This study evaluated 2-dimensional (2D) transthoracic echocardiography (TTE) using Valve Academic Research Consortium–2 (VARC-2) criteria and 3-dimensional (3D) TTE for assessment of aortic regurgitation (AR) after transcatheter aortic valve implantation (TAVI) in comparison with cardiac magnetic resonance (CMR) imaging. In 71 patients, 2D TTE, 3D TTE, and CMR imaging were performed to assess AR severity after TAVI. Using 2D TTE, AR severity was graded according to VARC-2 criteria and regurgitant volume (RVol) was determined. Three-dimensional color Doppler TTE allowed direct planimetry of the vena contracta area of the paravalvular regurgitation jet and calculation of the RVol as product with the velocity-time integral. RVol by CMR imaging was measured by phase-contrast velocity mapping in the ascending aorta. After TAVI, mean RVol determined by CMR imaging was 9.2 ± 9.6 ml/beat and mean regurgitant fraction was 13.3 ± 10.3%. AR was assessed as none or mild in 58 patients (82%) by CMR imaging. Correlation of 3D TTE and CMR imaging on RVol was better than correlation of 2D TTE and CMR imaging (r = 0.895 vs 0.558, p <0.001). There was good agreement between RVol by CMR imaging and by 3D TTE (mean bias = 2.4 ml/beat). Kappa on grading of AR severity was 0.357 between VARC-2 and CMR imaging versus 0.446 between 3D TTE and CMR imaging. Intraobserver variability for analysis of RVol of AR after TAVI was 73.5 ± 52.2% by 2D TTE, 16.7 ± 21.9% by 3D TTE, and 2.2 ± 2.0% by CMR imaging. In conclusion, 2D TTE considering VARC-2 criteria has limitations in the grading of AR severity after TAVI when CMR imaging is used for comparison. Three-dimensional TTE allows quantification of AR with greater accuracy than 2D TTE. Observer variability on RVol after TAVI is considerable using 2D TTE, significantly less using 3D TTE, and very low using CMR imaging.
Transcatheter aortic valve implantation (TAVI) has become a therapeutic option for high-risk or inoperable patients with severe symptomatic aortic valve stenosis. Paravalvular aortic regurgitation (AR) is a significant limitation with negative impact on patient outcome. Precise grading of paravalvular AR has considerable limitations using 2-dimensional (2D) transthoracic echocardiography (TTE). To improve uniformity in the assessment of AR severity after TAVI using 2D echocardiography, criteria for quantification have been defined by the Valve Academic Research Consortium (VARC). However, most of the parameters included in the VARC-2 criteria have initially been applied to evaluate native valve regurgitation, no guidelines are provided by VARC-2 in case of divergence in AR severity grading among the included parameters, and the grading system has not been validated against quantitative parameters. Cardiac magnetic resonance (CMR) imaging has gained acceptance for accurate quantification of valvular regurgitation. Three-dimensional (3D) TTE has been shown to allow accurate quantification of regurgitation severity in native valves and to improve quantification of regurgitation severity with surgical prosthetic valves. This study sought (1) to define the severity of AR after TAVI using CMR imaging, (2) to determine the accuracy of 2D TTE considering VARC-2 criteria and 3D TTE for grading of paravalvular AR severity after TAVI compared with CMR imaging, and (3) to evaluate the intra- and inter-observer variabilities of 2D TTE, 3D TTE, and CMR imaging for analysis of AR volume and AR fraction after TAVI.
Methods
In 90 consecutive patients (age 81 ± 6 years), in whom TAVI either with the CoreValve ReValving system (Medtronic, Minneapolis, Minnesota) or the Edwards SAPIEN XT valve (Edwards Lifesciences, Irvine, California) was performed 3 months earlier, 2D TTE, 3D TTE, and CMR imaging were planned to assess paravalvular AR severity. Nineteen patients with atrial fibrillation, more than trace pulmonary valve regurgitation, or contraindications for CMR imaging (device therapy) were excluded. Thus, 71 patients formed the study group. In these patients, all imaging studies were performed on the same day. This study was approved by the local ethical committee. All patients gave written informed consent.
