Background
Mitral valve (MV) repair is the procedure of choice for patients with degenerative MV disease (DMVD) with severe mitral regurgitation. The aim of this study was to identify specific quantitative MV parameters from preoperative three-dimensional (3D) transesophageal echocardiography that are associated with the length of the mitral annuloplasty band implanted and the performance of leaflet resection in patients with DMVD undergoing MV repair.
Methods
Ninety-four patients (mean age, 60 ± 11 years; 68% men) referred for MV surgery with adequate-quality preoperative 3D transesophageal echocardiographic studies were retrospectively identified. Parametric maps of the MV were generated using semiautomated MV modeling software. Annular and valvular parameters were measured and indexed to body surface area. The implanted annuloplasty band size and leaflet resection were determined on the basis of surgical reports.
Results
Three-dimensional annular circumference correlated best ( r = 0.74) with the implanted annuloplasty band length and remained an independent predictor on multivariate linear regression analysis. A third of our cohort ( n = 33) had posterior leaflet resection. On receiver operating characteristic curve analysis, P2 segment length ≥ 20 mm (area under the curve, 0.86; sensitivity, 88%; specificity, 74%) and P2 leaflet area ≥ 3.4 cm 2 (area under the curve, 0.84; sensitivity, 85%; specificity, 74%) best discriminated the need for leaflet resection.
Conclusions
In DMVD, quantitative 3D annular circumference obtained from semiautomatically generated parametric maps of the MV from 3D transesophageal echocardiographic data was associated with the surgically implanted annuloplasty band length, while P2 leaflet length ≥ 20 mm and area ≥ 3.4 cm 2 were associated with the performance of leaflet resection. These parameters should be further investigated for preoperative planning in patients with DMVD undergoing MV repair.
Degenerative mitral valve (MV) disease (DMVD) affects approximately 2% of the population and is the leading cause of mitral regurgitation (MR) in developed countries. In patients requiring surgery for severe MR due to DMVD, American and European societal guidelines recommend valve repair over replacement, provided that the valve is suitable and that the institution has surgeons with appropriate expertise. Valve repair as opposed to replacement is associated with improved event-free survival. Two important aspects of MV repair are the choice of mitral annuloplasty ring or band size and the decision regarding leaflet modification. These decisions are most often made intraoperatively on the basis of surgical experience and judgement. This reduces reproducibility. Several tools such as ring sizers have been used to help make this process more objective, but they too have limitations.
Echocardiography is the modality of choice for the assessment of MV disease. Specifically, three-dimensional (3D) transesophageal echocardiography (TEE) allows superior visualization of the MV anatomy and morphology, allowing better assessment of the lesion, its complexity, and its suitability for repair compared with two-dimensional (2D) TEE. In addition, 3D MV modeling generates a plethora of quantitative measurements of the MV obtained in its physiologic state. This offers an opportunity for preoperative surgical planning and individualizing the surgical approach to the patient. However, before using 3D MV modeling for surgical planning, it is important to first demonstrate associations between measurements obtained by preoperative 3D modeling and components of the surgical technique. The aim of our study was to identify specific quantitative parameters on a preoperative 3D TEE using a semiautomated MV modeling technique that would correlate with the annuloplasty band length used and the performance of leaflet resection in patients with DMVD who underwent MV repair. We hypothesized that mitral annular and leaflet parameters commonly measured and used intraoperatively along with novel 3D parameters can be obtained by 3D TEE–based semiautomated MV modeling preoperatively and will correlate with the annuloplasty band length used and the performance of leaflet resection during the surgery.
Methods
Patients
We retrospectively identified all adult patients referred to an expert surgeon (T.D.), who underwent MV repair for severe MR secondary to DMVD between 2010 and 2013 at Toronto General Hospital (Toronto, Ontario, Canada). Patients who had diagnostic-quality pre- or intraoperative 3D TEE, received annuloplasty bands, and underwent predischarge echocardiography were included. The study protocol was approved by the institutional research ethics board.
