Unsupervised Machine Learning for Assessment of Left Ventricular Diastolic Function and Risk Stratification.

BACKGROUND: The 2016 American Society of Echocardiography guidelines have been widely used to assess left ventricular diastolic function. However, limitations are present in the current classification system. The aim of this study was to develop a data-driven, unsupervised machine learning approach for diastolic function classification and risk stratification using the left ventricular diastolic function parameters recommended in the 2016 American Society of Echocardiography guidelines; the guideline grading was used as the reference standard. METHODS: Baseline demographics, heart failure hospitalization, and all-cause mortality data were obtained for all adult patients who underwent transthoracic echocardiography at Mayo Clinic Rochester in 2015. Patients with prior mitral valve intervention, congenital heart disease, cardiac transplantation, or cardiac assist device implantation were excluded. Nine left ventricular diastolic function variables (mitral E- and A-wave peak velocities, E/A ratio, deceleration time, medial and lateral annular e' velocities and E/e' ratio, and tricuspid regurgitation peak velocity) were used for an unsupervised machine learning algorithm to identify different phenotype clusters. The cohort average of each variable was used for imputation. Patients were grouped according to the algorithm-determined clusters for Kaplan-Meier survival analysis. RESULTS: Among 24,414 patients (mean age, 63.6 ± 16.2 years), all-cause mortality occurred in 4,612 patients (18.9%) during a median follow-up period of 3.1 years. The algorithm determined three clusters with echocardiographic measurement characteristics corresponding to normal diastolic function (n = 8,312), impaired relaxation (n = 11,779), and increased filling pressure (n = 4,323), with 3-year cumulative mortality of 11.8%, 19.9%, and 33.4%, respectively (P < .0001). All 10,694 patients (43.8%) classified as indeterminate were reclassified into the three clusters (n = 3,324, n = 5,353, and n = 2,017, respectively), with 3-year mortality of 16.6%, 22.9%, and 34.4%, respectively. The clusters also outperformed guideline-based grade for prognostication (C index = 0.607 vs 0.582, P = .013). CONCLUSIONS: Unsupervised machine learning identified physiologically and prognostically distinct clusters on the basis of nine diastolic function Doppler variables. The clusters can be potentially applied in echocardiography laboratory practice and future clinical trials for simple, replicable diastolic function-related risk stratification.

Marked Disparity in Regional and Transmural Cardiac Mechanics in the Athlete's Heart.

INTRODUCTION: Regional heterogeneity of the human heart plays an important role in left ventricular (LV) and right ventricular (RV) function and may contribute to enhanced myocardial efficiency in the athlete's heart. PURPOSE: This study comprehensively characterized regional and transmural myocardial tissue deformation (strain) in recreationally active (RA) and endurance-trained (ET) men to determine if regional nonuniformity evolves alongside morphological adaptations associated with endurance training. METHODS: Echocardiography was used to measure LV and RV global, regional (apical, mid, basal) and transmural (endocardial, epicardial) longitudinal strain in 30 endurance-trained (ET) (age, 31 ± 2 yr; body mass index, 23.1 ± 0.5 kg·m; V˙O2peak, 60.2 ± 6.5 mL·kg·min) and 30 RA (age: 29 ± 2 yr; body mass index, 23.4 ± 0.4 kg·m; V˙O2peak: 42.6 ± 4.6 mL·kg·min). Nonuniformity was characterized using apex-to-base and transmural (endocardial-to-epicardial) strain gradients. RESULTS: Global longitudinal strain was similar in ET and RA in the left (-17.4% ± 0.4% vs -17.8% ± 0.5%, P = 0.662) and right ventricle (-25.8% ± 0.8% vs 26.4% ± 1.0%, P = 0.717). The apex-to-base strain gradient was greater in ET than RA in the left (-6.5% ± 0.7% vs -2.7% ± 0.8%, P = 0.001) and right ventricle (-9.6% ± 1.8% vs -3.0% ± 1.6%, P = 0.010). The LV transmural strain gradient was greater than RV in both groups, but similar in ET and RA (-4.7% ± 0.2% vs -4.7% ± 0.2%, P = 0.850), whereas RV transmural strain gradient was greater in ET than RA (-3.4% ± 0.3% vs -1.6% ± 0.4%, P = 0.003). RV apex-to-base and transmural strain gradients correlated with RV end-diastolic area (R = 0.536 & 0.555, respectively, P < 0.01) and V˙O2peak (R = 0.415 & 0.677, respectively, P < 0.01). CONCLUSIONS: Transmural nonuniformity is more pronounced in the left ventricle than the RV free wall; however, RV functional nonuniformity develops markedly after endurance training. Differences in myocardial architecture and exercise-induced wall stress in the left and right ventricles are possible explanations for the marked functional nonuniformity throughout the myocardium and in response to endurance exercise training.

Preoperative Dobutamine Stress Echocardiography and Clinical Factors for Assessment of Cardiac Risk after Noncardiac Surgery.