Echocardiographic studies were performed with a commercially available echocardiographic system (Vivid E9; General Electric Vingmed, Horten, Norway) and 2D transthoracic probe (M5S, General Electric Vingmed, Horten, Norway). In the parasternal short-axis view, color flow Doppler was used to localize and define the circumferential extent of a paravalvular leak. Diastolic flow reversal was assessed in the descending aorta by pulse-wave Doppler from the suprasternal window. Paravalvular regurgitant volume (RVol) was calculated as the difference between stroke volume in the left ventricular outflow tract and pulmonary flow in systole assessed by pulse-wave Doppler echocardiography. Regurgitant fraction (RF) was calculated by dividing RVol by stroke volume in the left ventricular outflow tract. The velocity-time integral of the AR jet was assessed by planimetry of the area under the continuous-wave Doppler velocity curve from an apical 5-chamber view in diastole. Effective regurgitant orifice area was calculated as ratio of RVol/velocity-time integral of the AR jet. The pressure halftime of the continuous-wave regurgitant velocity curve was defined as a further parameter, with the pressure halftime being set at 600 ms in case of very faint signal due to at most minimal regurgitant. AR after TAVI was graded as none or mild, moderate, or severe according to an integrative approach as recommended by the VARC. This considers a combined analysis of (1) diastolic flow reversal in the descending aorta, (2) circumferential extent of a paravalvular leak determined by color Doppler in the short-axis view, (3) RVol, (4) RF, and (5) effective regurgitant orifice area by 2D TTE. In case of divergence in AR grading among the 5 parameters of the VARC-2 criteria, the grading reached in most parameters was used for final categorization. In addition, AR severity was categorized based on RF defined by 2D TTE only, with AR severity being graded as none or mild if RF was <30%, as moderate if RF was 30% to 49%, and as severe if RF was ≥50%, according to the VARC-2 recommendations.
Three-dimensional echocardiographic studies were performed with the same echocardiographic system used for 2D TTE using a 3D active matrix volume transthoracic probe (4V; General Electric Vingmed). Three-dimensional data sets were obtained by acquiring 4 consecutive beats of the whole left ventricle from the apical 4-chamber view. Three-dimensional color Doppler data sets were obtained after adjusting the color Doppler sample volume from the apical 5-chamber view. Care was taken that the sample volume contended the whole prosthetic valve within the left ventricular outflow tract. Three-dimensional analysis was done offline with the aid of a dedicated software package including an automatic quantification tool for assessment of left ventricular stroke volume by subtracting end-systolic volume from end-diastolic volume (EchoPAC, General Electric Vingmed). Cropping of the 3D color Doppler data set allowed direct visualization of the regurgitant orifice by cutting the jet zone from distal to proximal orthogonal to its origin. This position corresponded to the junction of the smallest cross section between the passage adjacent to the prosthetic valve and regurgitant jet spray. For optimal timing of the measurement of the vena contracta cross-sectional area (VCA) in diastole, the electrocardiogram and the guiding parallel plane analysis windows were considered. The VCA was measured by manual tracing at the smallest site adjacent to the TAVI prosthesis. In case of multiple AR jets, all cross-sectional areas were summed up to a total VCA. RVol by 3D TTE was calculated by multiplying total VCA and velocity-time integral of the AR jet obtained by continuous-wave Doppler from an apical 5-chamber view ( Figure 1 ). RF was calculated by dividing RVol by left ventricular stroke volume assessed by 3D automatic quantification. RF by 3D TTE analysis <30% was graded as none or mild, 30% to 49% as moderate, and ≥50% as severe, according to VARC-2 recommendations. Analysis of echocardiograms was performed by blinded readers with >3 years of 3D echocardiographic experience.
