Assessing left ventricular function is a common indication for echocardiography. It generally requires expert echocardiographer estimation and is somewhat subjective and prone to reader discordance. Mitral annular plane systolic excursion (MAPSE) has been suggested as a surrogate measurement for left ventricular function. The aim of this study was to examine the accuracy of MAPSE for predicting left ventricular ejection fraction (EF) on the basis of a large cohort of consecutive echocardiograms.
The study design was a retrospective analysis of 600 two-dimensional echocardiographic studies performed in a single laboratory. MAPSE measurement was performed by an untrained observer and compared with the EF as determined by an expert echocardiographer. The first 300 studies served as a calibration cohort to establish an algorithm for predicting EF on the basis of MAPSE measurement. The following 300 studies served as a verification cohort to test the accuracy of the established algorithm.
Using the first 300 studies, an algorithm was developed to predict EF. Cutoff values for normal EF (≥11 mm for women and ≥13 mm for men) and severely reduced EF (<6 mm for men and women) were identified. For the intermediate-range MAPSE values, a gender-specific regression equation was calculated to generate a predicted EF. Using this algorithm, predicted EFs were determined for the 300 patients in the verification cohort. By comparing the predicted EF and the expert-reported EF, positive and negative predictive values, sensitivity (73%–92%), specificity (81%–100%), and accuracy (82%–86%) of MAPSE for predicting EF were calculated.
MAPSE measurement by an untrained observer was found to be a highly accurate predictor of EF.
Assessing left ventricular systolic function is a common indication for transthoracic echocardiography. Stroke volume determination via echocardiography has been clinically relevant since the 1960s. Left ventricular function is often expressed as an assessment of left ventricular ejection fraction (EF), which over time has been determined using numerous methods. Each of these requires an expert physician echocardiographer and is somewhat subjective and prone to reader discordance and variability. Also, assessing left ventricular EF using currently available techniques is highly dependent on adequate endocardial resolution and the technical quality of the echocardiographic study.
During systole, both longitudinal and circumferential fibers contribute to myocardial contraction. Gibson and colleagues have done extensive work studying the importance of longitudinal fiber shortening and its relationship to left ventricular function. He and others demonstrated that the movement of the mitral annulus toward the apex is a result of long-fiber contraction. During diastole, the annulus moves back away from the apex. Left ventricular apical motion is limited throughout the cardiac cycle, such that the distance between the apex of the heart and the chest surface is almost constant during contraction. The magnitude of the displacement of the mitral annulus during myocardial contraction can be measured from M-mode images of the mitral annulus ( Figure 1 ).
Mitral annular plane systolic excursion (MAPSE), also referred to as mitral annular motion or atrioventricular displacement, was measured as early as 1967, when Zaky et al. described a “curve contour” using M-mode echocardiography through the mitral ring, which measured 1.6 ± 0.4 cm. They found “deviations” from this normal value in the movement of the mitral ring in patients with heart disease. MAPSE more recently has been suggested as a surrogate measurement for left ventricular function. Some have shown linear correlations between expert-measured EF and MAPSE; one study showed that a MAPSE value of <7 mm had sensitivity of 92% and specificity of 67% for detecting severe left ventricular dysfunction. A separate study demonstrated that a MAPSE value of <12 mm had 90% sensitivity and 88% specificity for the detection of EF <50%. Others have shown that MAPSE measurements correlate well with other techniques for left ventricular functional assessment, including three-dimensional echocardiography and cardiac magnetic resonance imaging (MRI). Tsang et al. studied the correlation of MAPSE, as derived from speckle-tracking echocardiography, with MRI-determined EF. They found a very strong correlation using this alternative MAPSE technique, suggesting a true inherent strong relation between mitral annular motion and global left ventricular systolic function.
Multiple studies have shown that a decrease in MAPSE correlates with many factors affecting left ventricular function, including atrial fibrillation, myocardial infarction, dilated cardiomyopathy, and age.
The major benefit of using MAPSE for left ventricular function assessment is in the simplicity of the measurement. It is a simple, one-dimensional measurement that can be easily performed by even novice practitioners with little training in echocardiography. Also, MAPSE measurement is much less dependent on endocardial resolution and can be performed even in technically challenging studies. Our hypothesis was that MAPSE, measured by an untrained observer, can accurately predict EF.
The study design was a retrospective analysis of echocardiograms obtained for clinical indications. Six hundred consecutive two-dimensional (2D) echocardiographic studies done in the echocardiography laboratory at Lenox Hill Hospital (an urban teaching hospital) were included in our study (Philips iE33 xMATRIX Ultrasound System; Philips Medical Systems, Andover, MA). M-mode echocardiography through the mitral valve annulus, from the apical four-chamber view at both the septal and lateral annuli, is routinely performed and recorded in all studies in our laboratory. No specific image orientation was used for the purpose of MAPSE measurement; the images were optimized for routine four-chamber view evaluation. The M-line was positioned through the medial and lateral annulus and included the tissue-blood border for easy identification of the motion of the annulus. The trough of the motion was defined as the end-diastolic position of the annulus, measured at the tip of the QRS complex. The peak was defined as the maximal systolic excursion point. MAPSE measurement was performed by a fourth-year medical student, blinded to the official report of the echocardiogram. The student was initially trained by the attending echocardiographer to accurately measure MAPSE, identifying the required images and the trough and peak of annular motion, and using the calipers on the digitized system. Once the student demonstrated adequate skill at performing these measurements, the student’s independent MAPSE measurements began for the study purpose.
