Multimodality imaging derived energy loss index and outcome after transcatheter aortic valve replacement

Abstract

Aims 

To assess whether the combination of transthoracic echocardiography (TTE) and multidetector computed tomography (MDCT) data affects the grading of aortic stenosis (AS) severity under consideration of the energy loss index (ELI) in patients undergoing transcatheter aortic valve replacement (TAVR).

Methods and results 

Multimodality imaging was performed in 197 patients with symptomatic severe AS undergoing TAVR at the University Hospital Zurich, Switzerland. Fusion aortic valve area index (fusion AVAi) assessed by integrating MDCT derived planimetric left ventricular outflow tract area into the continuity equation was significantly larger as compared to conventional AVAi (0.41 ± 0.1 vs. 0.51 ± 0.1 cm²/m²; P < 0.01). A total of 62 patients (31.4%) were reclassified from severe to moderate AS with fusion AVAi being >0.6 cm²/m². ELI was obtained for conventional AVAi and fusion AVAi based on sinotubular junction area determined by TTE (ELI_LTL 0.47 ± 0.1 cm²/m²; fusion ELI_LTL 0.60 ± 0.1 cm²/m²) and MDCT (ELI_MDCT 0.48 ± 0.1 cm²/m²; fusion ELI_MDCT 0.61 ± 0.05 cm²/m²). When ELI was calculated with fusion AVAi the effective orifice area was >0.6 cm²/m² in 85 patients (43.1%). Survival rate 3 years after TAVR was higher in patients reclassified to moderate AS according to multimodality imaging derived ELI (78.8% vs. 67%; P = 0.01).

Conclusion 

Multimodality imaging derived ELI reclassifies AS severity in 43% undergoing TAVR and predicts mid-term outcome.

Introduction

Accurate assessment of aortic stenosis (AS) severity is critical for the correct management of such patients.^1–3 This has become particularly important during the last years because transcatheter aortic valve replacement (TAVR) has markedly increased the number of patients eligible for aortic valve replacement.^4,5 Two-dimensional transthoracic echocardiography (TTE) is the first-line imaging modality applied for the diagnosis of AS. The current criteria defining severe AS include peak transvalvular flow velocity >4.0 m/s, mean transvalvular pressure gradient >40 mmHg, and indexed aortic valve area (AVAi) <0.6 cm²/m² as determined by the continuity equation.^6–8

A constellation of echocardiographic parameters such as AVAi <0.6 cm²/m² and mean transvalvular pressure gradient <40 mmHg is defined as low-gradient AS and raises uncertainty about the true stenosis severity and the appropriate therapeutic decision.⁹ Recent studies have suggested that underestimation of the left ventricular outflow tract (LVOT) area by 2D echocardiography due to the false assumption of a circular LVOT area and difficulties in defining the correct LVOT diameter may represent a particularly important problem leading to AVA underestimation and that this systematic error can be corrected by planimetry of the LVOT area in multidetector computed tomography (MDCT).^10–14

Assessment of AS severity by TTE is based on the continuity equation and does not account for the phenomenon of pressure recovery which is known to occur in the proximal ascending aorta, mainly in individuals with small aortic diameter.^15,16 To overcome this limitation, an adjustment of the Doppler derived AVA including the sinotubular (ST) junction area and accounting for pressure recovery has been introduced and termed energy loss index (ELI). ELI provides prognostic information in patients with asymptomatic AS independent of peak transvalvular flow velocity or mean transvalvular pressure gradient.^17,18 However, calculation of ELI is affected by the systematic error introduced through underestimation of LVOT area in 2D echocardiography.

This study sought to investigate the diagnostic impact of ELI on the grading of AS severity in patients undergoing TAVR, and to define its role in outcome prediction after TAVR.

Methods

Patient selection

A total of 554 consecutive patients with symptomatic severe AS as determined by an indexed AVA <0.6 cm²/m² and undergoing TAVR at the University Hospital Zurich, Switzerland, were retrospectively analysed for the current study. Only patients with on-site TTE and on-site MDCT examination as well as on-site analysis were included in the study. Hence, a total of 197 patients with documented on-site TTE and MDCT were evaluated. Patients with previous aortic valve replacement and severe mitral regurgitation were excluded. Data collection was performed according to national as well as institutional guidelines and in the context of a nation-wide prospective registry (SWISS TAVI Registry NCT01368250). All patients had provided written informed consent for data analysis.