CMR imaging was carried out on a 1.5-T magnetic resonance scanner (Achieva; Philips Healthcare, Best, the Netherlands) equipped with a 5-element cardiac synergy coil for signal reception and a vector electrocardiograph for cardiac synchronization. Based on survey and standard cine imaging, a through-plane velocity-encoded phase-contrast sequence (Q-flow: 35 phases per cardiac cycle; spatial resolution 1.4 × 1.4 × 10 mm; repetition time/echo time/flip angle: 3.9 ms/2.4 ms/15°; breath-hold duration 12 to 18 seconds) was planned orthogonal to the ascending aorta just above the cage of the TAVI prosthesis. Maximum velocity encoding was adapted individually to avoid aliasing. CMR-RVol and CMR-RF were assessed offline on a dedicated magnetic resonance workstation (Extended Workspace, Philips Healthcare, Best, the Netherlands) by blinded readers with >3 years of CMR imaging experience. CMR-RVol was determined as diastolic aortic backward flow. CMR-RF was calculated by division of aortic backward flow by aortic forward flow ( Figure 2 ). RF by 3D TTE analysis <19% was graded as none or mild, 19% to 29% as moderate, and ≥30% as severe.
Inter- and intra-observer variabilities were assessed by analysis of 40 patients. In 20 consecutive patients the CoreValve ReValving system was used, and in 20 consecutive patients the Edwards SAPIEN XT valve was implanted. For each imaging method, 2 observers independently measured aortic RVol and RF to assess interobserver variability. These same studies were reexamined by 1 observer ≥4 weeks apart to determine intraobserver variability.
Statistical analysis was performed with dedicated analysis programs (version 12.2.1.0, MedCalc Software, Mariakerke, Belgium; Simstat 2.5, Kovach Computing Services, Anglesey, Wales; and IBM SPSS Statistics 21.0, IBM, Armonk, New York). Continuous data are presented as mean ± SD. The Pearson correlation coefficient (r) was determined and Bland-Altman analysis was performed to evaluate agreement between RVol and RF determined by 2D TTE, 3D TTE, and CMR imaging. For comparison of correlated correlation coefficients, Meng’s z test was used. Analysis of variance assessed RVol and RF by 3D TTE and CMR imaging as graded by an integrative approach by 2D TTE analysis. Student-Newman-Keuls test was used for post hoc pairwise comparisons. Paired t test was performed to assess 2-tailed probability between the different methods. Kappa statistics was applied to determine the agreement on grading of AR severity by the different imaging methods. Comparison of correlated kappa values was performed as described by McKenzie et al by means of 1,000 bootstrap iterations using Simstat, version 2.6 (Provalis Research, Montreal, Quebec). Inter- and intra-observer variabilities on RVol and RF measurement were determined by analysis of the deviation between remeasurements divided by the mean of both measurements. A p <0.05 was considered significant.
Results
Patient and paravalvular AR characteristics are given in Table 1 . Mean RVol after TAVI determined by CMR imaging was 9.2 ± 9.6 ml and mean RF was 13.3 ± 10.3%. Grading of AR severity based on VARC-2 criteria and based on RF determined by 2D TTE, 3D TTE, and CMR imaging are given in Table 2 . Based on VARC-2 criteria, paravalvular regurgitation severity after TAVI was graded as none or mild in 61 patients (86%), moderate in 8 patients (11%), and severe in 2 patients (3%). In contrast, based on CMR-RF, regurgitation severity after TAVI was graded as none or mild in 58 patients (82%), moderate in 7 patients (10%), and severe in 6 patients (8%). The agreement of AR severity grading between VARC-2 and 2D and 3D TTE-RF compared with CMR-RF is given in Table 3 .