BACKGROUND: The role of dobutamine stress echocardiography (DSE) in the risk stratification of patients undergoing noncardiac surgery in the current era is unclear. The aim of this study was to evaluate the yield of DSE and the additive role of DSE to clinical criteria for preoperative risk stratification of patients undergoing noncardiac surgery. METHODS: The study included 4,494 patients undergoing DSE ≤90 days before noncardiac surgery. The primary outcome was a composite of postoperative myocardial infarction, cardiac arrest, and all-cause mortality ≤30 days after noncardiac surgery. RESULTS: The overall 30-day postoperative cardiac event rate was 2.3%. The mortality rate was 0.9% overall and 0.7% and 1.3% after normal and abnormal results on DSE, respectively. Among clinical variables, the modified Revised Cardiac Risk Index score demonstrated the strongest association with postoperative risk (P < .001). Patients with Revised Cardiac Risk Index scores of ≥3 had an event rate of 7.5%. The event rates for patients with wall motion score index ≥1.7 at baseline, left ventricular ejection fractions <40% at peak stress, or ischemic thresholds <70% of age-predicted maximal heart rate were 7.1%, 8.6%, and 7.9%, respectively. After adjusting for clinical variables, the overall result of DSE (P < .001), baseline and peak-stress wall motion score index (P < .001 and P = .014, respectively), peak-stress left ventricular ejection fraction (P < .001), and the number of ischemic segments (P = .027) were all associated with postoperative cardiac events. Incremental multivariate analysis demonstrated that an overall abnormal result on DSE, added to clinical variables, was associated with an increased risk for postoperative cardiac events (odds ratio, 2.07; 95% CI, 1.35-3.17; P < .001). CONCLUSIONS: Baseline and peak-stress findings on preoperative DSE add to the prognostic utility of clinical variables for stratifying cardiac risk after noncardiac surgery.

Incidence and Management of Hemopericardium: Impact of Changing Trends in Invasive Cardiology.

OBJECTIVE: As invasive cardiovascular care has become increasingly complex, cardiac perforation leading to hemopericardium is a progressively prevalent complication. We sought to assess the frequency, etiology, and outcomes of hemorrhagic pericardial effusions managed through a nonsurgical echo-guided percutaneous strategy. PATIENTS AND METHODS: Over a 10-year period (January 1, 2007, to December 31, 2016), 1097 unique patients required pericardiocentesis for clinically important pericardial effusions. Of these 411 had drainage of hemorrhagic effusions (defined as a pericardial hemoglobin level >50% of serum hemoglobin or frank blood in the setting of cardiac perforation). Clinical characteristics, echocardiographic data, details of the procedure, and outcomes were determined. RESULTS: Median patient age was 67 years (interquartile range, 56-76 years), and 60% were men. The procedure was emergent in 83% and elective in 17%. The site of pericardiocentesis was determined by echo-guidance in all: 68% from the left para-apical region, 18% from the left or right parasternal areas, and 14% were subxyphoid. Half (n=215 [52%]) occurred after cardiac perforation with percutaneous interventional procedure (ablation, n=94; device lead implantation, n=65; percutaneous coronary intervention, n=22; other, n=34), whereas 30% followed cardiac or thoracic surgery. Pericardial fluid volume drained was 546±440 mL. In 94% of cases, echo-guided pericardiocentesis was the only treatment of the effusion needed, whereas definitive surgery was required in 25 (6%) cases for persistent bleeding or acute management of the underlying etiology. There was no procedural mortality. Late mortality was better for hemorrhagic effusions compared with a contemporary cohort with nonhemorrhagic effusions. CONCLUSION: Echocardiographic guidance allows rapid successful pericardiocentesis in the setting of hemopericardium related to microperforation with interventional procedures, malignancy, or pericarditis, with most not requiring surgical intervention. Surgery should remain the first-line approach for aortic dissection or myocardial rupture.

Outcome prediction in sepsis: speckle tracking echocardiography based assessment of myocardial function.

INTRODUCTION: Speckle tracking echocardiography (STE) is a relatively novel and sensitive method for assessing ventricular function and may unmask myocardial dysfunction not appreciated with conventional echocardiography. The association of ventricular dysfunction and prognosis in sepsis is unclear. We sought to evaluate frequency and prognostic value of biventricular function, assessed by STE in patients with severe sepsis or septic shock. METHODS: Over an eighteen-month period, sixty patients were prospectively imaged by transthoracic echocardiography within 24 hours of meeting severe sepsis criteria. Myocardial function assessment included conventional measures and STE. Association with mortality was assessed over 12 months. RESULTS: Mortality was 33% at 30 days (n = 20) and 48% at 6 months (n = 29). 32% of patients had right ventricle (RV) dysfunction based on conventional assessment compared to 72% assessed with STE. 33% of patients had left ventricle (LV) dysfunction based on ejection fraction compared to 69% assessed with STE. RV free wall longitudinal strain was moderately associated with six-month mortality (OR 1.1, 95% confidence interval, CI, 1.02-1.26, p = 0.02, area under the curve, AUC, 0.68). No other conventional echocardiography or STE method was associated with survival. After adjustment (for example, for mechanical ventilation) severe RV free wall longitudinal strain impairment remained associated with six-month mortality. CONCLUSION: STE may unmask systolic dysfunction not seen with conventional echocardiography. RV dysfunction unmasked by STE, especially when severe, was associated with high mortality in patients with severe sepsis or septic shock. LV dysfunction was not associated with survival outcomes.

