Publications

BACKGROUND: Abnormal ventricular depolarization, evident as a broad QRS complex on an ECG, is traditionally categorized into left bundle-branch block (LBBB) and right bundle-branch block or nonspecific intraventricular conduction delay. This categorization, although physiologically accurate, may fail to capture the nuances of diseases subtypes.

METHODS: We used unsupervised machine learning to identify and characterize novel broad QRS phenogroups. First, we trained a variational autoencoder on 1.1 million ECGs and discovered 51 latent features that showed high disentanglement and ECG reconstruction accuracy. We then extracted these features from 42 538 ECGs with QRS durations >120 milliseconds and employed a reversed graph embedding method to model population heterogeneity as a tree structure with different branches representing phenogroups.

RESULTS: Six phenogroups were identified, including phenogroups of right bundle-branch block and LBBB with varying risk of cardiovascular disease and mortality. The higher risk right bundle-branch block phenogroup exhibited increased risk of cardiovascular death (adjusted hazard ratio [aHR], 1.46 [1.30-1.63], P<0.0001) and all-cause mortality (aHR, 1.24 [1.16-1.33], P<0.0001) compared with the baseline phenogroup. Within LBBB ECGs, tree position predicted future cardiovascular disease risk differentially. Additionally, for subjects with LBBB undergoing cardiac resynchronization therapy, tree position predicted cardiac resynchronization therapy response independent of covariates, including QRS duration (adjusted odds ratio [aOR], 0.47 [0.25-0.86], P<0.05).

CONCLUSIONS: Our findings challenge the current paradigm, highlighting the potential for these phenogroups to enhance cardiac resynchronization therapy patient selection for subjects with LBBB and guide investigation and follow-up strategies for subjects with higher risk right bundle-branch block.

BACKGROUND AND AIMS: Valvular heart disease (VHD) is a significant source of morbidity and mortality, though early intervention can improve outcomes. This study aims to develop artificial intelligence-enhanced electrocardiography (AI-ECG) models to diagnose and predict future moderate or severe regurgitant VHDs (rVHDs), including mitral regurgitation (MR), tricuspid regurgitation (TR), and aortic regurgitation (AR).

METHODS: The AI-ECG models were developed in a data set of 988 618 ECG and transthoracic echocardiogram pairs from 400 882 patients from Zhongshan Hospital, Shanghai, China. The AI-ECG models used a residual convolutional neural network with a discrete-time survival loss function. External evaluation was performed in outpatients from a secondary care data set from Beth Israel Deaconess Medical Center, Boston, USA, consisting of 34 214 patients with linked echocardiography.

RESULTS: In the internal test set, the AI-ECG models accurately predicted future significant MR [C-index 0.774, 95% confidence interval (CI) 0.753-0.792], AR (0.691, 95% CI 0.657-0.720), and TR (0.793, 95% CI 0.777-0.808). In age- and sex-adjusted Cox models, the highest risk quartile had a hazard ratio (HR) of 7.6 (95% CI 5.8-9.9, P < .0001) for risk of future significant MR, compared with the lowest risk quartile. For future AR and TR, the equivalent HRs were 3.8 (95% CI 2.7-5.5) and 9.9 (95% CI 7.5-13.0), respectively. These findings were confirmed in the transnational external test set. Imaging association analyses demonstrated AI-ECG predictions were associated with subclinical chamber remodelling.

CONCLUSIONS: This study developed AI-ECG models to diagnose and predict rVHDs and validated the models in a transnational and ethnically distinct cohort. The AI-ECG models could be utilized to guide surveillance echocardiography in patients at risk of future rVHDs, to facilitate early detection and intervention.

BACKGROUND: Cardiac magnetic resonance imaging (CMR) contributes to the diagnostic evaluation of cardiomyopathy and procedural planning. Many patients referred for clinical CMR have cardiac implantable electrical devices (CIEDs). Few studies have described the feasibility of CMR at 3T in patients with CIED.

