The aim of this study was to validate a new real-time three-dimensional echocardiography (RT3DE) score for evaluating patients with mitral stenosis (MS).
A two-staged study was conducted. In the first stage, the feasibility of a new RT3DE score was assessed in 17 patients with MS. The second stage was planned to validate the RT3DE score in 74 consecutive patients undergoing percutaneous mitral valvuloplasty. The new RT3DE score was constructed by dividing each mitral valve (MV) leaflet into 3 scallops and was composed of 31 points (indicating increasing abnormality), including 6 points for thickness, 6 for mobility, 10 for calcification, and 9 for subvalvular apparatus involvement. The total RT3DE score was calculated and defined as mild (<8), moderate (8-13), or severe (≥14). MV morphology was assessed using Wilkins’s score and compared with the new RT3DE score.
In the first stage, the RT3DE score was feasible and easily applied to all patients, with good interobserver and intraobserver agreement. In the second stage, RT3DE improved MV morphologic assessment, particularly for the detection of calcification and commissural splitting. Both scores were correlated for assessment of thickness and calcification ( r = 0.63, P < .0001, and r = 0.44, P < .0001, respectively). Predictors of optimal percutaneous mitral valvuloplasty success by Wilkins’s score were leaflet calcification and subvalvular apparatus involvement, and those by RT3DE score were leaflet mobility and subvalvular apparatus involvement. The incidence and severity of mitral regurgitation were associated with high-calcification RT3DE score.
The new RT3DE score is feasible and highly reproducible for the assessment of MV morphology in patients with MS. It can provide incremental prognostic information in addition to Wilkins’s score.
Rheumatic mitral valve (MV) disease remains a major health problem, particularly in developing countries. Two-dimensional echocardiography (2DE) plays a crucial role in selecting the proper therapeutic strategy in patients with mitral stenosis (MS). Apart from estimation of the transmitral pressure gradient, 2DE can calculate MV area (MVA) by planimetry and pressure half-time. Also, 2DE is essential in the selection of patients for percutaneous mitral valvuloplasty (PTMV). Among all available MV scoring systems, Wilkins’s score is the most commonly used. This score evaluates MV thickness, mobility, leaflet calcification, and the degree of subvalvular apparatus thickening. A favorable Wilkins’s score (<8 points) is highly predictive of a good outcome after PTMV. More recently, it has been shown that other factors, such as commissural fusion and calcification, are also strong predictors of outcomes after PTMV.
The introduction of real-time three-dimensional echocardiography (RT3DE) and advances in analysis software have improved MV orientation and evaluation. Several studies have reported improved accuracy and superiority of RT3DE over 2DE for MVA estimation. RT3DE allows a unique “en face” view and morphologic analysis of the entire MV apparatus, including the MV annulus, the subvalvular apparatus, and its anatomic relation to other nearby structures. In the current study, we assessed a new RT3DE score for the assessment of the MV in patients with MS in comparison with Wilkins’s score for the prediction of PTMV outcomes.
The study was designed in two stages. In the first stage, we studied the feasibility and applicability of the proposed RT3DE score in 17 patients with MS (mean age, 30.6 ± 7.8 years; 70% women) and adequate image quality on 2DE. All patients were in sinus rhythm. In the second stage, we validated this score in 74 consecutive patients (mean age, 34.4 ± 5.9 years; 62% women) with tight MS preselected according to Wilkins’s score for PTMV. Sixty-seven patients (74%) were in sinus rhythm, while 24 patients (26%) were in atrial fibrillation. Finally, the RT3DE score was applied to all 91 patients and compared with Wilkins’s score. To assess the interobserver and intraobserver variability, all data were analyzed twice by two independent experienced observers blinded to each other’s results in two different sessions (A.M.A. and O.I.I.S. in the first stage and A.M.A. and W.M.A. in the second stage). The observers were also blinded to PTMV outcomes to avoid bias toward any of both scores.
The first stage was performed at Thoraxcenter, Erasmus Medical Center (Rotterdam, The Netherlands), and the second stage was conducted at Al-Hussein University Hospital, Al-Azhar University (Cairo, Egypt). The study was approved by the institutional review boards at both centers and carried out in accordance with the universal ethics standards established by the Declaration of Helsinki.