Outcome prediction by quantitative right ventricular function assessment in 575 subjects evaluated for pulmonary hypertension.

BACKGROUND: Although right ventricular (RV) dysfunction is a major determinant of outcome in patients with pulmonary hypertension (PH), the optimal measure of RV function is poorly defined. We hypothesized that RV strain measured by speckle-tracking echocardiography predicts outcome in PH over a broad range of pulmonary pressures. METHODS AND RESULTS: Prospective peak RV longitudinal systolic strain measurement was performed on 575 patients (mean age, 56 ± 18 years; 63% women) referred for echocardiography for known or suspected PH. Survival status was assessed over a median of 16.5 (interquartile range, 7.6-20.0) months. There were 406 patients with PH (71%) (74% group 1, 14% group 3, and 12% group 4) and 169 patients without evidence of PH (29%). Among patients with PH, 46% were World Health Organization functional class III-IV. The mean RV strain was -21.2 ± 6.7% for all patients. RV strain declined with worse functional class, shorter 6-minute walk distances, higher N-terminal pro-B-type natriuretic peptide levels, and the presence of right heart failure. RV strain predicted outcome across pulmonary pressures and groups of PH. Eighteen-month survival was 92%, 88%, 85%, and 71% according to RV strain quartile (P<0.001), with a 1.46 higher risk of death (95% confidence interval, 1.05-2.12) per 6.7% decline in RV strain. RV strain predicted survival when adjusted for pulmonary pressure, pulmonary vascular resistance, and right atrial pressure and provided incremental prognostic value over conventional clinical and echocardiographic variables. CONCLUSIONS: Quantitative assessment of RV free-wall systolic strain is feasible and is a powerful predictor of the clinical outcome of patients with known or suspected PH.

Role of serial quantitative assessment of right ventricular function by strain in pulmonary arterial hypertension.

The aim of this study was to assess whether serial quantitative assessment of right ventricular (RV) function by speckle-based strain imaging is affected by pulmonary hypertension-specific therapies and whether there is a correlation between serial changes in RV strain and clinical status. RV longitudinal systolic function was assessed using speckle-tracking echocardiography in 50 patients with pulmonary arterial hypertension (PAH) before and after the initiation of therapy. The mean interval to follow-up was 6 ± 2 months. Subsequent survival was assessed over 4 years. Patients demonstrated a mean increase in RV systolic strain from -15 ± 5 before to -20 ± 7% (p = 0.0001) after PAH treatment. Persistence of or progression to a severe reduction in free wall systolic strain (<-12.5%) at 6 months was associated with greater disease severity (100% were in functional class III or IV vs 42%, p = 0.005), greater diuretic use (86% vs 40%, p = 0.02), higher mean pulmonary artery pressure (67 ± 20 vs 46 ± 17 mm Hg, p = 0.006), and poorer survival (4-year mortality 43% vs 23%, p = 0.002). After adjusting for age, functional class, and RV strain at baseline, patients with ≥ 5% improvement in RV free wall systolic strain had a greater than sevenfold lower mortality risk at 4 years (hazard ratio 0.13, 95% confidence interval 0.03 to 0.50, p = 0.003). In conclusion, serial echocardiographic assessment of RV longitudinal systolic function by quantitative strain imaging independently predicts clinical deterioration and mortality in patients with PAH after the institution of medical therapy.

Pulmonary hypertension: diagnosis and management.

Pulmonary arterial hypertension is a progressive, symptomatic, and ultimately fatal disorder for which substantial advances in treatment have been made during the past decade. Effective management requires timely recognition and accurate diagnosis of the disorder and appropriate selection among therapeutic alternatives. Despite progress in treatment, obstacles remain that impede the achievement of optimal outcomes. The current article provides an overview of the pathobiologic mechanisms of pulmonary arterial hypertension, including genetic substrates and molecular and cellular mechanisms, and describes the clinical manifestations and classification of pulmonary arterial hypertension. The article also reviews established approaches to evaluation and treatment, with emphasis on the appropriate application of calcium channel blockers, prostacyclin analogues, endothelin receptor antagonists, and phosphodiesterase 5 inhibitors. In addition, the authors discuss unresolved issues that may complicate patient management, such as the clinical importance of mild or exercise-related pulmonary arterial hypertension, and they identify avenues by which treatment may advance in the future through the use of combination treatment, outcomes assessment, and exploration of alternative pharmacologic strategies.