Spontaneous plaque rupture in mouse models of atherosclerosis is controversial, although numerous studies have discussed so-called “vulnerable plaque” phenotypes in mice. We compared the morphology and biomechanics of two acute and one chronic murine model of atherosclerosis to human coronaries of the thin-cap fibroatheroma (TCFA) phenotype. Our acute models were apolipoprotein E-deficient (ApoE −/− ) and LDL receptor-deficient (LDLr −/− ) mice, both fed a high-fat diet for 8 wk with simultaneous infusion of angiotensin II (ANG II), and our chronic mouse model was the apolipoprotein E-deficient strain fed a regular chow diet for 1 yr. We found that the mouse plaques from all three models exhibited significant morphological differences from human TCFA plaques, including the plaque burden, plaque thickness, eccentricity, and amount of the vessel wall covered by lesion as well as significant differences in the relative composition of plaques. These morphological differences suggested that the distribution of solid mechanical stresses in the walls may differ as well. Using a finite-element analysis computational solid mechanics model, we computed the relative distribution of stresses in the walls of murine and human plaques and found that although human TCFA plaques have the highest stresses in the thin fibrous cap, murine lesions do not have such stress distributions. Instead, local maxima of stresses were on the media and adventitia, away from the plaque. Our results suggest that if plaque rupture is possible in mice, it may be driven by a different mechanism than mechanics.
Pediatric obesity is a growing public health problem, which is associated with increased risk of cardiovascular disease and premature death. Left ventricular (LV) remodeling (increased myocardial mass and thickness) and contractile dysfunction (impaired longitudinal strain) have been documented in obese children, but little attention has been paid to the right ventricle (RV). We hypothesized that obese/overweight children would have evidence of RV remodeling and contractile dysfunction.
Background Measures of left ventricular cardiac mechanics such as strains and torsion are becoming increasingly important for assessing heart function. Cardiac magnetic resonance (CMR) can be used to quantify cardiac mechanics using several methods such as tagged CMR or cine Displacement Encoding with Stimulated Echoes (DENSE). These images are generally acquired during an end-expiratory breath-hold. Unfortunately, it is difficult for subjects to hold their breath at the exact same position when undergoing a series of breath-holds during a typical CMR study. For example, end-expiratory breath-hold positions have an average range of about 8 millimeters (mm). The effects of different breath-hold positions on measures of cardiac mechanics have not been investigated. We hypothesized that the normal variability in breath-hold positions would significantly affect the quantification of left ventricular strains and torsion.
Objectives: Patients with repaired tetralogy of Fallot (rTOF) suffer from progressive ventricular dysfunction decades after their initial surgical repair. This study sought to determine whether measures of ventricular strain and dyssynchrony could predict deterioration of ventricular function in patients with rTOF. Methods: A database search identified all patients at a single institution with rTOF who underwent cardiac magnetic resonance (CMR) at least twice, greater than 6 months apart, without intervening surgical or catheter procedures. Five primary potential predictors were derived from the first CMR using a custom feature tracking algorithm: left (LV), right (RV) and inter-ventricular dyssynchrony, and LV and RV peak circumferential strains. Three outcomes were defined as measures of progression over time: RV end-diastolic volume, RV ejection fraction (EF), and LV EF. A multivariate linear mixed model, subject to backward elimination of extraneous terms, was fit to investigate relationships between predictors and outcomes. Ten potential confounders (see table footnote) measured at baseline were also included in the model. Results: A cohort of 153 patients with rTOF (23±14 years, 50% male) were included. The mean time between the first and last CMR was 2.9±1.3 years. None of the 5 primary predictors were significantly associated with change over time in the 3 outcomes in the multivariate model, though RV circumferential strain was associated with RV EF at baseline (table). Only 1-16% of the variability in the change over time in the 3 outcomes could be explained by the predictors in the multivariate model. Conclusions: In patients with rTOF, measures of ventricular dyssynchrony and strain were not significantly related to changes in ventricular size and function over time in a multivariate analysis. The ability to predict deterioration in ventricular function using all potential predictors is limited.