Echocardiography
All patients underwent preoperative transthoracic echocardiography. Left ventricular (LV) volumes and ejection fraction were measured using the biplane Simpson method. The severity of MR was graded per American Society of Echocardiography guidelines using a multiparametric approach. All patients then underwent clinical 2D TEE using an iE33 system (Philips Medical Systems, Andover, MA) equipped with an X7-2t transesophageal probe. Three-dimensional assessment of the MV was performed using full-volume (median volume rate, 20 volumes/sec; interquartile range, 16–26 volumes/sec) or real-time (median volume rate, 9 volumes/sec; interquartile range, 7–13 volumes/sec) 3D acquisitions of the MV from either the midesophageal four- or three-chamber view. Four-beat gated acquisitions were used for the full-volume data sets. Withholding of respiration was performed whenever possible. All patients underwent predischarge transthoracic echocardiography from which the MV peak and mean gradients and residual MR were assessed by two level 3–trained echocardiographers (F.P., P.T.) using the integrative approach, as recommended by the American Society of Echocardiography. The result of the surgical repair was considered optimal on the basis of a ≤5mm Hg mean transvalvular gradient, mild or less MR, and absence of systolic anterior motion (SAM) or LV outflow tract obstruction on predischarge echocardiography.
Three-Dimensional Quantitative Measurements
The 3D transesophageal echocardiographic data sets were first assessed for gating artifacts by examining the studies in a plane perpendicular to the plane of acquisition. Studies with gating artifacts were excluded. The studies were then analyzed offline by a single operator (A.C.) blinded to clinical, echocardiographic, and surgical findings using semiautomated valve software (eSie Valves; Siemens Medical Solutions USA, Inc, Mountain View, CA) which has been previously described in detail. In brief, first 3D Digital Imaging and Communications in Medicine data were loaded. A midsystolic frame was chosen for analysis because it was where leaflet billowing and/or prolapse was best visualized, and on the basis of the annular dynamics, it was felt to represent the average measure of annular circumference (smallest circumference at early systole and largest at end-systole). The valve was segmented automatically using a machine learning algorithm. The position of the valve and its orientation and dimension in the image are first detected. This forms a region of interest, within which key landmarks (e.g., trigones and commissures) as well as the annulus and leaflet free edges are detected. Then an average surface model of the anterior and posterior leaflets is fitted to the landmarks. These contours are then deformed to match the atrial side of the leaflets in the image ( Figure 1 A). To facilitate automated computation of complex measurements, the surface model returned by eSie Valves is represented uniformly. This is established from a landmark-based resampling procedure. Each vertex of the valve surface model is uniquely defined by two coordinates, the u-coordinate, tangential to the valve circumference, from anterior to posterior, and the v-coordinate, perpendicular to the valve circumference, from annulus to free edge ( Figure 1 B). Each stage of the automated valve modeling is performed by robust detectors trained on a large database of cases covering a wide range of normal and pathologic patient data. When computing the annular circumference, the application goes through the vertices of the anterior leaflet defined by a zero v-coordinate, creating the anterior perimeter, followed by the same process for the posterior leaflet. The total circumference is then the sum of the anterior and posterior perimeters. Leaflet segments are defined using geometric features to cope with the lack of clear image features. More precisely, the A1 to A3 and P1 to P3 segments were defined by dividing the leaflets at the u-coordinates 0 to 1/3, 1/3 to 2/3 and 2/3 to 1.
Once the valve is modeled, to confirm the predetermined landmarks and edit the segmented valve, different work-flow options were available. First, commonly used valve orientations (long axis through A2 to P2, commissural view, and en face view) are displayed for editing ( Figure 2 A). To view the model in more detail, parallel sagittal and coronal or rotational cut planes (which go through a 360° rotation of the valve) are used ( Figures 2B and 2C ). Using those planes, data sets in which shift and stitch artifacts interfered with the modeling were excluded. Quantitative parameters automatically generated from the geometric model included (1) 3D annular measurements: total area, anterior, posterior, and total circumference; (2) anteroposterior and anterolateral-posteromedial (ALPM) diameter; (3) 3D leaflet measurements: anterior and posterior leaflet areas, A1 to A3 leaflet segment areas and lengths, P1 to P3 leaflet segment areas and lengths; (4) intertrigonal distance (ITD); and (5) inter-commissural distance (ICD) ( Figure 3 ). The annular and leaflet measurements were true 3D surface measurements. For example, the area measurements of the leaflet were true 3D surface area, and the length of the leaflets included 3D length of the prolapsing segment.
Surgical Technique
Surgical repair was performed by a single experienced surgeon (T.D.) blinded to all quantitative 3D transesophageal echocardiographic MV data. Grading of DMVD was based on the intraoperative leaflet and annular changes seen during surgical inspection. Mild myxomatous degeneration was defined as leaflets that were thin, were fairly normal in size, were transparent (except for the prolapsing segment), and had chordae tendineae that were thin and attenuated. This included Carpentier fibroelastic deficiency and cases with minimal myxomatous changes. Moderate degeneration was defined as opaque leaflets due to myxoid infiltration of the spongiosa, increased leaflet size but still elastic and not excessively thick (<3 mm thick), and myxoid infiltration of the chordae tendineae. Severe degeneration was considered to exist when leaflets were voluminous, aneurysmal, and thickened (≥3 mm), the annulus was massively dilated (i.e., ≥40 mm), and often posterior displacement of the mitral annulus ≥5 mm, with thick and obviously myxomatous chordae tendineae.
