The assessment of left ventricular (LV) systolic function using conventional echocardiographic measures is problematic in the setting of mitral regurgitation (MR) given that altered loading conditions can mask underlying ventricular dysfunction. The purpose of this study was to characterize LV function and deformation before and after effective mitral valve repair or replacement to determine echocardiographic measures associated with early postoperative myocardial dysfunction.
Baseline LV function was assessed retrospectively by conventional echocardiography and speckle-tracking strain analysis pre- and postoperatively in patients diagnosed with MR between January 2000 and March 2013, excluding patients with less than mild to moderate MR preoperatively, left-sided obstructive lesions, large septal defects, or more than mild MR postoperatively.
Forty-six pediatric patients were evaluated (average age, 8.2 ± 6.4 years). Thirteen patients had normal preoperative ejection fractions but significant postoperative dysfunction (defined as an ejection fraction < 50%). Compared with the 33 patients with normal postoperative function, age (11.5 ± 7.1 vs 7.3 ± 5.7 years, P = .04), global circumferential strain (−13.2 ± 5.6% vs −17.1 ± 4.6%, P = .02), and global circumferential strain rate (−0.94 ± 0.40 vs −1.36 ± 0.42 sec −1 , P = .004) were found to be statistically different. Using receiver operating characteristic curves, an older preoperative age (area under the curve, 0.67; P = .03), lower global circumferential strain magnitude (area under the curve, 0.74; P = .007), and lower global circumferential strain rate magnitude (area under the curve, 0.80; P = .0004) were determined to be factors associated with early postoperative LV dysfunction after surgical repair of MR.
Strain measurements may be useful as part of the echocardiographic assessment of patients with MR and can guide timing for surgical repair in the pediatric population.
Mitral regurgitation (MR) is a common finding in the pediatric congenital and acquired heart disease population, with varying etiologies. In patients with MR, assessment of left ventricular (LV) systolic function is problematic given that altered loading conditions can mask underlying ventricular dysfunction. Thus, conventional measurements of function such as ejection fraction may not reflect true myocardial contractility.
Myocardial deformation evaluation is often used in adults as an additional measure of LV function. Speckle-tracking-derived myocardial strain assessment is useful in that it provides an angle-independent measure of global and regional myocardial deformation, which has been shown to correlate with normal and abnormal ventricular function.
Identifying LV dysfunction early is crucial to allow timely surgery to prevent early postoperative dysfunction and improve survival after surgical correction. The purposes of this study were to characterize LV function and deformation using conventional measures of function and strain analysis using speckle-tracking echocardiography before and after effective mitral valve repair or replacement and to determine preoperative echocardiographic measures associated with postoperative myocardial dysfunction. We hypothesized that global and regional measurements of strain before valvular repair or replacement would be decreased and can identify patients at risk for ventricular failure, which may be helpful in guiding the timing of surgical intervention.
The Lucile Packard Children’s Hospital Heart Center database was queried to retrospectively identify all eligible patients diagnosed with MR between January 2000 and March 2013. Patients with less than mild to moderate MR preoperatively, coexistent left-sided obstructive lesions such as mitral or aortic stenosis, large septal defects, or more than mild MR postoperatively were excluded from this study. Criteria for inclusion were thus pediatric patients ≤21 years of age with at least mild to moderate MR preoperatively and adequate imaging for analysis on preoperative and initial postoperative transthoracic echocardiography (TTE).
Each patient’s diagnosis, gender, date of birth, height, weight, body surface area (calculated using the formula of Du Bois and Du Bois), dates of surgical procedure, types of surgical procedure, period between surgery and postsurgical TTE, and type of prior surgical intervention were obtained from the electronic medical record. Postsurgical TTE included analysis both initially and, if available, 4 to 15 weeks postoperatively.
The etiology and duration of MR before surgical intervention were recorded, with the duration defined as either acute (present for <6 months) or chronic (present for ≥6 months). To account for the influence of varying surgical technique over the 13-year period, aortic cross-clamp time and number of cardioplegia doses were obtained. The percentage of patients with postoperative LV dysfunction was also measured for two different surgical eras, 2000 to 2006 and 2007 to 2013.
