Deep Learning-Enabled Assessment of Left Heart Structure and Function Predicts Cardiovascular Outcomes.

BACKGROUND: Deep learning interpretation of echocardiographic images may facilitate automated assessment of cardiac structure and function. OBJECTIVES: We developed a deep learning model to interpret echocardiograms and examined the association of deep learning-derived echocardiographic measures with incident outcomes. METHODS: We trained and validated a 3-dimensional convolutional neural network model for echocardiographic view classification and quantification of left atrial dimension, left ventricular wall thickness, chamber diameter, and ejection fraction. The training sample comprised 64,028 echocardiograms (n = 27,135) from a retrospective multi-institutional ambulatory cardiology electronic health record sample. Validation was performed in a separate longitudinal primary care sample and an external health care system data set. Cox models evaluated the association of model-derived left heart measures with incident outcomes. RESULTS: Deep learning discriminated echocardiographic views (area under the receiver operating curve >0.97 for parasternal long axis, apical 4-chamber, and apical 2-chamber views vs human expert annotation) and quantified standard left heart measures (R2 range = 0.53 to 0.91 vs study report values). Model performance was similar in 2 external validation samples. Model-derived left heart measures predicted incident heart failure, atrial fibrillation, myocardial infarction, and death. A 1-SD lower model-left ventricular ejection fraction was associated with 43% greater risk of heart failure (HR: 1.43; 95% CI: 1.23-1.66) and 17% greater risk of death (HR: 1.17; 95% CI: 1.06-1.30). Similar results were observed for other model-derived left heart measures. CONCLUSIONS: Deep learning echocardiographic interpretation accurately quantified standard measures of left heart structure and function, which in turn were associated with future clinical outcomes. Deep learning may enable automated echocardiogram interpretation and disease prediction at scale.

Identifying high risk clinical phenogroups of pulmonary hypertension through a clustering analysis.

INTRODUCTION: The classification and management of pulmonary hypertension (PH) is challenging due to clinical heterogeneity of patients. We sought to identify distinct multimorbid phenogroups of patients with PH that are at particularly high-risk for adverse events. METHODS: A hospital-based cohort of patients referred for right heart catheterization between 2005-2016 with PH were included. Key exclusion criteria were shock, cardiac arrest, cardiac transplant, or valvular surgery. K-prototypes was used to cluster patients into phenogroups based on 12 clinical covariates. RESULTS: Among 5208 patients with mean age 64±12 years, 39% women, we identified 5 distinct multimorbid PH phenogroups with similar hemodynamic measures yet differing clinical outcomes: (1) "young men with obesity", (2) "women with hypertension", (3) "men with overweight", (4) "men with cardiometabolic and cardiovascular disease", and (5) "men with structural heart disease and atrial fibrillation." Over a median follow-up of 6.3 years, we observed 2182 deaths and 2002 major cardiovascular events (MACE). In age- and sex-adjusted analyses, phenogroups 4 and 5 had higher risk of MACE (HR 1.68, 95% CI 1.41-2.00 and HR 1.52, 95% CI 1.24-1.87, respectively, compared to the lowest risk phenogroup 1). Phenogroup 4 had the highest risk of mortality (HR 1.26, 95% CI 1.04-1.52, relative to phenogroup 1). CONCLUSIONS: Cluster-based analyses identify patients with PH and specific comorbid cardiometabolic and cardiovascular disease burden that are at highest risk for adverse clinical outcomes. Interestingly, cardiopulmonary hemodynamics were similar across phenogroups, highlighting the importance of multimorbidity on clinical trajectory. Further studies are needed to better understand comorbid heterogeneity among patients with PH.

Indexing Left Atrial Volumes: Alternative Indexing Methods Better Predict Outcomes in Overweight and Obese Populations.

BACKGROUND: Left atrial volume (LAV) is often adjusted for body surface area (BSA). In overweight individuals this may result in underestimation of left atrial (LA) dilation. The authors investigated whether alternative indexing techniques better predict mortality and cardiovascular (CV) events. OBJECTIVES: The purpose of this study was to evaluate the efficacy of different methods of indexing LAV in predicting mortality and CV events across a range of body sizes. METHODS: LAV was adjusted for BSA, idealized BSA (iBSA), height, and height-squared (H2) in patients aged over 50 years who underwent outpatient echocardiography and longitudinal follow-up at our institution. LA dilation was categorized using published criteria. Mortality and CV events were assessed via medical records. RESULTS: LAVs were calculated in 17,454 individuals. In this study, 71.2% were overweight or obese. Indexing using iBSA, height, and H2 resulted in reclassification of LA size in up to 28.4% (P < 0.001) compared with indexing using BSA. In severely obese individuals (body mass index [BMI] ≥40 kg/m2), LA dilation indexed for BSA no longer predicted mortality (P = 0.70). Other indexing methods remained predictive of mortality. Height, H2, and iBSA all had greater performance, compared with BSA, for prediction of mortality and CV events in all overweight patients with H2 showing the best overall performance (P < 0.001). Net reclassification index for mortality was significant for all alternative indexing techniques (P < 0.001) and patients whose LA was reclassified from normal to dilated had increased risk of mortality (P < 0.001) and CV events (P < 0.001) across all BMI categories. CONCLUSIONS: LA dilation based on standard indexing using BSA is nondiscriminatory for prediction of mortality in the severely obese. Indexing using height, H2, or iBSA to diagnose LA dilation better predicts mortality in this population and has better overall predictive performance across all overweight and obese populations. Using BSA indexing may lead to underappreciation of LA dilation and underestimation of patients at increased risk.

