Article Text

Download PDFPDF

Original article
Cardiac magnetic resonance versus transthoracic echocardiography for the assessment and quantification of aortic regurgitation in patients undergoing transcatheter aortic valve implantation
  1. Henrique B Ribeiro,
  2. Florent Le Ven,
  3. Éric Larose,
  4. Abdellaziz Dahou,
  5. Luis Nombela-Franco,
  6. Marina Urena,
  7. Ricardo Allende,
  8. Ignacio Amat-Santos,
  9. Maria de la Paz Ricapito,
  10. Christophe Thébault,
  11. Marie-Annick Clavel,
  12. Robert Delarochelliére,
  13. Daniel Doyle,
  14. Éric Dumont,
  15. Jean G Dumesnil,
  16. Philippe Pibarot,
  17. Josep Rodés-Cabau
  1. Quebec Heart & Lung Institute, Laval University, Quebec City, Quebec, Canada
  1. Correspondence to Dr Josep Rodés-Cabau, Quebec Heart & Lung Institute, Laval University, 2725 Chemin Ste-Foy, Quebec City, Quebec, Canada G1V 4G5; josep.rodes{at}


Background The transthoracic echocardiographic (TTE) evaluation of the severity of residual aortic regurgitation (AR) following transcatheter aortic valve implantation (TAVI) has been controversial and lacks validation.

Objectives This study sought to compare TTE and cardiac magnetic resonance (CMR) for assessment of AR in patients undergoing TAVI with a balloon-expandable valve.

Methods TTE and CMR exams were performed pre-TAVI in 50 patients and were repeated postprocedure in 42 patients. All imaging data were analysed in centralised core laboratories.

Results The severity of native AR as determined by multiparametric TTE approach correlated well with the regurgitant volume and regurgitant fraction determined by CMR prior to TAVI (Rs=0.79 and 0.80, respectively; p<0.001 for both). However, after TAVI, the correlation between the prosthetic AR severity assessed by TTE and regurgitant volume and fraction measured by CMR was only modest (Rs=0.59 and 0.59, respectively; p<0.001 for both), with an underestimation of AR severity by TTE in 61.9% of patients (1 grade in 59.5%). The TTE jet diameter in parasternal view and the multiparametric approach (Rs=0.62 and 0.59, respectively; both with p<0.001) showed the best correlation with CMR regurgitant fraction post-TAVI. The circumferential extent of prosthetic paravalvular regurgitation showed a poor correlation with CMR regurgitant volume and fraction (Rs=0.32, p=0.084; Rs=0.36, p=0.054, respectively).

Conclusions The severity of AR following TAVI with a balloon-expandable valve was underestimated by echocardiography as compared with CMR. The jet diameter, but not the circumferential extent of the leaks, and the multiparametric echocardiography integrative approach best correlated with CMR findings. These results provide important insight into the evaluation of AR severity post-TAVI.

View Full Text

Statistics from


Transcatheter aortic valve implantation (TAVI) has been established as the gold standard treatment for patients with severe symptomatic aortic stenosis considered inoperable and a reasonable alternative in those deemed at high risk for conventional surgical aortic valve replacement (SAVR).1 However, the occurrence of residual aortic regurgitation (AR) secondary to paravalvular leaks (PAR) remains a major limitation of the procedure.2 ,3 The presence of moderate or severe residual AR has been associated with increased short-term and long-term mortality following TAVI, and even mild PARs have been linked to poorer outcomes.4 ,5 An accurate evaluation of the presence and severity of residual AR post-TAVI is, therefore, of major prognostic importance.