Echocardiographic studies were routinely performed in all patients as part of their pre- and postoperative evaluations. Images were acquired according to American Society of Echocardiography guidelines and stored on our institution’s secure server. The ultrasound equipment used for the echocardiographic studies was either the Siemens Sequoia C512 (Siemens Medical Solutions USA, Inc, Mountain View, CA) or the Philips iE33 (Philips Medical Systems, Bothell, WA).
For each patient, pre- and postoperative transthoracic echocardiographic studies were selected before and after mitral valve repair or replacement at our institution. The primary investigator made offline measurements using the syngo Dynamics workstation (Siemens Medical Solutions USA, Inc; syngo Dynamics Solutions, Ann Arbor, MI).
MR was graded by the vena contracta jet width, left atrial–to–aortic valve ratio, and pulmonary venous Doppler profile. While blinded to each patient’s clinical information, pre- and post-operative MR grading was conducted ≥30 days apart by a senior investigator. MR was graded as mild, moderate, or severe on the basis of defined criteria by the American Society of Echocardiography’s Nomenclature and Standards Committee and the Task Force on Valvular Regurgitation. Inclusion criteria for the final analyses were therefore patients with more than mild MR preoperatively with complete correction after surgery with no more than mild postoperative MR. Given the retrospective nature of the study, several studies were excluded for inadequate pre- and/or postoperative imaging.
LV ejection fraction (LVEF) was measured using the 5/6 area × length method. LV end-diastolic volume (LVEDV), as derived by the 5/6 area × length method, was indexed to body surface area for further comparison. LV end-diastolic dimension (LVEDD), measured by M-mode echocardiography in the short-axis view, was reported with standard Z scores. Velocity of circumferential fiber shortening (VCFc), circumferential wall stress (WSc), and meridional wall stress were calculated for all studies by the primary investigator. Wall stress calculations used noninvasive systolic blood pressure measurements performed at the time of echocardiography. Pre- and postoperative offline analyses were performed separately to maintain objectivity and reduce bias. An LVEF < 50% was defined as the threshold for ventricular dysfunction. Normal LVEF preoperatively was not a requirement for inclusion.
A second investigator performed measurements of LVEF by the same method on a randomly selected subset of 10 patients (using the random number generator in Microsoft Excel; Microsoft Corporation, Redmond, WA) to determine interobserver variability and reproducibility. The second reader was blinded to the initial analysis, the two measurements were separated by ≥30 days, and the reader was free to independently choose images for analysis within the designated study date.
The highest quality apical four-chamber and parasternal short-axis view images were identified to perform systolic strain measurements using Velocity Vector Imaging version 2.0 (Siemens Medical Solutions USA, Inc). This software provides vendor-independent deformation analysis using a speckle-tracking algorithm to provide angle-independent two-dimensional velocity, strain, and strain rate. The average frame rates for the apical four-chamber and parasternal short-axis views were 44 and 45 Hz, respectively. The primary investigator performed all strain measurements. The endocardial border was manually traced and automatically tracked by the Velocity Vector Imaging software, producing graphs of strain and strain rate of six segments of myocardium in the parasternal short-axis view and four segments of myocardium in the apical four-chamber view over time ( Figure 1 ). The apical segments in the four-chamber view were excluded from longitudinal strain calculations because of the poor tracking capability of these segments. Therefore, global longitudinal strain (GLS), global circumferential strain (GCS), global longitudinal strain rate (GLSr), and global circumferential strain rate (GCSr) measurements were the average of basal and midsegment values, while regional measurements were the average of basal and midsegment values of the lateral or septal wall.
All continuous data are presented as mean ± SD. Parametric testing was used to compare data with normal distributions, such as age, body surface area, and echocardiographic indices. All unpaired comparisons were performed with Student t tests, while paired t tests were used to compare echocardiographic indices before surgery, immediately after surgery, and at 4 to 15 weeks postoperatively.
A similar subanalysis of echocardiographic indices was performed on only patients who underwent mitral valve repair to minimize the influence of type of surgical intervention. Fisher exact tests were used to compare postoperative LV dysfunction when stratified by degree of preoperative MR and to compare postoperative LV dysfunction when stratified by surgical era. The Fisher exact test was also used to compare the percentage of mitral valve repair in those patients with and without postoperative LV dysfunction.
