Background
The American Society of Echocardiography recommends calculating left atrial (LA) biplane volume because of its greater accuracy and prognostic value over LA diameter. However, biplane methods are not always feasible. The aim of this study was to assess the correlation between the echocardiographic LA biplane and single-plane volumes and their agreement in the classification of LA size when American Society of Echocardiography cutoffs are applied.
Methods
Two-dimensional echocardiography was performed on the participants of the population-based Cardiovascular Abnormalities and Brain Lesions study. LA volume was calculated by the biplane area-length and single-plane modified Simpson’s methods and validated against three-dimensional echocardiography.
Results
The study sample consisted of 527 participants (mean age 69.6 ± 9.7 years; 61.9% women). Both single-plane and biplane LA volumes correlated well with three-dimensional echocardiography ( r = 0.93, P < .001). The correlation between the single-plane and biplane methods was excellent ( r = 0.95, P < .001; intraclass correlation coefficient, 0.92; 95% confidence interval, 0.80–0.96). Categorical agreement between the single-plane and biplane methods was modest (κ = 0.51; 95% confidence interval, 0.45–0.57; disagreement rate, 26.0%), mainly because of overestimation by the single-plane method. The correction of the single-plane volume by a regression equation improved the agreement (κ = 0.70; 95% confidence interval, 0.64–0.76), but misclassifications remained in 14.0% of cases.
Conclusions
Single-plane and biplane LA volume measurements have strong correlations, but their agreement for categorical classification is suboptimal. Specific cutoff points should be developed for the single-plane method.
Enlargement of the left atrium is an independent predictor of adverse cardiovascular outcomes, and its prognostic value in a variety of cardiovascular diseases, such as atrial fibrillation, myocardial infarction, and dilated and hypertrophic cardiomyopathy, is well documented. The anteroposterior (AP) diameter of the left atrium, assessed by two-dimensional echocardiography, is the most widely used measure of left atrial (LA) size in epidemiologic studies and in clinical practice. However, AP diameter is inaccurate, because it relies on several geometric assumptions and often results in an underestimation of LA size compared with LA volumes. Echocardiographic measures of LA volume rely on fewer geometric assumptions than AP diameter and have been validated against cine computed tomography, contrast ventriculography, and magnetic resonance imaging. The American Society of Echocardiography (ASE) guidelines recommend measuring LA volume using either the biplane area-length formula or the biplane modified Simpson’s rule, because of their high accuracy and their stronger prognostic value compared with linear LA dimension. However, biplane planimetry of the left atrium is not always feasible, because the apical two-chamber view sometimes provides suboptimal LA border visualization.
The aim of our study was to compare a biplane method for LA volume measurement with a measurement obtained from a single-plane four-chamber view in a population-based cohort with a wide range of LA sizes. Besides assessing the correlation and agreement between the two methods, we tested the ability of the single-plane LA volume determination to correctly classify LA size into different categories when the ASE suggested cutoffs are used. Furthermore, for comparison and validation of the two-dimensional methods, we calculated the LA volume by real time three-dimensional (3D) echocardiography in a subgroup of subjects.
Methods
The study was conducted at the Adult Cardiovascular Ultrasound Laboratories of Columbia University Medical Center. The study sample was derived from the National Institutes of Health–sponsored Cardiac Abnormalities and Brain Lesions (CABL) study, whose aim is to assess the relationship between cardiovascular subclinical disease and silent brain infarctions in a community-based cohort. Participants in CABL were drawn from the Northern Manhattan Study (NOMAS), an epidemiologic study carried out in New York City. Extensive details about the population and enrollment of NOMAS have been published previously. Briefly, subjects were eligible if they (1) had never been diagnosed with stroke, (2) were aged ≥ 50 years, and (3) resided in northern Manhattan for ≥3 months in a household with a telephone. The study was approved by the institutional review board of Columbia University Medical Center, and informed consent was obtained from all study participants.
Cardiovascular risk factors were ascertained through direct examination and interview by trained research assistants. Among the variables used in the analyses, hypertension was defined as systolic blood pressure ≥ 140 mm Hg or diastolic blood pressure ≥ 90 mm Hg (mean of two readings) or a patient’s self-reported history of hypertension or antihypertensive medication use. Coronary artery disease (CAD) was defined as a history of myocardial infarction, coronary artery bypass grafting, percutaneous coronary intervention, typical angina, or use of anti-ischemic medications.
Echocardiographic Evaluation
Extensive two-dimensional echocardiography was performed in all participants in the lateral recumbent position, using a commercially available system (iE33; Philips Medical Systems, Andover, MA) equipped with a 2.5-MHz to 3.5-MHz transducer, by a trained sonographer following a standardized protocol. All the exams were stored on digital media for subsequent analysis. Left ventricular (LV) diameters and wall thickness were measured according ASE guidelines, and LV ejection fraction (LVEF) was computed using the modified Simpson’s formula. LVEF was considered abnormal if <50%.