Doppler-echocardiography has been the most common method used for AR assessment following TAVI, but the echocardiographic grading of paravalvular AR is challenging and has not yet been validated.24 The limited body of evidence for the echocardiographic parameters and criteria used in the evaluation of paravalvular AR post-TAVI and the need for their further validation have been highlighted in the recommendations of the Valve Academic Research Consortium (VARC-2).6 Additionally, the Placement of Aortic Transcatheter Valve (PARTNER) trial,7 ,8 a randomised study that included centrally analysed echocardiography, weighted more heavily on the circumferential extent of paravalvular AR,9 although this parameter has neither been well validated.6

Cardiovascular magnetic resonance (CMR) is accepted as a non-invasive and safe technique that permits serial assessment of LV mass, volume and function, and it is considered the ‘gold standard’ method for measuring these parameters.3 ,10 ,11 Furthermore, CMR allows the direct measurement of the severity of AR with high accuracy and reproducibility by using the technique of phase-contrast velocity mapping. This method has been correlated with long-term clinical outcomes in patients with native valve AR.12 However, data on the CMR evaluation of AR in the setting of TAVI have been limited to studies including very small number of patients who had received most likely a self-expandable valve.13 ,14 The aim of this prospective study was to compare AR detection and severity grading using multiparametric transthoracic Doppler-echocardiography (TTE) approach versus CMR in patients undergoing TAVI with a balloon-expandable valve.


Study population

We prospectively enrolled 50 patients diagnosed with severe symptomatic aortic stenosis who were accepted for a TAVI procedure at a single Canadian centre. All patients had a CMR performed within 7 days (median 1 (1, 2) days) before TAVI, and 42 of them had a repeated CMR exam within 30 days (median 6 (3–24) days) following TAVI. The reasons to not repeat the CMR exam after TAVI were pacemaker implantation post-TAVI (n=4), death (n=2) and logistic reasons (n=2). TTE exams were performed before the TAVI procedure (median 15 (7, 35) days) and within 30 days (median 6 (6, 22) days) after the procedure. CMR and TTE exams were performed <7 days apart after TAVI in all cases (median 0 (−1, 5) days) in similar haemodynamic conditions. All clinical events during the follow-up period were defined according to the VARC-2 criteria.6 Details about the TAVI procedure have been provided elsewhere.1 All patients received a balloon-expandable valve (Edwards-SAPIEN, SAPIEN XT or SAPIEN 3 valves, Edwards Lifesciences, Irvine, California). All baseline and procedural characteristics were prospectively collected on preset data collection forms. Baseline comorbidities were defined according to the Society of Thoracic Surgeons predicted risk of mortality criteria and periprocedural events according to the VARC-2 criteria.6

Doppler-echocardiography measurements

All TTE exams were analysed in a central echocardiography laboratory by experienced technicians supervised by a cardiologist. The following measurements were obtained in all patients: aortic annulus diameter, left ventricular ejection fraction (LVEF) calculated with the biplane Simpson method, mean transvalvular gradient calculated with the Bernoulli formula and the valve effective orifice area calculated by the continuity equation. The AR was graded using an integrative multiparametric approach based on semiquantitative and qualitative parameters, which mainly included visual assessment of the number of jets, jet(s) width (parasternal and apical views) and the circumferential extent of PAR regurgitation, as recommended by the American Society of Echocardiography guidelines and the VARC-2, respectively.6 ,15 ,16 AR was classified as none/trace, mild, moderate and severe.7 ,8 ,16 ,17 The number of AR jets as well as the jet(s) width and extent in the ventricle characteristics were assessed in parasternal short-axis and long-axis views and apical three-chamber and five-chamber views. The jet width was measured just below the native or prosthetic valve cusps for central AR and just below the apical border of the prosthetic stent for PAR. In case of disagreement between different echocardiographic parameters in the evaluation of AR severity, a final decision was taken after assessing and interpreting all parameters. The circumferential extent (%) of the prosthetic PAR was assessed in the parasternal short-axis view and classified according to the following definition: no or trace (no regurgitant colour flow or pinpoint jet), mild (extent <10%), moderate (extent of 10%–29%) and severe (extent ≥30%) (figure 1).6

Figure 1

Examples of discrepancies between echocardiographic and cardiac magnetic resonance (CMR) quantification of aortic regurgitation (AR). Example 1: transthoracic echocardiography showing jet arc length in the short-axis view that covers >30% of circumference (A) consistent with severe AR, whereas CMR shows a regurgitant fraction (RF) of 14% (B), which is consistent with mild AR by CMR. Example 2: transthoracic echocardiography showing jet arc length in the short-axis view that covers 10%–20% of circumference (C) consistent with moderate AR, while CMR showed a RF of 36%, which is consistent with severe AR.