Complete 2-dimensional echocardiographic studies were performed 24 hours prior to PTMV using a 3.5-MHz (S3) probe and a commercially available ultrasound system (Sonos 7500; Philips Medical Systems, Best, The Netherlands). The morphology score according to Wilkins’s score was applied for the evaluation of 4 factors: leaflet thickness, mobility, calcification, and the degree of subvalvular thickness. Each factor is given a score of 0 for normal, 1 for mild MV involvement, 2 for moderate, and 3 or 4 for severe, yielding a summed score of 16 points. The higher the score, the greater the morphologic abnormality of the MV. According to Wilkins’s score, <5 points defines mild MV involvement, 5 to 8 moderate involvement, and >8 severe involvement. In addition, the following measurements were obtained: (1) MVA was defined as the narrowest orifice at the time of maximal MV opening and measured by two-dimensional (2D) planimetry in the short-axis view, (2) mean transmitral pressure gradient was calculated using continuous-wave Doppler, (3) mitral regurgitation was graded according to the vena contracta using color Doppler flow, and (4) commissural splitting was assessed for each commissure (anterior and posterior). Commissural splitting was defined as the terminal distance between the two MV leaflets at an end-diastolic still frame from the parasternal short-axis view at the MV level. The splitting was scored as 0 for no splitting, 0.5 for partial splitting, and 1 for complete splitting. All measurements were obtained <48 hours after PTMV. One-year post-PTMV clinical and 2D echocardiographic examination follow-up was obtained for all patients included in the second stage. The endpoints included recurrent symptoms, restenosis, and need for reintervention (either PTMV or MV surgery).
Transesophageal 2DE was performed 24 hours before PTMV for the assessment of MV morphology, to exclude evidence of left atrial and/or left atrial appendage thrombi, and to measure interatrial septal thickness.
RT3DE was performed immediately after 2DE using the same machine, with an X4 matrix transducer capable of providing real-time B-mode and color Doppler images. Full-volume images of 3 cardiac cycles were acquired within 5 to 7 seconds of breath holding in patients with sinus rhythm. In patients with atrial fibrillation, 8 separate live three-dimensional (3D) images were acquired instead of full-volume images because of variability in cardiac cycle duration. The 3D data were transferred to an offline analysis system (first stage: TomTec GmbH, Munich, Germany; second stage: (QLAB version 6, Philips Medical Systems) ( Video 1 ; view video clip online.). Data were stored digitally and subsequently evaluated by two independent experienced observers. Data analysis of 3D echocardiographic imaging was based on the acquired images from the apical and parasternal positions in patients with sinus rhythm. Two additional views (apical long axis and parasternal short axis) were needed in patients with atrial fibrillation. By using the crop function for image formatting, parallel sections through scanned volumes in 3 perpendicular planes were obtained. The narrowest slice to enable visualization of the MV (leaflets and subvalvular apparatus) was selected.
The new RT3DE morphology score ( Table 1 ) was derived to include both MV leaflets and the subvalvular apparatus, guided by Wilkins’s score. Each leaflet was divided into 3 scallops (anterolateral, A1 and P1; middle, A2 and P2; and posteromedial, A3 and P3) and scored separately ( Figure 1 ). The subvalvular apparatus was divided into 3 cut sections of the anterior and posterior chordae at 3 levels: proximal (valve level), middle, and distal (papillary muscle level). Each cut section was scored separately for chordal thickness and separation in between.
|Anterior leaflet||Posterior leaflet|
|Thickness (0-6) (0 = normal, 1 = thickened) ∗||0-1||0-1||0-1||0-1||0-1||0-1|
|Mobility (0-6) (0 = normal, 1 = limited) ∗||0-1||0-1||0-1||0-1||0-1||0-1|
|Calcification (0-10) (0 = no, 1-2 = calcified) †||0-2||0-1||0-2||0-2||0-1||0-2|
|Subvalvular apparatus †|
|Proximal third||Middle third||Distal third|
|Thickness (0-3) (0 = normal, 1 = thickened)||0-1||0-1||0-1|
|Separation (0-6) (0 = normal, 1 = partial, 2 = no)||0, 1, 2||0, 1, 2||0, 1, 2|
Thickness, mobility, and calcification for each MV scallop were assessed qualitatively and scored as follows: normal thickness and mobility were scored as 0 and abnormal thickness or restricted mobility as 1. An absence of calcification was scored as 0, calcification in middle scallop (A2 or P2) was scored as 1, and calcification of the commissural scallops of both leaflets (A1, A3, P1, and P3) was scored as 2.