MV repair was performed with or without neochord placement or leaflet resection, followed by annuloplasty band placement using a Medtronic Simplici-T band (Medtronic, Inc, Minneapolis, MN). Prolapse of the anterior leaflet and of the commissural areas was corrected with chordal replacement. Posterior leaflet prolapse was corrected mostly with neochords, when the leaflet was normal in size or when its height was not significantly increased (<20 mm); however, partial resection of the base and occasional triangular resection of the free margin were done in patients with excessively large posterior leaflets (height > 20 mm). The Simplici-T annuloplasty band has been used exclusively by our surgeon since October 2005. No pre- or intraoperative measurements are used to determine the length of the band used. First annuloplasty sutures are placed through the posterior mitral annulus from the lateral to the medial fibrous trigones. Then the annuloplasty band is secured to these sutures. Annular reduction is performed by passing the sutures closer together in the band in areas of commissures, false commissures, or partially resected segments, on the basis of subjective observation of disproportion between the sizes of the leaflets and the mitral annulus. The excess length of the 100-mm band is then cut.
Interobserver and Intraobserver Variability
For intraobserver variability, 10 randomly selected studies were reanalyzed by the same observer (A.C.) 3 to 5 months after the initial analysis, blinded to the original measurement or the precise frame used previously. For interobserver variability, the measurements were repeated by a second observer (F.P.) blinded to the original measurements and to the precise frame used. Variability was assessed for relevant annular and leaflet parameters.
Statistical Analysis
Categorical variables are expressed as frequencies and continuous variables as mean ± SD or median and interquartile range, depending on normality. Univariate and multivariate linear regression analysis was performed to identify 3D transesophageal echocardiographic parameters associated with the annuloplasty band size used in patients with optimal surgical repair (defined above) using the bootstrap resampling method (10,000 samples). The bias-corrected and accelerated method was used to generate CIs. In the multivariate model, the best parameters on univariate analysis and other parameters that are clinically used intraoperatively to determine annuloplasty band size length were included. A partial F test was used to compare a model with single parameter with a more comprehensive model using all the parameters commonly used intraoperatively to determine annuloplasty band size. Comparisons between patients with and those without leaflet resection were performed using Pearson’s χ 2 tests for categorical variables and Student’s unpaired t test or the Wilcoxon signed rank sum test for continuous data. Receiver operating characteristic curves were created to assess the discriminatory value of P2 length and P2 area for the need for leaflet resection. Inter- and intraobserver variability was assessed using the intraclass correlation coefficient and the coefficient of variation. A level of significance of .05 was used. All statistical analyses were conducted using SPSS version 20 (SPSS, Inc, Chicago, IL).
Results
Patient Population and Baseline Characteristics
A total of 285 patients had MV repair during the study period. Among these patients, preoperative 3D TEE was performed in 189 patients. The proportion of patients undergoing 3D TEE increased over time, extending from 42% in 2010 to 91% in 2013, reflecting availability of equipment and experience of the operators. Among these cases, only 106 patients had available Digital Imaging and Communications in Medicine data for analysis. However, 12 were excluded because of low volume rates of acquisition (≤5 volumes/sec). A total of 94 patients with a mean age of 60 ± 11 years, 68% of whom were men, were included. Among the patients, 61% had New York Heart Association functional class ≥ II symptoms at the time of surgery, with normal LV size (end-systolic diameter < 40 mm) in 86% and normal function (LV ejection fraction ≥ 60%) in 62%. Baseline characteristics of the 94 patients are summarized in Table 1 . The 12 excluded patients had similar clinical characteristics (mean age, 56 ± 15 years; New York Heart Association class ≥ II in 42%; no LV dilatation; and normal LV function in 83%).
| Variable | Value |
|---|---|
| Age (y) | 60 ± 11 |
| Men | 64 (68%) |
| BSA (m 2 ) | 1.91 ± 0.23 |
| NYHA class | |
| I | 37 (39%) |
| II | 46 (49%) |
| III | 11 (12%) |
| IV | — |
| Hypertension | 32 (35%) |
| Diabetes | 5 (5%) |
| Atrial fibrillation ∗ | 15 (16%) |
| 2D echocardiography | |
| LV end-diastolic diameter (cm) | 5.3 ± 0.5 |
| LV end-systolic diameter (cm) | 3.3 ± 0.5 |
| LV ejection fraction (%) | 61 ± 7 |
| RVSP (mm Hg) | 36 ± 13 |
∗ None of the patients were in atrial fibrillation at the time of echocardiography.