| Variable | n = 71 |
|---|---|
| Age (yrs) | 81 ± 6 |
| Men | 32 (45) |
| Edwards SAPIEN XT valve | 39 (55) |
| CoreValve ReValving system | 32 (45) |
| Logistic EuroSCORE (%) | 21 ± 14 |
| New York Heart Association functional classification | |
| I | 1 (1) |
| II | 9 (13) |
| III | 50 (70) |
| IV | 11 (16) |
| Diabetes mellitus | 23 (32) |
| Hypertension, blood pressure >160/90 mm Hg or medically treated | 58 (81) |
| Hypercholesterolemia, cholesterol level >250 mg/dl or medical treated | 48 (67) |
| Renal insufficiency, glomerular filtration rate <60 ml/min | 26 (36) |
| Smoker | 15 (21) |
| Two-dimensional echocardiography | |
| Ejection fraction (%) | 52 ± 12 |
| Paravalvular aortic jet length (cm) | 1.2 ± 1.3 |
| Paravalvular aortic jet area/LVOT area (%) | 1.3 ± 1.8 |
| Pressure halftime of paravalvular regurgitation (ms) | 529 ± 171 |
| Aortic RVol (ml) | 8.7 ± 9.3 |
| Aortic RF (%) | 13.8 ± 11.3 |
| Three-dimensional echocardiography | |
| Aortic RVol (ml) | 6.8 ± 10.5 |
| Aortic RF (%) | 9.7 ± 13.3 |
| CMR | |
| Aortic RVol (ml) | 9.2 ± 9.6 |
| Aortic RF (%) | 13.3 ± 10.3 |
| Variable | Grading of AR Severity | ||
|---|---|---|---|
| AR None or Mild | AR Moderate | AR Severe | |
| VARC-2 criteria ∗ | 61 (86) | 8 (11) | 2 (3) |
| Two-dimensional TTE-RF | 61 (86) | 10 (14) | 0 (0) |
| Three-dimensional TTE-RF | 64 (90) | 6 (9) | 1 (1) |
| CMR-RF | 58 (82) | 7 (10) | 6 (8) |
| Imaging modality | CMR-RF | ||
|---|---|---|---|
| 1 | 2 | 3 | |
| VARC-2 ∗ | |||
| 1 | 54 | 5 | 2 |
| 2 | 4 | 2 | 2 |
| 3 | 0 | 0 | 2 |
| Two-dimensional TTE-RF † | |||
| 1 | 53 | 6 | 2 |
| 2 | 5 | 1 | 4 |
| 3 | 0 | 0 | 0 |
| Three-dimensional TTE-RF ‡ | |||
| 1 | 58 | 5 | 1 |
| 2 | 0 | 2 | 4 |
| 3 | 0 | 0 | 1 |
∗ Kappa 0.357, 95% CI 0.106 to 0.608.
† Kappa 0.158, 95% CI −0.029 to 0.344.
In 58 patients (82%) there was agreement between VARC-2 and CMR grading of paravalvular AR severity. Considering RF determined by 2D TTE and by CMR imaging, there was agreement in grading of paravalvular AR severity in 54 patients (76%). In 61 patients (86%) there was agreement between 3D TTE and CMR imaging in grading of paravalvular AR severity. Comparison of kappa values on the agreement between 3D TTE and CMR imaging was numerically higher than that of VARC-2 and CMR imaging (0.446 vs 0.357). However, this did not reach statistical significance (p = 0.503). The kappa value on the agreement between 3D TTE and CMR imaging was higher than the kappa value on agreement between 2D TTE and CMR imaging (p = 0.023).
Aortic RVol and RF determined by 2D TTE, 3D TTE, and CMR imaging related to AR severity grade defined by VARC-2 criteria are given in Table 4 . There were significant differences in aortic RVol and RF among the 3 VARC severity groups. In patients with none or mild paravalvular AR, RVol (p <0.001) and RF (p = 0.0002) determined by 3D TTE were lower compared with measurements by CMR imaging. There were no differences in RVol between measurements by 3D TTE and CMR imaging in patients with moderate paravalvular regurgitation (p = 0.6891) and RF (p = 0.8318). Similarly, there were no differences between measurements by 3D TTE and CMR imaging in patients with severe paravalvular regurgitation as defined by VARC-2 criteria (p = 0.930 and p = 0.8767, respectively) ( Table 4 ).