Background Displacement Encoding with Stimulated Echoes (DENSE) directly measures tissue displacements and can be used to quantify cardiac mechanics. Multi-dimensional DENSE results in lengthy scans that require respiratory gating, acquiring data only while the diaphragm is within a prespecified “acceptance window.” Because it is not possible to perform respiratory gating during data acquisition, DENSE can employ the following respiratory gating strategies: 1) acquire data and keep it if the diaphragm is inside the window after acquisition (retrospective) 2) acquire data and keep it only if the diaphragm was inside the window right before data acquisition (prospective) or 3) a combination of retrospective and prospective where the diaphragm must be inside the window both before and after data acquisition (combined) (Fig 1). Combined respiratory gating is not used often because more data is discarded, resulting in longer scan times. It is possible, however, that with only retrospective or prospective respiratory gating, the diaphragm may not be within the acceptance window for the entirety of data acquisition (i.e. drifting into or out of the window), negatively affecting image quality. We hypothesized that the combined respiratory gating would result in significantly different estimates of left ventricular strains compared to either retrospective or prospective respiratory gating.
Despite marked benefits in many heart failure patients, a considerable proportion of patients treated with cardiac resynchronization therapy (CRT) fail to respond appropriately. Recently, a "U-shaped" (type II) wall motion pattern identified by cardiovascular magnetic resonance (CMR) has been associated with improved CRT response compared to a homogenous (type I) wall motion pattern. There is also evidence that a left ventricular (LV) lead localized to the latest contracting LV site predicts superior response, compared to an LV lead localized remotely from the latest contracting LV site. We prospectively evaluated patients undergoing CRT with pre-procedural CMR to determine the presence of type I and type II wall motion patterns and pre-procedural echocardiography to determine end systolic volume (ESV). We assessed the final LV lead position on post-procedural fluoroscopic images to determine whether the lead was positioned concordant to or remote from the latest contracting LV site. CRT response was defined as a ≥ 15 % reduction in ESV on a 6 month follow-up echocardiogram. The study included 33 patients meeting conventional indications for CRT with a mean New York Heart Association class of 2.8 ± 0.4 and mean LV ejection fraction of 28 ± 9 %. Overall, 55 % of patients were echocardiographic responders by ESV criteria. Patients with both a type II pattern and an LV lead concordant to the latest contracting site (T2CL) had a response rate of 92 %, compared to a response rate of 33 % for those without T2CL (p = 0.003). T2CL was the only independent predictor of response on multivariate analysis (odds ratio 18, 95 % confidence interval 1.6-206; p = 0.018). T2CL resulted in significant incremental improvement in prediction of echocardiographic response (increase in the area under the receiver operator curve from 0.69 to 0.84; p = 0.038). The presence of a type II wall motion pattern on CMR and a concordant LV lead predicts superior CRT response. Improving patient selection by evaluating wall motion pattern and targeting LV lead placement may ultimately improve the response rate to CRT.
Background The amount and location of left ventricular (LV) mechanical dyssynchrony affects an individual’s ability to respond positively to cardiac resynchronization therapy (CRT) [Bax et al JACC 2005]. By using high temporal resolution short-axis cines, it is possible to derive radial motion curves throughout the LV. These radial motion curves can be used to create maps showing dyssynchronous regions in patients enrolled for CRT. The objective of this study was to characterize the size and location of areas of mechanical dyssynchrony in patients scheduled for CRT by comparing their radial wall motion curves to radial motion curves from normal subjects.
Predicting future clinical events helps physicians guide appropriate intervention. Machine learning has tremendous promise to assist physicians with predictions based on the discovery of complex patterns from historical data, such as large, longitudinal electronic health records (EHR). This study is a first attempt to demonstrate such capabilities using raw echocardiographic videos of the heart. We show that a large dataset of 723,754 clinically-acquired echocardiographic videos (~45 million images) linked to longitudinal follow-up data in 27,028 patients can be used to train a deep neural network to predict 1-year mortality with good accuracy (area under the curve (AUC) in an independent test set = 0.839). Prediction accuracy was further improved by adding EHR data (AUC = 0.858). Finally, we demonstrate that the trained neural network was more accurate in mortality prediction than two expert cardiologists. These results highlight the potential of neural networks to add new power to clinical predictions.
Predicting future clinical events helps physicians guide appropriate intervention. Machine learning has tremendous promise to assist physicians with predictions based on the discovery of complex patterns from historical data, such as large, longitudinal electronic health records (EHR). This study is a first attempt to demonstrate such capabilities using raw echocardiographic videos of the heart. We show that a large dataset of 723,754 clinically-acquired echocardiographic videos (~45 million images) linked to longitudinal follow-up data in 27,028 patients can be used to train a deep neural network to predict 1-year mortality with good accuracy (area under the curve (AUC) in an independent test set = 0.839). Prediction accuracy was further improved by adding EHR data (AUC = 0.858). Finally, we demonstrate that the trained neural network was more accurate in mortality prediction than two expert cardiologists. These results highlight the potential of neural networks to add new power to clinical predictions.