Preoperative Quantitative 3D Echocardiography
Automated valve modeling was performed in <1 min, but additional time was necessary for manual editing of annular landmarks and leaflet contours. The latter required on average 8 min, given that 50% of our patients had bileaflet disease, and 65% had moderate to severe myxomatous valvular changes. Quantitative annular and leaflet measurements are summarized in Table 2 .
| Quantitative 3D parameter | All ( n = 94) | No resection ( n = 61) | Resection ( n = 33) | P † |
|---|---|---|---|---|
| Annular dimensions (indexed to BSA ∗ ) | ||||
| Annular area (cm 2 ) | 14.2 ± 3.7 (7.5 ± 1.9) | 13.6 ± 3.3 (7.5 ± 1.9) | 15.5 ± 4.0 (7.7 ± 1.8) | .03 |
| Anterior circumference (mm) | 63 ± 8 (33.6 ± 5.1) | 62 ± 8 (33.8 ± 5.5) | 66 ± 8 (33.1 ± 1.2) | .01 |
| Posterior circumference (mm) | 75 ± 11 (39.6 ± 6.7) | 73 ± 10 (40.2 ± 6.9) | 77 ± 12 (38.6 ± 6.2) | .09 |
| Total annular circumference (mm) | 138 ± 17 (73.2 ± 10.7) | 135 ± 17 (74.0 ± 11.5) | 144 ± 17 (71.6 ± 9.1) | .02 |
| AP diameter (mm) | 37 ± 5 (19.5 ± 3.1) | 35 ± 5 (19.6 ± 3.3) | 39 ± 6 (19.4 ± 2.9) | .003 |
| ALPM diameter (mm) | 44 ± 6 (23.4 ± 3.7) | 43 ± 6 (23.6 ± 4.0) | 46 ± 6 (22.9 ± 3.3) | .04 |
| ITD (mm) | 27 ± 3 (14.6 ± 2.2) | 27 ± 4 (14.8 ± 2.4) | 29 ± 3 (14.3 ± 1.7) | .02 |
| ICD (mm) | 27 ± 5 (14.5 ± 2.9) | 27 ± 5 (14.6 ± 3.0) | 29 ± 5 (14.3 ± 2.5) | .09 |
| Leaflet measurements (indexed to BSA ∗ ) | ||||
| A1 length (mm) | 16 ± 3 (8.7 ± 1.8) | 16 ± 3 (8.9 ± 1.9) | 17 ± 3 (8.2 ± 1.6) | .58 |
| A1 area (cm 2 ) | 2.2 ± 0.7 (1.2 ± 0.4) | 2.1 ± 0.7 (1.2 ± 0.5) | 2.3 ± 0.7 (1.1 ± 0.3) | .29 |
| A2 length (mm) | 22 ± 4 (11.8 ± 2.7) | 22 ± 4 (12.1 ± 3.0) | 22 ± 4 (11.2 ± 2.2) | .36 |
| A2 area (cm 2 ) | 3.3 ± 1.0 (1.8 ± 0.6) | 3.2 ± 0.8 (1.8 ± 0.5) | 3.6 ± 1.2 (1.8 ± 0.6) | .12 |
| A3 length (mm) | 17 ± 3 (8.9 ± 1.9) | 16 ± 3 (9.0 ± 1.9) | 18 ± 4 (8.9 ± 1.8) | .13 |
| A3 area (cm 2 ) | 2.0 ± 0.7 (1.1 ± 0.7) | 1.9 ± 0.6 (1.0 ± 0.4) | 2.2 ± 0.9 (1.1 ± 0.4) | .04 |
| Anterior leaflet area (cm 2 ) | 7.5 ± 2.2 (4.0 ± 1.2) | 7.2 ± 1.9 (4.0 ± 1.2) | 8.2 ± 2.7 (4.1 ± 1.2) | .11 |
| P1 length (mm) | 16 ± 5 (8.4 ± 2.5) | 15 ± 5 (8.2 ± 2.6) | 18 ± 5 (8.7 ± 2.3) | .003 |
| P1 area (cm 2 ) | 2.8 ± 1.2 (1.5 ± 0.6) | 2.7 ± 1.0 (1.5 ± 0.6) | 3.0 ± 1.4 (1.5 ± 0.7) | .32 |
| P2 length (mm) | 19 ± 6 (8.2 ± 2.6) | 17 ± 5 (9.5 ± 2.9) | 24 ± 5 (12.0 ± 2.2) | <.001 |
| P2 area (cm 2 ) | 3.4 ± 1.4 (1.8 ± 0.7) | 2.8 ± 1.1 (1.6 ± 0.7) | 4.5 ± 1.4 (2.2 ± 0.6) | <.001 |
| P3 length (mm) | 16 ± 4 (8.3 ± 2.4) | 15 ± 4 (8.1 ± 2.5) | 17 ± 4 (8.6 ± 2.3) | .003 |
| P3 area (cm 2 ) | 2.7 ± 1.1 (1.5 ± 0.6) | 2.5 ± 1.1 (1.4 ± 0.6) | 3.1 ± 1.1 (1.5 ± 0.6) | .02 |