OBJECTIVE: This study aimed to assess the safety and quality of CMR at 3T field strength in patients with conditional CIEDs.

METHODS: We implemented an abbreviated CMR protocol including cardiac cine and late gadolinium enhancement (LGE) sequences. We identified patients with magnetic resonance imaging-conditional CIEDs at 3T referred for a clinical CMR from September 2020 to February 2024. CIED function was assessed after each scan. Cardiac cine and LGE sequences were evaluated for quality and artifacts affecting interpretation.

RESULTS: We evaluated 87 patients (22 women, 66 ± 11.7 years) who underwent 90 scans. No adverse events were observed during any scan. No changes in battery voltage or lead parameters required device reprogramming; 82 scans (91%) were diagnostic. In 50 cine sequences (59%) and 49 LGE sequences (58%), the quality was determined to be good or good/intermediate. In 73 cine (86%) and 47 LGE sequences (55%), there was agreement that there were no artifacts affecting interpretation.

CONCLUSION: Our single-center real-world experience confirms the safety of CIED-CMR at 3T for appropriately labeled CIEDs, with no adverse events or changes to device parameters requiring reprogramming. Although images may be suboptimal owing to artifacts, more than 90% of scans were diagnostic.

INTRODUCTION: Complete heart block (CHB) is a life-threatening condition that can lead to ventricular standstill, syncopal injury, and sudden cardiac death, and current electrocardiography (ECG)-based risk stratification (presence of bifascicular block) is crude and has limited performance. Artificial intelligence-enhanced electrocardiography (AI-ECG) has been shown to identify a broad spectrum of subclinical disease and may be useful for CHB.

OBJECTIVE: To develop an AI-ECG risk estimator for CHB (AIRE-CHB) to predict incident CHB.

DESIGN, SETTING, AND PARTICIPANTS: This cohort study was a development and external validation prognostic study conducted at Beth Israel Deaconess Medical Center and validated externally in the UK Biobank volunteer cohort.

EXPOSURE: Electrocardiogram.

MAIN OUTCOMES AND MEASURES: A new diagnosis of CHB more than 31 days after the ECG. AIRE-CHB uses a residual convolutional neural network architecture with a discrete-time survival loss function and was trained to predict incident CHB.

RESULTS: The Beth Israel Deaconess Medical Center cohort included 1 163 401 ECGs from 189 539 patients. AIRE-CHB predicted incident CHB with a C index of 0.836 (95% CI, 0.819-0.534) and area under the receiver operating characteristics curve (AUROC) for incident CHB within 1 year of 0.889 (95% CI, 0.863-0.916). In comparison, the presence of bifascicular block had an AUROC of 0.594 (95% CI, 0.567-0.620). Participants in the high-risk quartile had an adjusted hazard ratio (aHR) of 11.6 (95% CI, 7.62-17.7; P < .001) for development of incident CHB compared with the low-risk group. In the UKB UK Biobank cohort of 50 641 ECGs from 189 539 patients, the C index for incident CHB prediction was 0.936 (95% CI, 0.900-0.972) and aHR, 7.17 (95% CI, 1.67-30.81; P < .001).

CONCLUSIONS AND RELEVANCE: In this study, a first-of-its-kind deep learning model identified the risk of incident CHB. AIRE-CHB could be used in diverse settings to aid in decision-making for individuals with syncope or at risk of high-grade atrioventricular block.

BACKGROUND: Multiple risk scores and biomarkers have been proposed for the prediction of atrial fibrillation (AF), but it is unknown how these compare with each other and whether they could be combined.

OBJECTIVE: This study aimed to evaluate and compare approaches for incident AF prediction.

METHODS: The artificial intelligence-enhanced electrocardiogram risk estimator-AF (AIRE-AF), a convolutional neural network with a discrete-time survival loss function, was developed to predict incident AF. It was trained using a dataset of 1,163,401 electrocardiograms from 189,539 patients from the Beth Israel Deaconess Medical Center and externally validated in the UK Biobank (n = 38,892). AIRE-AF was compared with other risk prediction approaches including the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE)-AF, a clinical risk score.