Scoring of the subvalvular apparatus included chordal thickness and distance of separation between the chordae as follows: normal thickness was scored as 0 and abnormal thickness as 1, normal chordal separation (distance in between > 5 mm) was scored as 0, partial separation (distance in between < 5 mm) as 1, and absence of separation was scored as 2.
The individual RT3DE score points of leaflets and subvalvular apparatus were summed to calculate the total RT3DE score, ranging from 0 to 31 points. A total score of mild MV involvement was defined as <8 points, moderate MV involvement as 8 to 13 points, and severe MV involvement as ≥14 points.
In the second stage, all score components were used to create a logistic regression model with definition of success as the dichotomous dependent variable. This resulted in a multivariate coefficient for each variable with a standard error. None of the score components could be excluded, because all components had significant coefficients. Regarding the grading of each component and the total score points, receiver operating characteristic (ROC) curves were used to identify the cutoff points. The area under the ROC curve was used to describe the accuracy of both the RT3DE and Wilkins’s scores.
Seventy-four patients recruited in the second stage (mean age, 34.4 ± 5.9 years; 70% women) were class I candidate for PTMV according to American Heart Association and American College of Cardiology guidelines. All had Wilkins’s scores ≤ 9. PTMV was performed by two experienced operators who were unaware of the RT3DE assessments. PTMV was performed using the Inoue balloon technique. The optimal success was defined as achievement of a 50% increase in MVA compared with before PTMV or an achieved MVA ≥ 1.5 cm 2 (which is larger) with mitral regurgitation moderate or less. Suboptimal success was defined as an achieved MVA < 1.5 cm 2 with up to moderate regurgitation or severe regurgitation irrespective of the achieved MVA.
Clinical and 2D echocardiographic follow-up were completed for all patients who underwent PTMV. The soft endpoints were recurrence of symptoms and/or increased transmitral pressure gradient without need for intervention. The hard endpoints were restenosis, increased severity of mitral regurgitation, and need for reintervention.
All statistical analyses were performed using SPSS version 12.0.1 (SPSS, Inc, Chicago, IL). Each component of both scores was assessed separately, and then the total scores of 2DE and RT3DE were calculated for every patient. Both scores were investigated for correlations using Pearson’s correlation analysis to demonstrate the basic differences. Paired t tests were used, and P < .05 was set as the criterion for significance in all comparisons. Interobserver and intraobserver agreement were evaluated with the κ index, classified as poor (κ = 0.01-0.20), slight (κ = 0.21-0.40), fair (κ = 0.41-0.60), good (κ = 0.61-0.80), very good (κ = 0.81-0.92), or excellent (κ = 0.93-1.00). Univariate Cox regression analysis identified baseline variables that were significantly associated with outcomes (immediate and 1-year success and significant mitral regurgitation). Significantly associated variables with P values < .10 were then integrated into multivariate analysis using Cox proportional-hazards modeling with stepwise selection. The proportional-hazards assumptions were then validated in the final model for each categorical variable through visual inspection of log-log plots. For continuous variables, the linearity assumption was checked graphically using the residual plots. There were no signs of violation of the assumptions. ROC curves were used to define the best cutoff value that was associated with outcomes.
First-Stage Study Results
Acquisition of real-time 3D echocardiographic data was performed successfully in all 17 patients within a reasonable time (3 minutes for acquisition and 10 minutes for data analysis). The MV was assessed by both Wilkins’s and RT3DE scores. Assessment of MV leaflets and commissures by RT3DE was best performed from the parasternal window, while assessment of the subvalvular apparatus was optimally performed from the apical window. The RT3DE score was applied easily by the investigators in all patients, with identical scores. The distribution of patients by Wilkins’s score was as follows: 5 patients had mild, 3 had moderate, and 9 had severe MV involvement. The distribution by the new RT3DE score was as follows: 2 patients had mild, 4 had moderate, and 11 had severe MV involvement.