| Posterior leaflet area (cm 2 ) | 8.9 ± 3.3 (4.8 ± 1.8) | 8.0 ± 2.8 (4.5 ± 1.8) | 10.5 ± 3.5 (5.2 ± 1.6) | <.001 |
∗ Indexed measurements (area in square centimeters per square meter, length and circumference in millimeters per square meter).
Surgical Findings
Single-leaflet disease was seen in 50%, with the posterior leaflet more commonly affected ( Table 3 ). The degree of myxomatous disease was moderate or severe in 65% of the patients. Eighty-seven patients (93%) underwent neochord implantation (13 ± 7 chords/patient), and all patients underwent annuloplasty band (mean size, 66 ± 7 mm) placement. Posterior leaflet resection was performed in 33 patients (35%). Compared with the group without resection, these patients had more severe myxomatous changes (moderate or greater changes in 82%) and had longer annuloplasty bands implanted ( Table 3 ).
| Variable | All ( n = 94) | No resection ( n = 61) | Resection ( n = 33) | P ∗ |
|---|---|---|---|---|
| Leaflet involvement | .06 | |||
| Isolated anterior | 9 (10%) | 9 (15%) | 0 | |
| Isolated posterior | 38 (40%) | 23 (38%) | 15 (45%) | |
| Bileaflet | 47 (50%) | 29 (48%) | 18 (55%) | |
| Myxomatous change | .04 | |||
| Mild | 33 (35%) | 27 (44%) | 6 (18%) | |
| Moderate | 44 (47%) | 25 (41%) | 19 (58%) | |
| Severe | 17 (18%) | 9 (15%) | 8 (13%) | |
| Total number of chords placed | 12.6 ± 7.1 | 11.9 ± 6.2 | 14.1 ± 8.7 | .25 |
| Simplici-T band length (mm) | 66 ± 7 | 64 ± 6 | 69 ± 8 | .009 |
Predischarge Echocardiography
On the basis of predischarge 2D transthoracic echocardiography, optimal early postoperative results were seen in 86 patients (91%). The remaining eight had mild to moderate residual MR ( n = 3) or transmitral gradients > 5 mm Hg ( n = 5; mean gradient, 5.7 ± 0.3 mm Hg; mean heart rate, 80 ± 4 beats/min) ( Table 4 ). When patients with mean gradients ≤ 5 mm Hg ( n = 89) were compared with those with gradients >5 mm Hg ( n = 5), 31 of 89 patients with gradients ≤ 5 mm Hg underwent MV leaflet resection, while two of five with elevated gradients underwent leaflet resection. The mean length of the annuloplasty band in patients with elevated gradients was 62 ± 8 mm, while in the other group it was 66 ± 7 mm.
| Variable | All ( n = 94) | No resection ( n = 61) | Resection ( n = 33) | P ∗ |
|---|---|---|---|---|
| Mean transmitral gradient (mm Hg) | 3.3 ± 1.0 | 3.3 ± 1.0 | 3.2 ± 1.1 | .36 |
| Peak transmitral gradient (mm Hg) | 8.3 ± 2.9 | 8.3 ± 3.2 | 8.2 ± 2.4 | .81 |
| MR severity | .99 | |||
| None or trivial | 75 (80%) | 49 (80%) | 26 (79%) | |
| Mild | 16 (17%) | 10 (16%) | 6 (18%) | |
| Moderate | 3 (3%) | 2 (3%) | 1 (3%) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