CMR measurements

The CMR exams were performed using a 1.5 Tesla Philips Achieva (Philips Healthcare, Best, The Netherlands) scanner operating release 2.6 level 3 and dedicated phased-array cardiac coil during successive end-expiratory breath-holds. Cine imaging of cardiac function was performed by steady-state free precession technique at 30 phases per cardiac cycle (by vectorcardiographic gating) in 8–14 parallel short-axis and two-chamber, four-chamber and two orthogonal left ventricular outflow tract (LVOT) planes (8 mm thickness, 0 mm gap). Typical parameters included TR/TE of 3.4/1.2 ms, flip angle 40°, number of excitations (NEX) of 1, yielding in-plane spatial resolution of 1.6×2 mm. In addition, through-plane phase-contrast (sQFlow SENSE) imaging was performed in the aorta at the sino-tubular junction. Velocity encoding maximum value (Venc) was set at 200 cm/s. Caution was taken to exclude the prosthesis from acquisition slice to avoid artefacts. However, if significant turbulence, aliasing or prosthesis stent-related artefacts were seen in the velocity image, the acquisition was repeated a few millimetres downstream from the valve and/or with a higher-velocity window (velocity was increased by 50 cm/s). CMR imaging parameters were the following: TR/TE of 4.60–4.92/2.76–3.05 ms, flip angle 15°, 24 phases, pixel spacing 1.32–2.07 mm and slice thickness 10 mm and acquisition matrix of 256×208. Each phase-contrast velocity mapping acquisition produced two cine images: one magnitude image and one phase image.

The CMR analysis was performed by two investigators blinded to clinical and TTE results in a central CMR core laboratory. For assessment of AR, a region of interest identifying the aortic root was defined, and flow was integrated for the whole cardiac cycle to provide forward and regurgitant flow through the aortic valve per cardiac cycle. Regurgitant fraction (RF) was calculated as follows: [regurgitant volume (RV)/total forward volume]×100 (figure 1). CMR grades of AR were defined according to RF using similar reference cut-point values as previously described in native AR as follows: none/trace (RF<5%), mild (5%–19%), moderate (20%–29%) and severe (≥30%).17 LV volumes and EFs were calculated with the use of end-diastolic and end-systolic endocardial semiautomated tracings.

Before TAVI, intraobserver agreement for grading of AR weighted κ was 0.99 for CMR and 0.85 for TTE (p<0.001 for both). Interobserver agreement weighted κ was 0.96 for CMR and 0.85 for TTE (p<0.001 for both). After TAVI, intraobserver agreement weighted κ was 0.99 for CMR and 0.82 for TTE (p<0.001 for both). Interobserver agreement weighted κ was 0.92 for CMR and 0.68 for TTE (p<0.001 for both).

Statistical analysis

Continuous variables were tested for distribution normality with the Shapiro–Wilk test and expressed as mean±SD or median (25–75th IQR). Categorical variables are reported as n (%). For comparison of baseline and post-TAVI TTE and CMR parameters, paired Student's t test or Wilcoxon signed-rank test was used, based on the distribution of the change in values. Correlations and agreements between CMR and TTE parameters were assessed by Spearman coefficient, and rates of agreement were evaluated using the weighted κ and Bowker's test of symmetry. Comparison of the accuracy of pre-TAVI and post-TAVI agreement between TTE and CMR was performed with the generalised linear model. To test the interobserver and intraobserver variability, the measurements in 30 random patients were repeated after 2 weeks by the same and by another observer. The weighted κ was used to test the variability. The results were considered significant with p values <0.05. All analyses were conducted using the statistical package SPSS V.19 (SPSS, Chicago, Illinois, USA).


Baseline and procedural characteristics of the study population are shown in table 1.

Table 1

Clinical and procedural characteristics of the study population

Echocardiographic data at baseline and after TAVI

Baseline and post-TAVI TTE characteristics are shown in table 2. According to the multiparametric AR grade, 14.0% of the patients presented moderate or severe AR at baseline and 11.9% after TAVI. Also, following TAVI, the presence of AR was paravalvular in all cases, and there were four patients (9.5%) with concomitant mild central AR. According to the circumferential extent of prosthetic PAR jet(s), nine patients (21.4%) would be considered as having moderate AR and seven (16.7%) as severe AR.