Real-time 3D echocardiographic data acquisition and analysis could be performed in all patients for both stages (91 patients). The average time needed for data analysis and score calculation in patients with sinus rhythm (67 patients) was 10 minutes, compared with 13 minutes in patients with atrial fibrillation (24 patients).
RT3DE helped in more intensified descriptions of each MV scallop mobility and thickness than global evaluation by 2DE ( Video 2 ; view video clip online.). Calcification of MV leaflets and annulus was fairly detected by 2DE, while RT3DE added more information regarding the distribution of calcification on leaflet’s parts and its extent ( Figures 2 A-2C). RT3DE allowed detailed assessments of chordal thickening and separation, while on 2DE, chordal separation was not clear ( Figure 2 D). The degree of commissural splitting (partial or complete) after PTMV could be assessed well by RT3DE but was inadequate on 2DE.
Quantitative assessments of all patients (91 patients) by both Wilkins’s and RT3DE scores are presented in Table 2 . Both scores were correlated ( r = 0.60, P = .007). Correlations between both scores for the assessment of leaflet score and calcification were r = 0.63 ( P < .0001) and r = 0.44 ( P < .0001), respectively, while no correlations were detected for the assessment of leaflet mobility and subvalvular apparatus affection ( r = 0.047, P = .96, and r = 0.14, P = .91, respectively).
|Thickness ( r = 0.63, P < .0001)||Wilkins’s||0 (0%)||36 (40%)||50 (55%)||5 (5%)|
|RT3DE||0 (0%)||22 (24%)||56 (62%)||13 (14%)|
|Mobility ( r = 0.047, P = .96)||Wilkins’s||3 (3%)||44 (48%)||35 (39%)||9 (10%)|
|RT3DE||14 (16%)||31 (34%)||46 (51%)||0 (0%)|
|Calcification ( r = 0.44, P < .0001)||Wilkins’s||36 (40%)||35 (38%)||17 (19%)||3 (3%)|
|RT3DE||23 (25%)||35 (38%)||28 (31%)||5 (6%)|
|Subvalvular apparatus ( r = 0.14, P = .91)||Wilkins’s||0 (0%)||35 (39%)||41 (45%)||15 (16%)|
|RT3DE||2 (2%)||23 (25%)||50 (55%)||16 (18%)|
|Total score ( r = 0.60, P = .007)||Wilkins’s||0 (0%)||32 (35%)||50 (55%)||9 (10%)|
|RT3DE||0 (0%)||14 (16%)||54 (59%)||23 (25%)|
The total assessment of MV involvement showed that both scores were comparable for the definition of moderate severity ( r = 0.62, P = .007), while those for low and high severity were not. As shown in Table 3 and Figure 3 , the total RT3DE scores of patients who underwent PTMV (74 patients) were severe in 12 patients (16%). Compared with Wilkins’s score, the RT3DE scores for leaflet thickness, calcification, and subvalvular affection were significantly more often in the severe category.
