Predicting Left Atrial Stasis in Patients With AF
All patients during a 6 month time interval undergoing transthoracic and transesophageal echocardiogram were assessed for the presence of criteria allowing admission into the study. Patients with an AF episode lasting for longer than 48 hours and without effective anticoagulation in the preceding 3 weeks were considered eligible for the study except if any of the following exclusion criteria were met: lack of adequate endocardial border definition of the left atrium, presence of prosthetic heart valve or previous valve repair, significant aortic or mitral valve disease (any degree of aortic/mitral valve stenosis or mitral/aortic regurgitation > II/IV) and previous closure of the LAA.
This study was conducted with the approval of our Institution's Ethics Committee, Comissão de Ética da Faculdade de Medicina da Universidade de Coimbra. All subjects provided their informed consent to undergo the necessary investigations and to allow the usage of their data for research purposes, preserving their anonymity.
Preliminary transthoracic echocardiography was performed using standard views (GE Vivid 7 echocardiograph with a M4S probe). The frame rate was set > 60 frames per second. Since all patients were in AF at the time of procedure, all measurements were obtained from an average of 3 cycles. Left atrium volume was measured using the bi-plane area length method. Left ventricle ejection fraction was calculated with the Simpson method. The ratio between indexed left atrial volume and left ventricle ejection fraction, which has shown to be highly accurate at excluding the presence of an LAA thrombus in patients with AF who are candidates for AF catheter ablation or cardioversion, was calculated.
Pulsed-wave Doppler at the tips of the mitral valve was used for measuring early diastolic filling velocity (E). The early diastolic tissue velocity (E') was measured with tissue Doppler imaging of the lateral mitral annulus. E/E' ratio was calculated. Mitral regurgitation was semi-quantitatively assessed by color Doppler across mitral valve and graded as none/trace (0), mild (I/IV) and moderate (II/IV). Individuals with moderately severe (III/IV) and severe (IV/IV) mitral regurgitation were excluded from analysis.
Global longitudinal strain and strain rate of the left ventricle were assessed as previously described by other authors.
Transthoracic images were processed for assessing left atrial deformation through speckle tracking imaging using the GE Health Care EchoPac Dimension software, PC version 108.1.4 (featuring a software for speckle-tracking of the left ventricle) by two cardiologists who were blinded for the transesophageal echocardiogram results and clinical information of the patients.
The left atrium endocardial surface was manually traced using a point-and-click approach in apical four-chamber view, which allowed the automatic definition of a region of interest. This was manually adjusted, if necessary, to better suit the atrium anatomy. The cardiac cycle was demarcated by indicating QRS onset. The region of interest was divided into 6 segments by the software (Figure 1) and the resulting tracking quality for each segment was scored as either acceptable or non-acceptable. Segments classified as "non-acceptable" were rejected by the software and excluded from the analysis. In subjects with adequate image quality, a total of 6 segments were analyzed. Longitudinal strain and strain rate curves for each segment were analyzed and the following parameters collected (Figure 2): peak positive and peak negative strain, peak positive and peak negative strain rate and time to peak-positive strain. Peak-to-peak strain and strain rate were also calculated (i.e., peak positive – peak negative strain and peak positive – peak negative strain rate). These parameters referred to data from the whole cardiac cycle based on the assumption that since the left atrium was fibrillating during the entire cardiac cycle duration, there was no strong rationale for its division in different phases of atrial function (just like there would be no sense in assessing the different phases of ventricular function in a fibrillating ventricle). An average of the 6 segments in three consecutive heart cycles was estimated, except for time-to-peak positive strain, where standard deviation was calculated for three cycles also.
(Enlarge Image)
Figure 1.
Assessed segments in the left atrial region of interest.
(Enlarge Image)
Figure 2.
Left atrial longitudinal deformation and dyssynchrony measurements. Legend: Peak negative strain (PNS) and peak positive strain (PPS) were measured of each segment from baseline to the peak negative or peak positive value of longitudinal strain, respectively; Peak positive strain rate (PPSR) and peak negative strain rate (PNSR) of each segment were measured from baseline to the peak positive or peak negative value of longitudinal strain rate, respectively. The average of all 6 segments during 3 cardiac cycles was calculated. Time to peak positive strain (TPPS) was the time interval measured from the beginning of the cardiac cycle to the timing of peak positive strain; the standard deviation of all 6 segments during 3 cardiac cycles was calculated.