Association of Pulmonary Artery Pulsatility Index With Adverse Cardiovascular Events Across a Hospital-Based Sample.

BACKGROUND: The pulmonary artery pulsatility index (PAPi), calculated from the ratio of the pulmonary artery pulse pressure to right atrial pressure, is a predictor of right ventricular failure after inferior myocardial infarction and left ventricular assist device implantation. Whether PAPi is associated with adverse outcomes across a heterogeneous population is unknown. METHODS: We examined consecutive patients undergoing right heart catheterization between 2005 and 2016 in a hospital-based cohort. Multivariable Cox models were utilized to examine the association between PAPi and all-cause mortality, major adverse cardiac events, and heart failure hospitalizations. RESULTS: We studied 8285 individuals (mean age 63 years, 39% women) with median PAPi across quartiles 1.7, 2.8, 4.2, and 8.7, who were followed over a mean follow-up of 6.7±3.3 years. Patients in the lowest PAPi quartile had a 60% greater risk of death compared with the highest quartile (multivariable-adjusted hazard ratio, 1.60 [95% CI, 1.36-1.88], P<0.001) and a higher risk of major adverse cardiac events and heart failure hospitalizations (hazard ratio, 1.80 [95% CI, 1.56-2.07], P<0.001 and hazard ratio, 2.08 [95% CI, 1.76-2.47], P<0.001, respectively). Of note, patients in quartiles 2 and 3 also had increased risk of cardiovascular events compared with quartile 4 (multivariable P<0.05 for all). CONCLUSIONS: Compared with the highest PAPi quartile, patients in PAPi quartiles 1 to 3 had a greater risk of all-cause mortality, major adverse cardiac events, and heart failure hospitalizations, with greatest risk observed in the lowest quartile. A low PAPi, even at values higher than previously reported, may serve an important role in identifying high-risk individuals across a broad spectrum of cardiovascular disease.

Prognostic importance of the transmitral pressure gradient in mitral annular calcification with associated mitral valve dysfunction.

AIMS: The aim of this study was to define the natural history of patients with mitral annular calcification (MAC)-related mitral valve dysfunction and to assess the prognostic importance of mean transmitral pressure gradient (MG) and impact of concomitant mitral regurgitation (MR). METHODS AND RESULTS: The institutional echocardiography database was examined from 2001 to 2019 for all patients with MAC and MG ≥3 mmHg. A total of 5754 patients were stratified by MG in low (3-5 mmHg, n = 3927), mid (5-10 mmHg, n = 1476), and high (≥10 mmHg, n = 351) gradient. The mean age was 78 ± 11 years, and 67% were female. MR was none/trace in 32%, mild in 42%, moderate in 23%, and severe in 3%. Primary outcome was all-cause mortality, and outcome models were adjusted for age, sex, and MAC-related risk factors (hypertension, diabetes, coronary artery disease, chronic kidney disease). Survival at 1, 5, and 10 years was 77%, 42%, and 18% in the low-gradient group; 73%, 38%, and 17% in the mid-gradient group; and 67%, 25%, and 11% in the high-gradient group, respectively (log-rank P < 0.001 between groups). MG was independently associated with mortality (adjusted HR 1.064 per 1 mmHg increase, 95% CI 1.049-1.080). MR severity was associated with mortality at low gradients (P < 0.001) but not at higher gradients (P = 0.166 and 0.372 in the mid- and high-gradient groups, respectively). CONCLUSION: In MAC-related mitral valve dysfunction, mean transmitral gradient is associated with increased mortality after adjustment for age, sex, and MAC-related risk factors. Concomitant MR is associated with excess mortality in low-gradient ranges (3-5 mmHg) but gradually loses prognostic importance at higher gradients, indicating prognostic utility of transmitral gradient in MAC regardless of MR severity.

Lessons from implementing a combined workflow-informatics system for diabetes management.

Shortcomings surrounding the care of patients with diabetes have been attributed largely to a fragmented, disorganized, and duplicative health care system that focuses more on acute conditions and complications than on managing chronic disease. To address these shortcomings, we developed a diabetes registry population management application to change the way our staff manages patients with diabetes. Use of this new application has helped us coordinate the responsibilities for intervening and monitoring patients in the registry among different users. Our experiences using this combined workflow-informatics intervention system suggest that integrating a chronic disease registry into clinical workflow for the treatment of chronic conditions creates a useful and efficient tool for managing disease.