Receiver operating characteristic (ROC) curves were generated to evaluate preoperative echocardiographic indices as potential echocardiographic measures associated with postoperative dysfunction in the entire cohort, defined as an LVEF < 50%. Subsequent ROC analysis was performed on a randomly selected derivation group (using the random number generator in Microsoft Excel). The threshold values obtained were evaluated in a test group, composed of the remaining patients, to evaluate the sensitivity and specificity values for each variable.
Intraclass correlation analysis was used to compare LVEF measurements between two investigators. P values < .05 were considered statistically significant. All statistical calculations were performed using SAS Enterprise Guide version 4.2 (SAS Institute Inc, Cary, NC), Analyse-it Standard version 3.20.2 (Analyse-it, Leeds, United Kingdom), Microsoft Excel 2007, and IBM SPSS Statistics version 22 (IBM Corporation, Armonk, NY).
The study protocol was approved by the Stanford University Institutional Review Board (Protocol No. 20116).
Between January 2000 to March 2013, 70 patients in the Lucile Packard Children’s Hospital Heart Center database met our inclusion criteria of more than mild MR preoperatively with no coincident left-sided obstructive lesions or septal defects. Of the 70 patients, 24 patients were excluded for more than mild postoperative MR and/or inadequate image quality. The remaining 46 patients constituted the study cohort. The cohort had an average age of 8.2 ± 6.4 years and an average body surface area of 0.9 ± 0.5 m 2 preoperatively, and there was a 22:24 male-to-female ratio. Thirty-nine of the patients underwent valve repair, and the remaining seven underwent valve replacement ( Table 1 ). The mechanism of MR was noted to be cleft mitral valve ( n = 21), failed mitral valve repair or iatrogenic disruption of the valve ( n = 8), congenital leaflet abnormality ( n = 3), central regurgitation with annular dilation ( n = 3), mitral valve prolapse ( n = 2), and endocarditis, rheumatic fever, or lupus with leaflet disruption ( n = 9). Seven patients had acute MR (<6 months), while the remaining 39 patients had chronic MR (≥6 months). Average aortic cross-clamp time was 83.5 ± 54.1 min, with 3.0 ± 2.5 cardioplegia doses per case.
|Age, median (range)||8.2 y (4 mo to 21 y)|
|Body surface area (m 2 ), mean ± SD||0.9 ± 0.5|
|Type of mitral valve repair|
|Severity of preoperative MR|
|Mild to moderate||8|
|Moderate to severe||8|
|Cleft mitral valve||21|
|Rheumatic heart disease||4|
|AV canal status after repair||4|
|Mitral valve prolapse||2|
|Systemic lupus erythematosus||1|
|No additional diagnosis||6|
|Duration of MR|
|Acute (<6 mo)||7|
|Chronic (≥6 mo)||39|
The average time between surgery and initial postoperative TTE was 4.0 ± 2.4 days. The average time between preoperative and initial postoperative TTE was 18.4 days. Echocardiographic functional parameters for the whole cohort, both before and initially after mitral valve surgery, are presented in Table 2 .
|LVEF (%) ( n = 46)||68.9 ± 7.0||56.8 ± 10.1||<.0001|
|Indexed LVEDV (mL/m 2 ) ( n = 46)||93.1 ± 41.7||56.2 ± 35.7||<.0001|
|LVEDD Z score ( n = 40)||0.95 ± 2.19||−0.83 ± 1.85||<.0001|
|VCFc (circ/sec) ( n = 38)||1.12 ± 0.38||1.15 ± 0.57||.02|
|WSc (dynes/cm 2 ) ( n = 33)||139.0 ± 52.7||129.4 ± 43.4||.22|
|WSm (dynes/cm 2 ) ( n = 33)||54.5 ± 29.8||61.3 ± 26.1||.38|
|GLS (%) ( n = 46)||−16.7 ± 6.5||−11.5 ± 5.8||<.0001|
|GLSr (sec −1 ) ( n = 46)||−1.11 ± 0.39||−0.93 ± 0.43||.01|
|Regional LS, lateral (%) ( n = 46)||−18.6 ± 8.1||−12.5 ± 7.4||<.0001|
|Regional LSr, lateral (sec −1 ) ( n = 46)||−1.22 ± 0.51||−1.02 ± 0.58||.03|
|Regional LS, septal (%) ( n = 46)||−15.0 ± 6.5||−10.6 ± 6.0||.0002|
|Regional LSr, septal (sec −1 ) ( n = 46)||−1.02 ± 0.35||−0.84 ± 0.46||.03|
|GCS (%) ( n = 43)||−16.0 ± 5.1||−12.9 ± 6.7||.01|
|GCSr (sec −1 ) ( n = 43)||−1.24 ± 0.44||−1.14 ± 0.52||.42|
|Regional CS, lateral (%) ( n = 43)||−15.1 ± 5.7||−13.4 ± 6.9||.17|
|Regional CSr, lateral (sec −1 ) ( n = 43)||−1.19 ± 0.46||−1.19 ± 0.56||.72|
|Regional CS, septal (%) ( n = 43)||−17.7 ± 7.2||−12.0 ± 7.6||.002|
|Regional CSr, septal (sec −1 ) ( n = 43)||−1.34 ± 0.59||−1.03 ± 0.54||.02|
A separate analysis was performed for a subgroup of 19 patients who underwent follow-up echocardiography 4 to 15 weeks after surgery. The average time between surgery and the later postoperative echocardiographic study was 9.5 ± 4.9 weeks. Echocardiographic functional parameters for this group are presented in Table 3 .
