92 [95% CI, 0.91-0.94]), CA versus hypertrophic cardiomyopathy (AUC, 0.91 [95% CI, 0.87-0.94]), CA versus aortic stenosis (AUC, 0.93 [95% CI, 0.90-0.95]), CA versus chronic kidney disease (AUC, 0.93 [95% CI, 0.91-0.95]). The deep-learning model was able to classify a greater proportion of patients compared with the AI-derived multiparametric echocardiographic score and achieved superior diagnostic accuracy (AUC, 0.93 [95% CI, 0.91-0.95] versus AUC, 0.88 [95% CI, 0.85-0.90]; CONCLUSIONS: Both the multiparametric echocardiographic score computed from AI-derived measurements and the fully automated deep-learning model can accurately identify patients with CA in globally diverse cohorts, with the deep-learning model providing superior performance.
2. Genotype-Guided vs Conventional Oral P2Y12 Inhibitors in Acute Coronary Syndrome: An Individual Patient Data Meta-analysis of Randomized Controlled Trials.
In an IPD meta-analysis of two RCTs in ACS PCI, genotype-guided therapy overall reduced MI and NACE. Notably, genotype-guided de-escalation reduced BARC 2/3/5 bleeding and NACE without increasing MACE, with the largest benefit in the first 90 days; escalation showed no advantage.
Impact: This analysis sharpens the precision-medicine case for genotype-guided P2Y12 de-escalation soon after ACS PCI, balancing bleeding and ischemic risk where clinical trade-offs are most consequential.
Clinical Implications: CYP2C19-guided de-escalation (e.g., short-term potent agent then clopidogrel in non-LOF carriers) may be prioritized in the first 1–3 months post-PCI to reduce bleeding and NACE without compromising ischemic protection.
Key Findings
- Overall genotype-guided therapy reduced myocardial infarction (RR 0.68) and NACE (RR 0.85) versus conventional therapy at 12 months.
- Genotype-guided de-escalation reduced BARC 2/3/5 bleeding (RR 0.77) and NACE (RR 0.77) without increasing MACE.
- Benefits of guided therapy and de-escalation were most pronounced within the first 90 days post-PCI.
Methodological Strengths
- Individual patient data meta-analysis of randomized trials enabling time-to-event and time-dependent analyses
- Stratified evaluation of escalation versus de-escalation strategies
Limitations
- Only two RCTs contributed IPD; limited power for some subgroups and implementation contexts
- Real-world logistics and turnaround times for genotyping may affect feasibility and adherence
Future Directions: Pragmatic, geographically diverse trials of genotype-guided de-escalation with rapid testing pathways; cost-effectiveness and implementation studies.
BACKGROUND: Genotype-guided P2Y OBJECTIVES: This study sought to assess the impact of guided therapy escalation or de-escalation vs conventional therapy. METHODS: Randomized controlled trials comparing guided therapy using CYP2C19 genetic testing vs conventional therapy among patients with ACS undergoing PCI were searched and individual participant-level data obtained. The primary safety endpoint was time-to-first type 2, 3, or 5 Bleeding Academic Research Consortium (BARC) bleeding at 12 months. The primary efficacy endpoint was time-to-first major adverse cardiovascular event (MACE) at 12 months. RESULTS: A total of 6,734 participants from 2 randomized controlled trials were available for analysis. After 1 year, there were no differences in the primary safety or efficacy endpoints with overall guided therapy vs conventional therapy. However, guided therapy reduced myocardial infarction (0.68; 95% CI: 0.48-0.97) and net adverse cardiovascular events (NACE) (0.85; 95% CI: 0.73-1.00) compared with conventional therapy. Guided therapy de-escalation reduced the primary safety endpoint (0.77; 95% CI: 0.62-0.97) and NACE (0.77; 95% CI: 0.62-0.94) with no significant difference in MACE, compared with conventional therapy. The primary safety and efficacy endpoints were similar between patients undergoing guided therapy escalation and conventional therapy groups. Time-dependent covariate analyses showed that overall guided therapy and de-escalation strategies reduced bleeding and NACE before 90 days, compared with conventional therapy. CONCLUSIONS: These findings support evaluating genotype-guided therapy by separately analyzing de-escalation and escalation. In ACS patients undergoing PCI, genotype-guided de-escalation reduces bleeding and NACE without increasing MACE, with the greatest benefit in the first 3 months post-PCI. (Genotype-Guided versus Conventional Oral P2Y12 Inhibitors in Acute Coronary Syndrome: An Individual Patient Data Meta-analysis of Randomized Controlled Trials; PROSPERO CRD42024580431).