RESULTS: In the Beth Israel Deaconess Medical Center cohort, AIRE-AF predicted incident AF with a C-index of 0.750 (0.743-0.758). AIRE-AF was superior to CHARGE-AF, left atrial size, and N-terminal pro-B-type natriuretic peptide. The addition of CHARGE-AF and left atrial size provided a minor improvement in performance (C-index improvement 0.017). There was no additive value of N-terminal pro-B-type natriuretic peptide in combination with AIRE-AF. The single best-performing single predictor in the volunteer population (UK Biobank) was CHARGE-AF (C-index 0.750 [0.734-0.769]). The best-performing combination of 2 predictors was AIRE-AF and CHARGE-AF (C-index 0.768 [0.743-0.792]). The addition of a polygenic risk score to AIRE-AF and CHARGE-AF provided a further significant improvement in performance (C-index 0.791 [0.766-0.816]).

CONCLUSION: We present the first comprehensive evaluation of methodologies for predicting incident AF. Risk prediction with a model including AIRE-AF and CHARGE-AF resulted in similar performance to more complex models.

OBJECTIVE: Despite advances in the management of chronic limb-threatening ischemia (CLTI), a large proportion of these patients are not candidates for traditional revascularization and may be destined for major amputation. Given this medically complex and no-option patient population, deep venous arterialization (DVA) has been recently revitalized as a limb salvage technique, whereby an arteriovenous fistula in the lower leg is created to supply more oxygenated blood via the venous system to the foot. Recently, PROMISE II (Percutaneous Deep Vein Arterialization for the Treatment of Late-Stage Chronic Limb-Threatening Ischemia) demonstrated a 6-month amputation-free survival (AFS) rate of 66% after DVA. With this trial in mind, our study aimed to evaluate the real-world outcomes of this procedure.

METHODS: The study population included all patients undergoing a DVA from January 1, 2021, through December 31, 2023 among fee-for-service beneficiaries identified in the Medicare Fee-for-Service Carrier Claims file. DVA procedures were identified using Current Procedural Terminology code 0620T. Outcomes included limb salvage, freedom from major adverse limb events (defined as major amputation or ipsilateral reintervention), survival, and AFS. Cumulative incidences for outcomes that include death were estimated from traditional Kaplan-Meier methods; for non-death end points, outcomes were estimated from the cumulative incidence function, accounting for the competing risk of death.

RESULTS: Between 2021 and 2023, 134 patients underwent a DVA for CLTI. Among these, the median age was 70 years and the majority of patients were male (66%), White (63%), and had tissue loss (72%), hypertension (99%), hyperlipidemia (96%), chronic kidney disease (89%), and diabetes (83%). After a DVA for CLTI, the 6-month and 1-year AFS incidences were 42% and 33%, respectively. One-year incidences of limb salvage, freedom from major adverse limb events, and survival were 53%, 36%, and 65%, respectively.

CONCLUSIONS: Among patients with no traditional options for revascularization, our data demonstrate that DVA is a procedure that is, by its nature, performed on high-risk individuals who continue to have a high risk of limb loss and mortality. Importantly, AFS in our analysis was notably worse than that reported in PROMISE II and, as such, raises questions about the generalizability of this procedure in real world practice. Further investigation is needed regarding patient selection criteria for and the clinical usefulness of the DVA procedure.