Transesophageal echocardiogram was performed without anaesthesia or sedation using a 6 T phased array multiplane transesophageal probe (frequency 7.0 MHz). The left atrium and LAA were imaged in different tomographic planes to detect the presence of LAA thrombus and spontaneous echocardiographic contrast, which was graded according to the classification (1 to 4+) proposed by Fatkin et al. When dense spontaneous echocardiographic contrast (grade 3+ or 4+) was present and organized into a dynamic and gelatinous, but not solid or well-formed, echodensity present throughout the cardiac cycle, sludge was reported.
LAA flow velocities were assessed with a pulsed Doppler sample placed 1 cm from the entry of the LAA into the body of the left atrium. Emptying and filling velocities were estimated from an average of five well-defined emptying and filling waves.
Patients were divided into 2 groups, according to the type of findings on transesophageal echocardiogram: group I, with LAA thrombus or sludge and group II, without any of these changes.
Comparisons were performed between the two patient groups. Chi-square was used for nominal variables and Student's t-test was used for comparison of continuous variables, where appropriate; the Levene's test was used in order to check the homogeneity of variance; equivalent non-parametric tests were used when Kolmogorov-Smirnov was in favor of absence of normal distribution. Pearson's r correlation coefficient was used for quantifying the degree of association between two quantitative variables. Results with p < 0.05 were regarded as significant.
Univariate analysis was performed using the chi-square test. Predictors from univariate analysis were used for obtaining logistic regression models (using the backward stepwise method likelihood ratio; probability for stepwise = 0.10) that could predict the presence of left atrial thrombi or sludge. Continuous variables which statistically differed between group I and II (or presented a non-significant trend: P < 0.1) were converted into qualitative variables and then used in the logistic regression analysis. Cut-off values were obtained through receiver operating characteristic (ROC) curves which allowed us to define the optimal cutoff point, which was the value combining the higher value of specificity plus sensitivity (Youden index). The Hosmer-Lemeshow summary statistic was used to assess the goodness-of-fit of the models. The coefficients from the obtained model (beta values from the incorporated variables and constant) were used for estimating the probability of event in each patient. Then, the discriminative capability of the obtained probabilities was also assessed through a ROC curve.
PASW Statistics version 18.0 was used for descriptive and inferential statistical analysis.
Inter-observer variability was assessed using Bland-Altman plots with data from the first 7 included patients in the study (average for each of the 6 segments shown), that were separately assessed by the two operators. MedCalc for Windows version 9.2.0.1 was used.
Methods
Patient Selection
All patients during a 6 month time interval undergoing transthoracic and transesophageal echocardiogram were assessed for the presence of criteria allowing admission into the study. Patients with an AF episode lasting for longer than 48 hours and without effective anticoagulation in the preceding 3 weeks were considered eligible for the study except if any of the following exclusion criteria were met: lack of adequate endocardial border definition of the left atrium, presence of prosthetic heart valve or previous valve repair, significant aortic or mitral valve disease (any degree of aortic/mitral valve stenosis or mitral/aortic regurgitation > II/IV) and previous closure of the LAA.
This study was conducted with the approval of our Institution's Ethics Committee, Comissão de Ética da Faculdade de Medicina da Universidade de Coimbra. All subjects provided their informed consent to undergo the necessary investigations and to allow the usage of their data for research purposes, preserving their anonymity.
Preliminary transthoracic echocardiography was performed using standard views (GE Vivid 7 echocardiograph with a M4S probe). The frame rate was set > 60 frames per second. Since all patients were in AF at the time of procedure, all measurements were obtained from an average of 3 cycles. Left atrium volume was measured using the bi-plane area length method. Left ventricle ejection fraction was calculated with the Simpson method. The ratio between indexed left atrial volume and left ventricle ejection fraction, which has shown to be highly accurate at excluding the presence of an LAA thrombus in patients with AF who are candidates for AF catheter ablation or cardioversion, was calculated.
Pulsed-wave Doppler at the tips of the mitral valve was used for measuring early diastolic filling velocity (E). The early diastolic tissue velocity (E') was measured with tissue Doppler imaging of the lateral mitral annulus. E/E' ratio was calculated. Mitral regurgitation was semi-quantitatively assessed by color Doppler across mitral valve and graded as none/trace (0), mild (I/IV) and moderate (II/IV). Individuals with moderately severe (III/IV) and severe (IV/IV) mitral regurgitation were excluded from analysis.
Global longitudinal strain and strain rate of the left ventricle were assessed as previously described by other authors.
Assessment of Left Atrial Deformation by Transthoracic Echocardiogram
Transthoracic images were processed for assessing left atrial deformation through speckle tracking imaging using the GE Health Care EchoPac Dimension software, PC version 108.1.4 (featuring a software for speckle-tracking of the left ventricle) by two cardiologists who were blinded for the transesophageal echocardiogram results and clinical information of the patients.