|Thickness ( r = 0.66, P = .0001)||Wilkins’s||0 (0%)||27 (36.7%)||42 (56.7%)||5 (6.7%)|
|RT3DE||0 (0%)||15 (20.3%)||49 (66.2%)||10 (13.5%)|
|Mobility ( r = .18, P = .87)||Wilkins’s||2 (2.7%)||32 (43.3%)||32 (43.3%)||8 (10.7%)|
|RT3DE||3 (4%)||27 (36.5%)||40 (54.1%)||4 (5.4%)|
|Calcification ( r = 0.58, P = .0001)||Wilkins’s||22 (30%)||34 (46%)||18 (24%)||0 (0%)|
|RT3DE||10 (13.5%)||34 (46%)||24 (32.4%)||6 (8%)|
|Subvalvular apparatus ( r = .47, P = .69)||Wilkins’s||0 (0%)||29 (39.2%)||37 (50%)||8 (10.7%)|
|RT3DE||0 (0%)||17 (23%)||44 (59.5%)||13 (17.5%)|
|Total score ( r = 0.33, P = .004)||Wilkins’s||0 (0%)||26 (35.2%)||48 (64.8%)||0 (0%)|
|RT3DE||0 (0%)||9 (12.3%)||53 (71.7%)||12 (16%)|
Wilkins’s score showed fair agreement for the assessment of leaflet thickness and mobility (κ = 0.55 and 0.56, respectively), while agreement was poor for calcification and subvalvular affection (κ = 0.03 and 0.01, respectively). For RT3DE, agreement was good for the assessment of leaflet thickness and mobility (κ = 0.66 and 0.63, respectively) and fair for the assessment of calcification (κ = 0.42). Assessment of subvalvular apparatus showed slight agreement (κ = 0.21) for chordal thickening, while for splitting, it was excellent (κ = 0.95).
For Wilkins’s score, agreement was good for the assessment of leaflet thickness (κ = 0.65), fair for leaflet mobility and calcification (κ = 0.58 and 0.41, respectively), and poor for the subvalvular apparatus (κ = 0.14). For RT3DE, agreement was good for the assessment of leaflet thickness, mobility, and calcification (κ = 0.71, 0.61, and 0.62, respectively). Assessment of subvalvular apparatus by RT3DE score showed very good agreement (κ = 0.91) for chordal splitting and fair agreement (κ = 0.41) for thickening.
Immediate Results of PTMV
Using the definition of success described above, optimal successful PTMV was achieved in 57 patients (77%) and suboptimal success in 17 patients (23%). The immediate echocardiographic and hemodynamic outcomes are shown in Table 4 . Serious complications, including cardiac tamponade and urgent surgical interference, were not present.
|Variable||Before PTMV||After PTMV||P|
|MVA, 2DE (cm 2 )||0.91 ± 0.13||1.9 ± 0.30||<.0001|
|MVA, RT3DE (cm 2 )||0.92 ± 0.14||2.1 ± 0.30||<.0001|
|Mean pressure gradient (mm Hg)||20.0 ± 5.8||7.0 ± 2.5||<.0001|
|LA diameter (cm)||5.1 ± 0.95||4.7 ± 7.7||<.001|
|LV fractional shortening (%)||31.7 ± 6.8||35.1 ± 5.2||.002|
|No||37 (50%)||5 (6.6%)|
|Grade 1||32 (43.3%)||37 (50%)|
|Grade 2||5 (6.7%)||15 (20%)|
|Grade 3-4||0 (0%)||17 (23.3%)|
By 2DE, complete splitting of the anterior commissure was achieved in 48 patients (65%) and partial splitting in 18 patients (25%). Complete splitting of the posterior commissure was achieved in 37 patients (50%) and partial splitting in 22 patients (30%). Bilateral complete commissural splitting was obtained in 37 patients (50%).
By RT3DE, splitting of both commissures could be assessed clearly ( Figure 4 ). Complete splitting of the anterior commissure was achieved in 46 patients (62%) and partial splitting in 22 patients (30%). Complete splitting of the posterior commissure was achieved in 37 patients (50%) and partial splitting in 18 patients (25%).
PTMV Immediate Results and Wilkins’s Score
Leaflet calcification was the only predictor of successful PTMV ( P = .025). No independent predictors could be identified for the development of grade >2 mitral regurgitation. Considering a score of 8 as the cutoff point for optimal success, a Wilkins’s score of 8 was seen in 77% of optimal successful PTMV procedures. The area under the ROC curve was 0.80.
Immediate Results of PTMV and RT3DE Score
Univariate analysis using Cox proportional-hazard ratios showed that leaflet mobility, calcification, and subvalvular thickness by RT3DE were the predictors of successful PTMV ( P = .071, .030, and .004, respectively). In a multivariate model using a forward selection algorithm, leaflet calcification and subvalvular thickness remained independent predictors of successful PTMV ( P = .019 and .023, respectively) ( Table 5 ).