|LVEF (%)||68.0 ± 7.2||59.9 ± 19.3||.088|
|Indexed LVEDV (mL/m 2 )||89.9 ± 33.0||70.6 ± 54.1||.015|
|GLS (%)||−14.8 ± 6.6||−12.2 ± 6.1||.018|
|GLSr (sec −1 )||−0.97 ± 0.40||−0.92 ± 0.47||.242|
|GCS (%)||−14.5 ± 5.0||−12.6 ± 6.4||.248|
|GCSr (sec −1 )||−1.16 ± 0.48||−1.03 ± 0.52||.418|
Mean conventional indices of LV systolic function for the entire cohort of patients with MR were predominantly in the normal range preoperatively, with a mean LVEF of 68.9 ± 7.0%. All 46 patients had normal preoperative LVEFs (>50%). The intraclass correlation coefficient to assess for reproducibility of LVEF as measured by the 5/6 area × length method was excellent at 0.87. VCFc was 1.12 ± 0.38 circ/sec, and WSc and meridional wall stress were 139.0 ± 52.7 and 54.5 ± 29.8 dynes/cm 2 , respectively, under these loading conditions.
There was a significant decline in LVEF from the preoperative to the initial postoperative study (68.9 ± 7.0% vs 56.8 ± 10.1%, P < .0001). The postsurgical alteration in loading conditions resulted in a change in WSc from 139.0 ± 52.7 dynes/cm 2 preoperatively to 129.4 ± 43.4 dynes/cm 2 postoperatively; however, the change was not statistically significant ( P = .22). There was a significant decrease in indexed LVEDV (93.1 ± 41.7 vs 56.2 ± 35.7 mL/m 2 , P < .0001) and LVEDD Z score (0.95 ± 2.19 vs −0.83 ± 1.85, P < .0001).
The average GLS and GCS in the cohort were normal before surgery. However, there was a significant magnitude decline from the preoperative to the initial postoperative study for both GLS (−16.7 ± 6.5% vs −11.5 ± 5.8%, P < .0001) and GCS (−16.0 ± 5.1% vs −12.9 ± 6.7%, P = .009). GLSr and GCSr were also normal before surgical intervention. The magnitude of GLSr significantly decreased after surgery (−1.11 ± 0.39 vs −0.93 ± 0.43 sec −1 , P = .01). A magnitude decrease in GCSr was also noted (−1.24 ± 0.44 vs −1.14 ± 0.52 sec –1 , P = .42), but this change was not significantly different.
To ensure that type of surgery had no influence on the results, a repeat subanalysis was performed with only those patients who underwent mitral valve repair ( n = 39). Similar findings were noted, with a statistically significant decrease in postoperative LVEF (69.7 ± 5.9% vs 57.2 ± 9.8%, P = .001), indexed LVEDV (93.5 ± 41.9 vs 53.0 ± 27.7 mL/m 2 , P < .0001), LVEDD Z score (0.68 ± 2.4 vs −1.2 ± 1.8, P < .0001), magnitude of GLS (−16.9 ± 6.0% vs −11.8 ± 5.7%, P < .0001), GLSr (−1.10 ± 0.34 vs −0.96 ± 0.42 sec −1 , P = .001), and GCS (−16.1 ± 4.9% vs −12.9 ± 6.7%, P = .019).