The average MAPSE value (mean of septal and lateral values) was obtained for each patient and used in the analysis. Each study was interpreted (clinically) by an expert echocardiographer, providing an expert estimation of the EF that was based on “eyeballing,” with or without the use of Simpson’s method of disks. Because the purpose of the study was to assess how MAPSE measurement performs relative to an expert interpretation of an echocardiogram, there was no prerequisite requirement for the readers as to how to determine the EF.
Baseline patient characteristics were collected for all studies; these included age, gender, height, weight, body surface area, and the technical quality of the study. Technical quality was graded on a numeric scale (1 = poor, 2 = fair, 3 = good, and 4 = excellent).
The first 300 echocardiograms served as a calibration cohort to help establish MAPSE’s correlation with EF in our laboratory. These studies were analyzed to determine appropriate MAPSE cutoff values for varying tiers of EF.
The following 300 echocardiograms served as a validation cohort. Using the algorithm we developed from the calibration cohort, MAPSE measurement was used by the medical student to predict left ventricular EF. The predicted EF was then compared with the expert-reported assessment. On the basis of this comparison, the sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of MAPSE measurements for EF were determined.
Data were examined for normality; statistical tests and presentation of results were chosen accordingly. Continuous normally distributed data were analyzed using unpaired t tests and are presented as mean ± SD. A simple linear regression procedure was undertaken using MAPSE values as the predictor and EF values as the criterion variable. Diagnostic properties (sensitivity, specificity, positive predictive value, negative predictive values, and total accuracy) were calculated for MAPSE cutoff values on the basis of the performance of these values in the verification cohort. Two-tailed tests of significance are reported, and P values < .05 were considered a priori to indicate statistical significance. Statistical analyses were performed using SPSS version 20 (SPSS, Inc., Chicago, IL).
Interobserver and intraobserver variability in MAPSE measurements were tested by having 20 random studies read by an independent observer, as well as remeasuring MAPSE by a single observer. Interobserver variability was calculated by averaging the values obtained and then dividing the absolute difference between the two observers’ measurements by the mean of the two measurements for every single patient from the randomly selected group. Similarly, intraobserver variability was calculated from the difference between the two measurements by the same reader.
This study was approved by the institutional review board for human research.
Three hundred consecutive 2D echocardiographic studies were reviewed by the medical student, and MAPSE values were measured in all 300 patients. No patients were excluded from data analysis. Data recorded for each patient are listed in Table 1 . Of all the patient characteristics documented, gender had the greatest correlation with EF. Although MAPSE values were similar between genders, we found that on average, a given MAPSE value correlated with a higher EF in women. Given this observation, for establishing an algorithm for determining EF from MAPSE, men and women were analyzed separately.
( n = 300)
( n = 147)
( n = 153)
|Age (y)||65.9 ± 17.4||63.2 ± 15.6||68.1 ± 18.8||.015|
|Age range (y)||18–100||20–100||18–100||—|
|Body surface area (m 2 )||1.86 ± .29||2.02 ± 0.26||1.71 ± 0.22||<.0001|
|Average MAPSE (mm)||10.9 ± 3.0||10.7 ± 3.4||10.9 ± 2.7||.53|
|Reported EF (%)||56.8 ± 13.4||53 ± 15||60 ± 10||<.0001|
|Average image quality ∗||2.82 ± 0.54||2.88 ± 0.60||2.80 ± 0.52||.223|
Our initial analysis of the 300 studies in the calibration cohort (in accordance with data from previous research ) showed that MAPSE values ≥ 13 mm in men consistently predicted a normal or increased EF. Similarly, we found that in women, MAPSE values ≥ 11 mm consistently predicted a normal or increased EF. Likewise, we found that a MAPSE value < 6 mm (for both men and women) served as an appropriate cutoff for predicting severely depressed EF (≥30%).
Gender-specific scatterplots were constructed to evaluate MAPSE versus EF for those patients who had MAPSE measurements between the previously mentioned cutoff values ( Figure 2 ). A regression line was obtained from each scatterplot; for men, 4.8 × MAPSE (mm) + 5.8 ( R = 0.644), and for women, 4.2 × MAPSE (mm) + 20 ( R = 0.470).
Given these data, we created an algorithm for predicting EF on the basis of MAPSE measurement. For women, a MAPSE value ≥ 11 mm determined a normal EF (≥55%), while in men, this value was 13 mm. A MAPSE value < 6 mm in either gender was considered to reflect a severely depressed EF (≤30%). A MAPSE measurement between these cutoff values was used in the gender-specific regression equation, and a predicted EF was calculated.
Another cohort of 300 consecutive 2D echocardiographic studies was reviewed by the medical student. Baseline patient characteristics, as well as general quantitative and qualitative data for the verification cohort, are listed in Table 2 . There were no significant differences between the calibration cohort and the verification cohort. Similarly to the calibration cohort, a difference in average EF between men and women despite lack of difference in average MAPSE was noted in the verification cohort ( Table 3 ). MAPSE values were measured by the medical student, blinded to the expert report. Using this measurement, a predicted EF was determined on the basis of the algorithm established in the calibration cohort.
|Variable||Training cohort |
( n = 300)
|Verification cohort |
( n = 300)
|Age (y)||65.87 ± 17.4||63.86 ± 17.17||.098|
|Age range (y)||18–100||22–94||—|
|Body surface area (m 2 )||1.86 ± .29||1.89 ± .27||.15|
|Average MAPSE (mm)||10.9 ± 3.0||11.3 ± 3.06||.051|
|Reported EF (%)||56.8 ± 13.4||56.47 ± 13.08||.378|
|Average image quality ∗||2.82 ± .54||2.82 ± .53||.316|