Table 2

Baseline and post-TAVI transthoracic echocardiography characteristics of the study population

CMR data at baseline and after TAVI

Baseline and post-TAVI CMR data are shown in table 3. There were no significant changes in left and right ventricles dimensions and functions parameters following TAVI. According to the CMR AR grade, eight patients (19.1%) would be considered as having moderate AR and three (7.1%) as severe AR post-TAVI.

Table 3

Baseline and post-TAVI cardiac magnetic resonance (CMR) characteristics of the study population

Correlation between TTE and CMR AR grades

The severity of AR at baseline and post-TAVI as determined by multiparametric TTE approach correlated with the RV and RF as determined by CMR, p<0.001 (figure 2). However, better correlations were observed pre-TAVI than post-TAVI for both with RV (Rs=0.79 pre-TAVI vs Rs=0.59 post-TAVI, p<0.001) and RF (Rs=0.80 pre-TAVI vs Rs=0.59 post-TAVI, p<0.001) (table 4). The number of jets and jet width also had modest correlation before TAVI (all with p<0.001), and this correlation was also weaker but still significant (all with p<0.01) after TAVI. The distribution of RF and RV by CMR after TAVI, according to the number of jets by echocardiography, is shown in online supplementary figure S1. On the other hand, the circumferential extent of prosthetic PAR measured on the short-axis view showed poor correlation with AR RV and RF measured by CMR (RV: Rs=0.32, p=0.084; RF: Rs=0.36, p=0.054). The AR grade defined according to the circumferential extent of PAR (VARC-2)6 also correlated poorly with CMR quantitative parameters of AR (RV: Rs=0.34, p=0.027; RF: Rs=0.33, p=0.034).

Table 4

Correlation between parameters of aortic regurgitation measured by transthoracic echocardiography and cardiac magnetic resonance

Figure 2

Quantification of regurgitant volume (A and B) and fraction (C and D) as determined by cardiac magnetic resonance, according to the aortic regurgitation (AR) grade as determined by a multiparametric echocardiography evaluation before (A and C) and after transcatheter aortic valve implantation (TAVI; B and D). Data presented as median (IQR 25–75) and errors bars represent 95% CI.

The agreement between TTE and CMR grades was stronger before TAVI (weighted κ=0.766, p<0.001) as compared with post-TAVI (weighted κ=0.300, p=0.375) (figure 3). Eighty per cent of the patients had the same AR grade at TTE and CMR before TAVI versus 33.3% after TAVI, and this difference was statistically significant (p<0.001). Before TAVI, discrepancies in AR grade were essentially found between grade none/trace and mild (16%) with both under and overestimation (12% and 4%, respectively), and no more than one grade change. After TAVI, the TTE AR grade (multiparametric) was lower than CMR grade in a large proportion of patients (61.9%) reaching one grade in 59.5% and two grades in 2.4% (figure 3). Furthermore, in 14.3% (n=6) of the patients, AR was evaluated as ≤ mild by TTE but was classified as ≥ moderate by CMR (figure 3). Finally, concordance between circumferential extent of AR by TTE (VARC-2) and CMR AR grade was also poor (45.2%) after TAVI, with overestimation and underestimation of at least one grade in 16 (38.1%) and 7 (16.7%) patients, respectively (figures 1 and 4).

Figure 3

Comparison of the aortic regurgitation grade as determined by a multiparametric echocardiographic approach in relation to the grade classification with cardiac magnetic resonance pre-transcatheter aortic valve implantation (TAVI) (A) and post-TAVI (B).

Figure 4

Comparison of the aortic regurgitation (AR) grade as determined by circumferential extent of prosthetic paravalvular regurgitation in relation to the AR grade classification with cardiac magnetic resonance post-transcatheter aortic valve implantation.