Endovascular renal denervation (RDN) is a catheter-based, procedural therapy for the treatment of hypertension that was approved by the US Food and Drug Administration (FDA) in November 2023. Early studies suggest that endovascular RDN reduces blood pressure (BP) in patients with hypertension. However, BP response to RDN is highly variable, optimal patient selection remains uncertain, and the procedure's high cost remains a significant challenge. The purpose of this review is to comprehensively examine the literature regarding the mechanism by which endovascular RDN reduces BP and the safety and effectiveness of RDN, and to discuss key considerations for selecting appropriate patients for endovascular RDN. Relevant studies in the field were identified from PubMed using search terms including 'renal denervation' and 'renal denervation for hypertension.' In conclusion, clinical trials have demonstrated a statistically significant BP-lowering effect of endovascular RDN, which based on multiple trials with long-term follow-up, appears to persist over several years with low complication rates. More research is needed to understand which patients benefit most from endovascular RDN and to evaluate the long-term outcomes, including the impact of endovascular RDN on cardiovascular events.

OBJECTIVE: Routine imaging surveillance following endovascular aortic aneurysm repair (EVAR) for abdominal aortic aneurysm (AAA) is critical for the timely diagnosis of late postoperative complications. Compliance with recommended EVAR surveillance is variable, and disparities in post-EVAR surveillance remain unclear. This study examines variability in EVAR surveillance and emergency health service use across several sociodemographic populations.

METHODS: All Medicare fee-for-service beneficiaries who underwent infrarenal EVAR for intact abdominal aortic aneurysm between January 2011 and December 2019 were included. Patients were stratified by several sociodemographic characteristics: age category (66-74, 75-84, >85 years), sex (male, female), race (White, Black, Asian, other), dual enrollment in Medicare and Medicaid (dual enrolled, Medicare only), and distressed communities index (distressed >80th percentile, nondistressed ≤80th percentile). The following postoperative health care use metrics were assessed: EVAR-related office visits, imaging studies, emergency department (ED) visits, and hospital readmissions. Annual incidence rates were calculated for each health care use metric at 2 and 5 years after EVAR and compared across groups using Poisson regression models, adjusting for sociodemographic and hospital characteristics and comorbidities.

RESULTS: In 111,381 Medicare beneficiaries who underwent EVAR, postoperative health care use varied substantially across sociodemographic groups. After adjustment, annual incidence rates of EVAR-related office visits at 2 years post EVAR were lower in patients who were >85 years vs 66-75 years (adjusted rate ratio [aRR], 0.95; 95% confidence interval [CI], 0.93-0.97), female vs male (aRR, 0.94; 95% CI, 0.93-0.95), dual enrolled vs Medicare only (aRR, 0.83; 95% CI, 0.81-0.85), and residing in distressed vs nondistressed communities (aRR, 0.95; 95% CI, 0.93-0.96). Rates of imaging studies were lower in patients who were >85 years (aRR, 0.98; 95% CI, 0.96-0.99), dual enrolled (0.97; 95% CI, 0.95-0.98), and residing in distressed communities (aRR, 0.97; 95% CI, 0.96-0.98). There was higher use of hospital services in patients who were >85 years (ED: aRR, 1.37; 95% CI, 1.33-1.41; readmission: aRR, 1.23; 95% CI,1.19, 1.28), female (ED: aRR, 1.19; 95% CI, 1.16-1.22; readmission: aRR, 1.15; 95% CI, 1.12-1.19), Black (ED: aRR, 1.10; 95% CI, 1.05-1.15; readmission: aRR, 1.15; 95% CI, 1.09-1.22), dual-enrolled (ED: 1.29; aRR, 95% CI, 1.26-1.33; readmission: aRR, 1.14; 95% CI, 1.09-1.18), and residing in distressed communities (ED: aRR, 1.03; 95% CI, 1.01-1.06; readmission: aRR, 1.02; 95% CI, 0.99-1.05). At 5 years post EVAR, similar trends across sociodemographic groups were observed, with the added finding of lower rates of EVAR-related office visits in Black vs White patients.

CONCLUSIONS: Significant variation in post-EVAR health care use exists among Medicare beneficiaries. Patients who were older age, of female sex, of Black race, or socioeconomically disadvantaged had lower rates of EVAR-specific follow-up and higher use of emergency health service. Barriers in access to care are apparent, underscoring the need for targeted interventions to enhance post-EVAR surveillance and improve outcomes in these populations.