Similar findings were discovered when comparing the pre- and postoperative studies at 4 to 15 weeks ( n = 19), with significant declines in indexed LVEDV (89.9 ± 33.0 vs 70.6 ± 54.1 mL/m 2 , P = .015) and magnitude of GLS (−14.8 ± 6.6 vs −12.2 ± 6.1%, P = .018). Although LVEF, GLSr, GCS, and GCSr all decreased in magnitude in the further postoperative study, the differences were not statistically significant ( Table 3 ).
The cohort of patients was then stratified into two groups, those with immediate postoperative dysfunction ( n = 13) and those without postoperative dysfunction ( n = 33) ( Table 4 ). The group with postoperative dysfunction was older (11.5 ± 7.1 vs 7.3 ± 5.7 years, P = .04). Preoperative LVEF was significantly lower in the group with postoperative dysfunction, as expected, but still within normal limits (65.1 ± 7.7% vs 70.4 ± 6.5%, P = .02). GCS was abnormal preoperatively in the group with immediate postoperative dysfunction (–13.2 ± 5.6% vs −17.1 ± 4.6%, P = .02), as well as GCSr (−0.94 ± 0.40 vs −1.36 ± 0.42 sec −1 , P = .004). Regional circumferential strain rate of the septum, regional circumferential strain of the lateral wall, and regional circumferential strain rate of the lateral wall were also significantly lower in magnitude in the cohort with postoperative dysfunction ( Table 4 ). The magnitude of GLSr was lower in the postoperative dysfunction group, though not statistically significantly (−0.91 ± 0.34 vs −1.14 ± 0.37 sec –1 , P = .058); however, the magnitude of regional longitudinal strain rate of the septum was significantly lower in the postoperative dysfunction group (−0.82 ± 0.34 vs −1.07 ± 0.30 sec −1 , P = .03).
|Measurement||Dysfunction ( n = 13)||No dysfunction ( n = 33)||P|
|Age (y)||11.5 ± 7.1||7.3 ± 5.7||.04|
|LVEF (%)||65.1 ± 7.7||70.4 ± 6.5||.02|
|Indexed LVEDV (mL/m 2 )||114.9 ± 40.4||87.8 ± 40.4||.046|
|LVEDD Z score||2.32 ± 1.28||0.49 ± 2.26||.009|
|VCFc (circ/sec)||1.08 ± 0.49||1.19 ± 0.35||.60|
|WSc (dynes/cm 2 )||161.9 ± 59.1||127.8 ± 48.7||.05|
|WSm (dynes/cm 2 )||67.5 ± 46.6||48.6 ± 38.4||.06|
|GCS (%)||−13.2 ± 5.6||−17.1 ± 4.6||.02|
|GCSr (sec −1 )||−0.94 ± 0.40||−1.36 ± 0.42||.004|
|Regional CS, septal (%)||−15.5 ± 7.8||−18.3 ± 7.2||.25|
|Regional CSr, septal (sec −1 )||−1.02 ± 0.48||−1.46 ± 0.62||.03|
|Regional CS, lateral (%)||−12.1 ± 5.6||−16.4 ± 5.2||.02|
|Regional CSr, lateral (sec −1 )||−0.90 ± 0.40||−1.31 ± 0.43||.007|
|GLS (%)||−15.3 ± 6.3||−16.7 ± 6.5||.49|
|GLSr (sec −1 )||−0.91 ± 0.34||−1.14 ± 0.37||.06|
|Regional LS, septal (%)||−12.8 ± 5.5||−15.8 ± 6.8||.16|
|Regional LSr, septal (sec −1 )||−0.82 ± 0.34||−1.07 ± 0.30||.03|
|Regional LS, lateral (%)||−17.8 ± 8.2||−18.0 ± 7.8||.92|
|Regional LSr, lateral (sec −1 )||−0.99 ± 0.44||−1.26 ± 0.50||.09|
|Aortic cross-clamp time (min)||122 ± 68||69 ± 38||.032|
|Number of cardioplegia doses||3.5 ± 2.8||2.8 ± 2.3||.347|
|Aortic cross-clamp time per dose of cardioplegia (min/dose)||49 ± 57||28 ± 13||.05|
|Greater than moderate MR||12 (92%)||16 (48%)||.0073|