For LA volume assessment, two methods were used (1) the biplane area-length method, using the formula V = 8( A _{1 })( A _{2 })/3π( L ), where A _{1 }and A _{2 }represent the LA planimetry respectively in the four-chamber and two-chamber views, and L is the shortest length from the middle of the plane of the mitral annulus to the superior aspect of the left atrium ( Figure 1 A); (2) modified single-plane Simpson’s rule, assuming the stacked disks are circular, using the formula V = π/4( h )Σ( D ) ^{2 }, where V is volume, h is the height of the disks, and D is the orthogonal axis of the disks ( Figure 1 B). Apical views were optimized to satisfactorily visualize LV and LA walls and to avoid chamber foreshortening. In particular, when needed, in the four-chamber view, the tail of the probe was slightly tilted downward to account for the physiologic angle between the atrial and ventricular long axes, and in the two-chamber view, the probe was adjusted to correctly visualize both anterior and inferior LV and LA walls. The LA endocardial border was traced, and the volumes were calculated by the online software package. The LA appendage and the pulmonary vein confluence were excluded from the LA tracings, and a straight line was traced between the attachment points of the mitral annulus with the valve leaflets. LA volumes were indexed to body surface area. To assess the correlation of LA volume with a widely used measure of LA size, AP diameter from the parasternal long-axis view was also measured and indexed to body surface area. The measurements of the single-plane and the biplane volumes were taken at different times by a single trained sonographer, blinded to the results of the first measurement. The choice of which image to consider for the measurement was left to the reader, because several cine loops of four-chamber and two-chamber views were stored for each patient, and every clip had four cardiac cycles in it. Measurements were taken at end-systole, defined as the frame immediately preceding the mitral valve opening. The average of two consecutive measurements was considered.
In a subset of 80 patients, LA volume measurement at end-systole was also obtained by real-time 3D echocardiography, which was used as a reference to evaluate the accuracy of the two-dimensional methods. A full volume loop was acquired from an apical window using an X3-1 matrix-array transducer over four cardiac cycles. Measurements of 3D LA volume were performed offline (QLAB Advanced Quantification version 7.0; Philips Medical Systems) by an experienced physician (C.R.). The software requires the reader to identify five anatomic landmarks (septal, lateral, anterior, and inferior mitral annulus and posterior wall of the left atrium) at end-diastole and end-systole; subsequently, semiautomated border detection is performed, and LA borders are tracked throughout the entire cardiac cycle. Manual correction on all possible 3D planes is performed by the reader in case of inaccurate endocardial automated detection.
Of the 583 study participants who underwent echocardiographic evaluation, 56 (9.0%) were excluded from analysis because the two-chamber views were suboptimal for LA border tracing. Excluded subjects were significantly older than the others (74.2 ± 9.4 vs 69.4 ± 9.8 years, P < .01) and had a higher prevalence of obesity (42.3% vs 29.0%, P < .05). The final study sample consisted therefore of 527 participants.
Statistical Analysis
All descriptive data are expressed as mean ± SD for continuous variables and as percentages for categorical variables. Correlations between LA size measurements were assessed by Pearson’s correlation coefficient ( r ). Agreement between different measurements of LA volume as continuous variables was assessed by the intraclass correlation coefficient (ICC) for absolute agreement. Pearson’s coefficients and ICCs were obtained in the overall study group and in several demographic and clinical subsets of patients. Mean differences between single-plane and biplane LA volumes were assessed using paired Student’s t tests. Bland-Altman plots were created to calculate the limits of agreement. A linear regression model was used to calculate the predicted biplane volume value from the single-plane volume measurement. Single-plane LA volume was set as the independent variable, and biplane volume was set as the dependent variable, and a regression equation was derived using the slope of the relation and the intercept. The regression equation was y = 0.855 x + 3.71, where y is the predicted biplane volume and x is the single-plane measured volume (standard errors, 0.65 for the constant and 0.013 for the slope). The regression equation was then applied to the single-plane volume to obtain a corrected single-plane measurement.
Categorization of LA size into four groups (normal, mild, moderate, and severe dilation) was performed using the cutoffs suggested by the ASE: ≤28, 29 to 33, 34 to 39, and ≥40 ml/m ^{2 }. Categorical agreement between the single-plane and biplane methods was evaluated using κ statistics. Agreement between the biplane and the corrected single-plane volumes was also assessed. For all statistical analyses, two-tailed P values < .05 were considered significant.
Results
Clinical Characteristics of the Study Sample
The study sample included 527 participants. The clinical characteristics are shown in Table 1 . The mean age was 69.6 ± 9.7 years, and 61.9% were women. Hypertension was present in 68.5%, CAD in 5.9%, and atrial fibrillation in 2.1%. As expected, the large majority of the study participants were of Hispanic ethnic background (71.3%), with a minority of Caucasian (11.0%) and African American (14.2%) participants, reflecting the racial and ethnic composition of the community living in northern Manhattan.