3. Updated 5-year outcomes of transcatheter versus surgical aortic valve replacement in patients with severe aortic stenosis at low- to intermediate-surgical risk.
A Bayesian meta-analysis of six RCTs (n=7249) with ≥5-year follow-up found higher all-cause mortality with TAVI versus SAVR (RR≈1.12) in low- to intermediate-risk severe aortic stenosis, with a high probability of increased stroke. Findings inform durability-focused decision-making in younger/healthier patients.
Impact: As TAVI use expands to younger, lower-risk populations, robust synthesis showing worse 5-year mortality versus SAVR highlights durability and cerebrovascular trade-offs central to guideline and shared decision-making.
Clinical Implications: For low- to intermediate-risk severe aortic stenosis, SAVR may offer superior 5-year survival and stroke profile. Patient selection for TAVI should weigh coronary access, lifetime management, and redo strategies.
Key Findings
- Across six RCTs (n=7249), 5-year all-cause mortality was higher with TAVI (29.7%) vs SAVR (27.6%); median RR 1.12 (95% CrI 1.02–1.22).
- High probability signal of increased 5-year stroke with TAVI; composite of mortality+stroke also favored SAVR.
- Findings were robust across sensitivity analyses and modeling approaches.
Methodological Strengths
- Bayesian hierarchical meta-analysis of randomized trials with ≥5-year outcomes
- Reconstruction of time-to-event data and extensive sensitivity analyses
Limitations
- Device generations and operator experience varied across trials and eras
- Study-level meta-analysis without patient-level data; applicability limited to low- to intermediate-risk populations
Future Directions: Head-to-head contemporary trials with latest-generation valves, patient-level meta-analyses, and registries focused on lifetime management and redo-TAVR planning.
OBJECTIVES: The comparative long-term safety and efficacy of transcatheter aortic valve implantation (TAVI) versus surgical aortic valve replacement (SAVR) remains under continued investigation, particularly in patients at low- to intermediate-surgical risk. This study aims to synthesise and update contemporary long-term TAVI versus SAVR data. METHODS: This study comprised a systematic review and meta-analysis and employed a Bayesian hierarchical design. Randomised controlled trials (RCTs) comparing TAVI to SAVR in low-risk to intermediate-risk patients with at least 5-year follow-up were included. The primary outcome was 5-year all-cause mortality; secondary outcomes were the 5-year incidence of stroke and the 5-year incidence of the composite of mortality and stroke. REVIEW METHODS: Time-to-event data were reconstructed. Relative risks (RRs) with 95% credible intervals (CrIs) were estimated from reported 5-year event rates using minimally informative priors. Sensitivity analyses were performed using various meta-analytical models, and using conventional frequentist random-effects and fixed-effects models for sensitivity purposes. RESULTS: A total of six RCTs, enrolling 7249 low- to intermediate-risk patients reported 5-year outcomes (TAVI n=3704, SAVR n=3545). The 5-year all-cause mortality rate was 29.7% (28.2-31.2%, TAVI) and 27.6% (26.1-29.1%, SAVR). The median RR for all-cause mortality was 1.12 (95% CrI 1.02-1.22, heterogeneity τ CONCLUSION: In this meta-analysis of RCTs, TAVI resulted in a clinically relevant increase in all-cause mortality, and a high probability of an increased risk of stroke, at 5 years of follow-up in low-risk to intermediate-risk patients, when compared to SAVR.