Despite recent advances in the transcatheter heart valve systems and the increasing experience of the centres/operators, TAVI is still associated with a much higher rate of PAR as compared with SAVR.2 ,3 ,18 Nonetheless, there has been considerable variability in the frequency and severity of PAR after TAVI, with a rate of moderate/severe AR ranging from none to as much as 24%.2 ,3 Several factors such as the use of different imaging modalities between studies (transthoracic vs transesophageal echocardiography, 2D vs 3D echocardiography or angiography), timing of assessment and the absence of standardised criteria and a central echocardiography core laboratory to evaluate PAR may in part explain these differences.

In the PARTNER trial cohorts A and B, the incidence of residual moderate/severe AR was of 12.2% and 11.8%, respectively, which is similar to the 11.9% observed in our study using a central core laboratory evaluation.7 ,8 ,19 However, the rate of moderate/severe AR post-TAVI as evaluated by CMR was about twofold higher (∼26%). To date, there have been very few data evaluating AR by CMR after TAVI. In a previous small study performed exclusively in patients receiving a self-expandable valve, moderate/severe AR as evaluated by CMR was observed in 12.5% of the patients.13 This may be explained by the fact that although AR quantification was performed by phase-contrast sequences, AR severity cut-points were chosen according to a prior study that used only a volumetric method. This led to higher cut-point values (ie, >30% for AR grades III and IV) and may have been influenced by concomitant valvular diseases such as mitral regurgitation, which is very frequent in patients referred for TAVI.20 Interestingly, a recent study using the same CMR methodology as in the present study (phase-contrast) in patients treated with both self-expandable and balloon-expandable valves, the rate of moderate/severe AR was 19%,14 consistent with our results.

The discrepancies in AR severity grading between TTE and CMR might be, in part, due to the fact that the PAR jets are often multiple, eccentric and of irregular shape.9 ,15 ,16 Furthermore, they are not free jets in the centre of the LVOT but are often confined against the LVOT wall. Acoustic shadowing from the annulus and LVOT calcifications and Doppler attenuation from the prosthetic valve stent may further jeopardise accurate quantification of regurgitant jets and may result in an underestimation of AR severity. Indeed, up to 14% of the patients with moderate/severe AR as determined by CMR were classified by TTE as AR grade ≤ mild, which is consistent with the results of a previous study using self-expandable valves.21 This inaccurate quantification of AR grade may also explain why even mild AR has been related to increased mortality at 2 years in the PARTNER trial.5

In the PARTNER trial,7 ,8 the severity of residual AR was graded according to the American Society of Echocardiography recommendations for native valves,16 but the circumferential extent of prosthetic PARs was weighted more heavily than other parameters for the final assessment of AR grade.9 Importantly, the present study revealed that this parameter has a poor correlation with CMR parameters of AR, mainly due to an overestimation of the severity of AR. In fact, PAR jet diameter showed a better correlation with CMR parameters for the evaluation of AR severity post-TAVI, similar to that obtained with the use of multiparametric TTE AR grading. Still, these methods in TTE presented a poor agreement with CMR AR grade classification, highlighting that multiple parameters and imaging modalities may be considered in the sake of proper grade classification, but not to rely only on the circumferential extent of PAR.

CMR allows accurate and highly reproducible estimation of AR through a direct quantitative assessment of the regurgitation volume and fraction, irrespective of the number and eccentricity of regurgitant jets and the presence of concomitant valve diseases.13 ,14 ,17 ,22 However, the higher costs, lower availability and the presence of relative or absolute contraindications in many patients (eg, those with pacemakers) make CMR unlikely to replace TTE as the primary modality for assessment of PAR and AR grading after TAVI. The results of the present study highlight the potential role of CMR in those patients in whom there is a discrepancy between TTE assessment of transcatheter valve AR and patient's clinical status. The findings of this study also emphasise the need for improving the accuracy of echocardiography for assessment of PAR severity. To this effect, the measurement of the left and right outflow tract stroke volumes may help to improve the estimation of AR RV (difference between left and right stroke volumes) and thus the quantitation of AR severity by TTE. Also 3D transthoracic or transesophageal echocardiography may enhance the detection and quantitation of PAR jets.22 ,23