Variable | Value |
---|---|
Age (years) | 69.6 ± 9.7 |
Men | 201 (38.1%) |
Body mass index (kg/m ^{2 }) | 28.1 ± 4.6 |
Hypertension | 361 (68.5%) |
Diabetes | 149 (28.3%) |
CAD | 31 (5.9%) |
Atrial fibrillation | 11 (2.1%) |
LVEF | 63 ± 7.2 |
Left ventricular mass (g/m ^{2 }) | 106.6 ± 26.3 |
Validation of the Biplane and Single-Plane Methods by 3D LA Volume Estimation
Three-dimensional LA volume was measured and compared with the single-plane and biplane methods in a subgroup of 80 participants. The clinical characteristics of this subgroup were not significantly different from the remainder of the study sample (all P values > 0.10, data not shown). Mean LA volumes were 23.8 ± 6.7 ml/m ^{2 }for the 3D method, 24.1 ± 7.2 mL/m ^{2 }for the biplane method, and 25.9 ± 8.2 ml/m ^{2 }for the single-plane method. Correlations with 3D LA volume were excellent both for the biplane ( r = 0.93, P < .01; Figure 2 A) and the single-plane ( r = 0.93, P < .01; Figure 2 B) methods. However, the ICC for 3D LA volume was higher for the biplane method (ICC, 0.93; 95% confidence interval [CI], 0.89–0.95) than for the single-plane method (ICC, 0.88; 95% CI, 0.68–0.95).
Comparison Between Biplane and Single-Plane LA Volumes in the Entire Sample
Reproducibility of the Measurements
Intraobserver variability, calculated as the mean difference between LA volumes in two separate readings on a sample of 20 randomly chosen exams, was 1.25 mL for the biplane method (standard error, 1.08 mL; ICC, 0.95) and 1.60 mL (standard error, 1.18 mL; ICC, 0.95) for the single-plane method. No significant differences were found between the first and second sets of measurements for both the biplane and the single-plane methods ( P = .26 and P = .19, respectively). To test the interobserver variability, LA volumes were reobtained by an experienced reader in 30 randomly chosen patients. The mean difference between the two sets of measurements was 1.48 mL (standard error, 1.30; ICC, 0.90) for the biplane method ( P = .26 for the difference between the two readings) and 2.13 mL (standard error, 1.67; ICC, 0.91) for the single-plane method ( P = .21).
Comparison Between Biplane and Single-Plane LA Volumes
The mean LA volume measured by the biplane area-length method was 25.9 ± 8.3 ml/m ^{2 }, the mean LA volume by the single-plane Simpson’s method was 27.9 ± 9.1 ml/m ^{2 }(mean difference, 1.9 mL/m ^{2 }; P < .01), and the mean AP diameter was 22.3 ± 3.1 mm/m ^{2 }.
The correlation between biplane and single-plane LA volumes was excellent ( r = 0.95, P < .01), and their relation was linear throughout the entire spectrum of volumes ( Figure 3 A). The ICC between the two methods was 0.92 (95% CI, 0.80–0.96). The correlation between LA AP diameter and LA volume was considerably weaker, though still statistically significant ( r = 0.66 with the biplane method, P < .01, Figure 3 B; r = 0.67 with the single-plane method, P < .01). When the study sample was divided into different demographic and clinical subgroups, the correlation and the agreement between biplane and single-plane volume measurements remained strong ( r = 0.95 in men, r = 0.94 in women, r = 0.95 in normotensive subjects, r = 0.95 in hypertensive subjects, r = 0.96 in those with CAD, r = 0.95 in those without CAD, r = 0.94 in those with normal LVEFs, and r = 0.99 in those with reduced LVEFs; all P values < .01; ICC range, 0.91–0.97).
The correlation between single-plane and biplane LA volumes was also analyzed separately in different age strata, and it was strong in all age groups ( r = 0.90 in subjects aged 50–60 years, r = 0.94 in those aged 60–70 years, and r = 0.95 in those aged > 70 years), with the single-plane method yielding bigger volumes than the biplane in all subgroups ( Table 2 ). Both single-plane and biplane LA volumes tended to increase in participants aged > 70 years, but this trend was not present when subjects with hypertension and CAD were excluded from the analysis ( Table 2 ).
Age (years) | ||||
---|---|---|---|---|
50–60 | 60–70 | >70 | P ^{∗ } | |
Overall population | n = 86 | n = 193 | n = 246 | |
Biplane | 25.7 ± 6.3 | 24.4 ± 7.6 | 27.2 ± 9.2 | .004 |
Single-plane | 27.1 ± 6.6 ^{‡ } | 26.5 ± 8.5 ^{‡ } | 29.3 ± 10.2 ^{‡ } | .002 |
Participants without hypertension or CAD | n = 40 | n = 60 | n = 56 | |
Biplane | 26.1 ± 6.4 | 23.5 ± 7.9 | 24.5 ± 6.5 | .21 |
Single-plane | 27.1 ± 5.9 ^{† } | 25.3 ± 8.9 ^{‡ } | 25.9 ± 7.8 ^{‡ } | .53 |