The left atrium endocardial surface was manually traced using a point-and-click approach in apical four-chamber view, which allowed the automatic definition of a region of interest. This was manually adjusted, if necessary, to better suit the atrium anatomy. The cardiac cycle was demarcated by indicating QRS onset. The region of interest was divided into 6 segments by the software (Figure 1) and the resulting tracking quality for each segment was scored as either acceptable or non-acceptable. Segments classified as "non-acceptable" were rejected by the software and excluded from the analysis. In subjects with adequate image quality, a total of 6 segments were analyzed. Longitudinal strain and strain rate curves for each segment were analyzed and the following parameters collected (Figure 2): peak positive and peak negative strain, peak positive and peak negative strain rate and time to peak-positive strain. Peak-to-peak strain and strain rate were also calculated (i.e., peak positive – peak negative strain and peak positive – peak negative strain rate). These parameters referred to data from the whole cardiac cycle based on the assumption that since the left atrium was fibrillating during the entire cardiac cycle duration, there was no strong rationale for its division in different phases of atrial function (just like there would be no sense in assessing the different phases of ventricular function in a fibrillating ventricle). An average of the 6 segments in three consecutive heart cycles was estimated, except for time-to-peak positive strain, where standard deviation was calculated for three cycles also.
(Enlarge Image)
Figure 1.
Assessed segments in the left atrial region of interest.
(Enlarge Image)
Figure 2.
Left atrial longitudinal deformation and dyssynchrony measurements. Legend: Peak negative strain (PNS) and peak positive strain (PPS) were measured of each segment from baseline to the peak negative or peak positive value of longitudinal strain, respectively; Peak positive strain rate (PPSR) and peak negative strain rate (PNSR) of each segment were measured from baseline to the peak positive or peak negative value of longitudinal strain rate, respectively. The average of all 6 segments during 3 cardiac cycles was calculated. Time to peak positive strain (TPPS) was the time interval measured from the beginning of the cardiac cycle to the timing of peak positive strain; the standard deviation of all 6 segments during 3 cardiac cycles was calculated.
Assessment of Left Atrial Stasis by Transesophageal Echocardiogram
Transesophageal echocardiogram was performed without anaesthesia or sedation using a 6 T phased array multiplane transesophageal probe (frequency 7.0 MHz). The left atrium and LAA were imaged in different tomographic planes to detect the presence of LAA thrombus and spontaneous echocardiographic contrast, which was graded according to the classification (1 to 4+) proposed by Fatkin et al. When dense spontaneous echocardiographic contrast (grade 3+ or 4+) was present and organized into a dynamic and gelatinous, but not solid or well-formed, echodensity present throughout the cardiac cycle, sludge was reported.
LAA flow velocities were assessed with a pulsed Doppler sample placed 1 cm from the entry of the LAA into the body of the left atrium. Emptying and filling velocities were estimated from an average of five well-defined emptying and filling waves.
Patients were divided into 2 groups, according to the type of findings on transesophageal echocardiogram: group I, with LAA thrombus or sludge and group II, without any of these changes.
Statistical Analysis
Comparisons were performed between the two patient groups. Chi-square was used for nominal variables and Student's t-test was used for comparison of continuous variables, where appropriate; the Levene's test was used in order to check the homogeneity of variance; equivalent non-parametric tests were used when Kolmogorov-Smirnov was in favor of absence of normal distribution. Pearson's r correlation coefficient was used for quantifying the degree of association between two quantitative variables. Results with p < 0.05 were regarded as significant.
Univariate analysis was performed using the chi-square test. Predictors from univariate analysis were used for obtaining logistic regression models (using the backward stepwise method likelihood ratio; probability for stepwise = 0.10) that could predict the presence of left atrial thrombi or sludge. Continuous variables which statistically differed between group I and II (or presented a non-significant trend: P < 0.1) were converted into qualitative variables and then used in the logistic regression analysis. Cut-off values were obtained through receiver operating characteristic (ROC) curves which allowed us to define the optimal cutoff point, which was the value combining the higher value of specificity plus sensitivity (Youden index). The Hosmer-Lemeshow summary statistic was used to assess the goodness-of-fit of the models. The coefficients from the obtained model (beta values from the incorporated variables and constant) were used for estimating the probability of event in each patient. Then, the discriminative capability of the obtained probabilities was also assessed through a ROC curve.
PASW Statistics version 18.0 was used for descriptive and inferential statistical analysis.
Inter-observer variability was assessed using Bland-Altman plots with data from the first 7 included patients in the study (average for each of the 6 segments shown), that were separately assessed by the two operators. MedCalc for Windows version 9.2.0.1 was used.
SHARE