Even though this is to date the largest cohort of patients evaluated with CMR before and after TAVI, the patients were not consecutive and a selection bias might have played a role in the results. However, the fact that TTE results were similar to those obtained in previous studies including native aortic valve and post-TAVI patients makes this possibility unlikely. In the context of PAR, many of the quantitative or semiquantitative parameters proposed in the American Society of Echocardiography/European Association of Echocardiography guidelines15 are difficult or impossible to measure (eg, vena contracta width, jet width to LVOT diameter ratio) or are less reliable (eg, pressure half time of the continuous wave AR envelope, flow reversal in the descending aorta) due to the acute nature of the regurgitation and the reduced compliance of the LV. There is still a lack of a definitive gold standard for the measurement of AR in the context of TAVI. In the present study, we used CMR as the gold standard and this is supported by previous studies in native AR showing high reproducibility and accuracy, the fact that CMR measurements may be performed away from the site of valve implant and the good correlation of RF as evaluated by CMR with clinical outcomes in native aortic valves.12 Nonetheless, the limited number of patients and the lack of long-term follow-up did not allow us to determine a RF cut-off associated with poorer clinical outcomes. Thus, future studies with a larger number of patients will have to appraise the accuracy of AR measurement by CMR and its correlation with clinical outcomes. The results of this study were obtained in patients undergoing TAVI with a balloon-expandable valve and may not apply to those patients receiving a self-expandable valve.


In patients undergoing TAVI with a balloon-expandable valve, TTE may underestimate or overestimate the severity of residual AR as compared with CMR in a substantial proportion of patients. Although modest, the multiparametric TTE integrative approach showed the best correlation with AR severity determined by CMR. On the other hand, the circumferential extent of prosthetic PAR correlated poorly with the CMR AR severity. The use of CMR in selected patients, particularly in those exhibiting discordances between echocardiography results and clinical outcomes, might help to better quantify the AR grade, and this may in turn translate into the implementation of additional measures (PAR closure, valve-in-valve, SAVR) to improve clinical outcomes.

Key messages

What is already known on this subject?

  • The occurrence of residual aortic regurgitation (AR) secondary to paravalvular leaks remains a major limitation of the transcatheter aortic valve implantation (TAVI) procedure. However, the evaluation of the severity of residual AR by transthoracic echocardiography (TTE) is challenging and has not yet been well validated.

What might this study add?

  • The echocardiographic grading of AR in native aortic valve disease showed a good correlation with cardiac magnetic resonance (CMR). However, this correlation decreased and was only modest following TAVI, with an underestimation of AR severity by TTE as compared with CMR. The jet diameter and the multiparametric echocardiography integrative approach, but not the circumferential extent of the leaks, best correlated with CMR findings after TAVI.

How might this impact on clinical practice?

  • The present results highlight the challenges of an accurate evaluation of AR after TAVI. The use of a multiparametric echocardiography approach, rather than a single parameter such as the circumferential extent of prosthetic AR, should be considered for the accurate quantification of AR post-TAVI. Also, CMR may help to better quantify the AR grade in selected patients, such as for those exhibiting discordances between TTE results and clinical outcomes.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Contributors All authors have read and approved submission of the manuscript. All authors have contributed to this work as follows: substantial contributions to the conception and design, acquisition of data or analysis and interpretation of data; drafting the article or revising it critically for important intellectual content; and final approval of the version to be published.

  • Funding This study was funded, in part, by research grants (MOP-57745 and MOP 126072) from the Canadian Institutes of Health Research (CIHR). HBR is supported by a research PhD grant from 'CNPq, Conselho Nacional de Desenvolvimento Científico e Tecnológico—Brasil'. FLV is supported by a clinical and research fellowship from the 'Fédération Française de Cardiologie'. LN-F received funding via a research grant from the 'Fundación Mutua Madrileña' (Spain). IA-S was supported by the 'Instituto de Salud Carlos III', through a research contract at the Institute of Heart Sciences (Valladolid, Spain). M-AC is supported by a postdoctoral fellowship from CIHR. PP holds the Canada Research Chair in Valvular Heart Disease supported by CIHR.

  • Competing interests RD is consultant for St. Jude Medical. ED is consultant for Edwards Lifesciences. JR-C is consultant for Edwards Lifesciences and St. Jude Medical.

  • Ethics approval The study protocol was performed in accordance with the institutional ethics committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles