Deep Learning Algorithm for Automated Cardiac Murmur Detection via a Digital Stethoscope Platform.

Background Clinicians vary markedly in their ability to detect murmurs during cardiac auscultation and identify the underlying pathological features. Deep learning approaches have shown promise in medicine by transforming collected data into clinically significant information. The objective of this research is to assess the performance of a deep learning algorithm to detect murmurs and clinically significant valvular heart disease using recordings from a commercial digital stethoscope platform. Methods and Results Using >34 hours of previously acquired and annotated heart sound recordings, we trained a deep neural network to detect murmurs. To test the algorithm, we enrolled 962 patients in a clinical study and collected recordings at the 4 primary auscultation locations. Ground truth was established using patient echocardiograms and annotations by 3 expert cardiologists. Algorithm performance for detecting murmurs has sensitivity and specificity of 76.3% and 91.4%, respectively. By omitting softer murmurs, those with grade 1 intensity, sensitivity increased to 90.0%. Application of the algorithm at the appropriate anatomic auscultation location detected moderate-to-severe or greater aortic stenosis, with sensitivity of 93.2% and specificity of 86.0%, and moderate-to-severe or greater mitral regurgitation, with sensitivity of 66.2% and specificity of 94.6%. Conclusions The deep learning algorithm's ability to detect murmurs and clinically significant aortic stenosis and mitral regurgitation is comparable to expert cardiologists based on the annotated subset of our database. The findings suggest that such algorithms would have utility as front-line clinical support tools to aid clinicians in screening for cardiac murmurs caused by valvular heart disease. Registration URL: https://clinicaltrials.gov; Unique Identifier: NCT03458806.

Global Coronary Flow Reserve Measured During Stress Cardiac Magnetic Resonance Imaging Is an Independent Predictor of Adverse Cardiovascular Events.

OBJECTIVES: The aim of this study was to evaluate the incremental prognostic value of global coronary flow reserve (CFR) in patients with known or suspected coronary artery disease who were undergoing stress cardiac magnetic resonance (CMR) imaging. BACKGROUND: Coronary microvascular dysfunction results in impaired global CFR and is implicated in the development of both atherosclerosis and heart failure. Although noninvasive assessment of CFR with positron emission tomography provides independent prognostic information, the incremental prognostic value of CMR-derived CFR remains unclear. METHODS: Consecutive patients undergoing stress perfusion CMR were prospectively enrolled (n = 507). Coronary sinus flow was measured using phase-contrast imaging at baseline (pre) and immediately after stress (peak) perfusion. CFR was calculated as the ratio of peak to pre-flow. Patients were followed for major adverse cardiac events (MACE): death, nonfatal myocardial infarction, heart failure hospitalization, sustained ventricular tachycardia, and late revascularization. Cox proportional hazards regression modeling was used to examine the association between CFR and MACE. The incremental prognostic value of CFR was assessed in nested models. RESULTS: Over a median follow-up of 2.1 years, 80 patients experienced MACE. By Kaplan-Meier analysis, the risk of MACE was significantly higher in patients with CFR lower than the median (2.2) (log-rank p < 0.001); this remained significant after adjustment for the presence of ischemia and late gadolinium enhancement (LGE) (log-rank p < 0.001). CFR was significantly associated with the risk of MACE after adjustment for clinical and imaging risk factors, including ischemia extent, ejection fraction, and LGE size (hazard ratio: 1.238; p = 0.018). Addition of CFR in this model resulted in significant improvement in the C-index (from 0.70 to 0.75; p = 0.0087) and a continuous net reclassification improvement of 0.198 (95% confidence interval: 0.120 to 0.288). CONCLUSIONS: CMR-derived CFR is an independent predictor of MACE in patients with known or suspected coronary artery disease, incremental to common clinical and CMR risk factors. These findings suggest a role for CMR-derived CFR in identifying patients at risk of adverse events following stress CMR, even in the absence of ischemia and LGE.

Feature-Tracking Global Longitudinal Strain Predicts Death in a Multicenter Population of Patients With Ischemic and Nonischemic Dilated Cardiomyopathy Incremental to Ejection Fraction and Late Gadolinium Enhancement.

OBJECTIVES: The aim of this study was to evaluate the prognostic value of cardiac magnetic resonance (CMR) feature-tracking-derived global longitudinal strain (GLS) in a large multicenter population of patients with ischemic and nonischemic dilated cardiomyopathy. BACKGROUND: Direct assessment of myocardial fiber deformation with GLS using echocardiography or CMR feature tracking has shown promise in providing prognostic information incremental to ejection fraction (EF) in single-center studies. Given the growing use of CMR for assessing persons with left ventricular (LV) dysfunction, we hypothesized that feature-tracking-derived GLS may provide independent prognostic information in a multicenter population of patients with ischemic and nonischemic dilated cardiomyopathy. METHODS: Consecutive patients at 4 U.S. medical centers undergoing CMR with EF <50% and ischemic or nonischemic dilated cardiomyopathy were included in this study. Feature-tracking GLS was calculated from 3 long-axis cine-views. The primary endpoint was all-cause death. Cox proportional hazards regression modeling was used to examine the association between GLS and death. Incremental prognostic value of GLS was assessed in nested models. RESULTS: Of the 1,012 patients in this study, 133 died during median follow-up of 4.4 years. By Kaplan-Meier analysis, the risk of death increased significantly with worsening GLS tertiles (log-rank p < 0.0001). Each 1% worsening in GLS was associated with an 89.1% increased risk of death after adjustment for clinical and imaging risk factors including EF and late gadolinium enhancement (LGE) (hazard ratio [HR]:1.891 per %; p < 0.001). Addition of GLS in this model resulted in significant improvement in the C-statistic (0.628 to 0.867; p < 0.0001). Continuous net reclassification improvement (NRI) was 1.148 (95% confidence interval: 0.996 to 1.318). GLS was independently associated with death after adjustment for clinical and imaging risk factors (including EF and late gadolinium enhancement) in both ischemic (HR: 1.942 per %; p < 0.001) and nonischemic dilated cardiomyopathy subgroups (HR: 2.101 per %; p < 0.001). CONCLUSIONS: CMR feature-tracking-derived GLS is a powerful independent predictor of mortality in a multicenter population of patients with ischemic or nonischemic dilated cardiomyopathy, incremental to common clinical and CMR risk factors including EF and LGE.

Pompe disease results in a Golgi-based glycosylation deficit in human induced pluripotent stem cell-derived cardiomyocytes.

Infantile-onset Pompe disease is an autosomal recessive disorder caused by the complete loss of lysosomal glycogen-hydrolyzing enzyme acid α-glucosidase (GAA) activity, which results in lysosomal glycogen accumulation and prominent cardiac and skeletal muscle pathology. The mechanism by which loss of GAA activity causes cardiomyopathy is poorly understood. We reprogrammed fibroblasts from patients with infantile-onset Pompe disease to generate induced pluripotent stem (iPS) cells that were differentiated to cardiomyocytes (iPSC-CM). Pompe iPSC-CMs had undetectable GAA activity and pathognomonic glycogen-filled lysosomes. Nonetheless, Pompe and control iPSC-CMs exhibited comparable contractile properties in engineered cardiac tissue. Impaired autophagy has been implicated in Pompe skeletal muscle; however, control and Pompe iPSC-CMs had comparable clearance rates of LC3-II-detected autophagosomes. Unexpectedly, the lysosome-associated membrane proteins, LAMP1 and LAMP2, from Pompe iPSC-CMs demonstrated higher electrophoretic mobility compared with control iPSC-CMs. Brefeldin A induced disruption of the Golgi in control iPSC-CMs reproduced the higher mobility forms of the LAMPs, suggesting that Pompe iPSC-CMs produce LAMPs lacking appropriate glycosylation. Isoelectric focusing studies revealed that LAMP2 has a more alkaline pI in Pompe compared with control iPSC-CMs due largely to hyposialylation. MALDI-TOF-MS analysis of N-linked glycans demonstrated reduced diversity of multiantennary structures and the major presence of a trimannose complex glycan precursor in Pompe iPSC-CMs. These data suggest that Pompe cardiomyopathy has a glycan processing abnormality and thus shares features with hypertrophic cardiomyopathies observed in the congenital disorders of glycosylation.

Chemical-genomic dissection of the CTD code.

Sequential modifications of the RNA polymerase II (Pol II) C-terminal domain (CTD) coordinate the stage-specific association and release of cellular machines during transcription. Here we examine the genome-wide distributions of the 'early' (phospho-Ser5 (Ser5-P)), 'mid' (Ser7-P) and 'late' (Ser2-P) CTD marks. We identify gene class-specific patterns and find widespread co-occurrence of the CTD marks. Contrary to its role in 3'-processing of noncoding RNA, the Ser7-P marks are placed early and retained until transcription termination at all Pol II-dependent genes. Chemical-genomic analysis reveals that the promoter-distal Ser7-P marks are not remnants of early phosphorylation but are placed anew by the CTD kinase Bur1. Consistent with the ability of Bur1 to facilitate transcription elongation and suppress cryptic transcription, high levels of Ser7-P are observed at highly transcribed genes. We propose that Ser7-P could facilitate elongation and suppress cryptic transcription.