Echocardiography

TTE was performed using commercially available ultrasound systems (Philips iE33, Philips Healthcare, Zürich, Switzerland; GE E9 or Vivid 7, GE Healthcare, Glattbrugg, Switzerland). Acquired images were analysed using a multi-modality image management system (Xcelera, Philips Healthcare). Certified staff members performed analysis of parameters and measurements in the on-site core laboratories. Interobserver variability analysis confirmed good correlation of TTE measurements for LVOT and ST junction diameters (LVOT: r = 0.85; ST: r = 0.88).

Multidetector computed tomography

MDCT examination was performed as previously described.¹⁴ For image analysis, dedicated software was applied (3mensio Structural Heart 7.2, PIE Medical Imaging BV, Maastricht, the Netherlands). For quantitative analysis, a centreline within the aortic root was drawn along the annulus plane (Supplementary data online, Figure S1A), which was defined by the nadir of each aortic leaflet (Supplementary data online, Figure S1B, level B, picture 2). Along the centreline within the aortic root the maximal and minimal diameter (i.e. the diameter orthogonal to the maximal diameter), the perimeter and the area were assessed at the following levels: the aortic annulus plane, the ST junction, and the LVOT (Supplementary data online, Figure S1B). At the level of the sinus of Valsalva, only the maximal and minimal diameters were measured. Additionally, the diameter equivalent to the echocardiographic LVOT diameter was determined in the long-axis view (Supplementary data online, Figures S2 and S3). As for TTE measurements, interobserver variability analysis confirmed excellent correlations between LVOT and ST diameter measurements (LVOT: r = 0.93; ST: r = 0.98).

Multimodality imaging for assessment of aortic valve area and ELI

In TTE, LVOT diameter was measured 5 mm below the aortic annulus during mid-systole in parasternal long-axis view and LVOT area was calculated assuming a circular shape of the LVOT as recommended.^6,16–18 The diameter of the ST junction was assessed in the same view using leading edge to leading edge (LTL) measurements.

In MDCT, LVOT area was planimetered 5 mm below the aortic annulus plane as defined by the nadir of each aortic valve leaflet. In patients with LVOT calcifications, LVOT area was planimetered both including and excluding the calcified area in order to quantify the impact of LVOT calcification. Similarly, ST junction area was determined by planimetry in MDCT. The flow velocity time integral (VTI) in the LVOT was measured in TTE by pulsed wave Doppler 5 mm below the aortic annulus in apical five-chamber view. The VTI across the aortic valve was determined using continuous wave Doppler. Indexed stroke volume was determined using calculated LVOT areas in TTE as well as planimetered LVOT areas in MDCT. Left ventricular volume and ejection fraction were quantified using the biplane Simpson’s method. Left ventricular mass was derived from M-mode recordings in parasternal long-axis view.

AVA was calculated in TTE by the continuity equation according to current guidelines and indexed to body surface area (AVAi). The so-called fusion AVAi was determined by inserting the planimetric LVOT area measured in MDCT into the continuity equation. To evaluate the impact of pressure recovery the ELI was assessed as previously described.^16–18 ELI was determined using calculated ST junction area in TTE or planimetered ST junction area in MDCT and calculated as follows: ELI = [(AVA × AA)/(AA − AVA)]/BSA, where AA is the cross-sectional area of the aorta measured at the ST junction and BSA is the body surface area.

Statistical analysis

The data are presented as mean (±standard deviation) or median values and interquartile range for normally and non-normally distributed continuous variables, respectively, and as counts/percentages for categorical variables. Comparisons between categorical variables were made using the Chi-square or Fisher’s exact test wherever appropriate. Continuous variables with a non-parametric distribution were compared using Mann–Whitney U test. Changes in pressure gradients and valve area were calculated and analysed using repeated measures ANOVA. Student’s t-test and one-way ANOVA were used for comparison of normally distributed continuous variables. Post-hoc comparisons were performed using the Bonferroni test. Kaplan–Meier curves for survival were generated to compare event rates among both groups at 3 years. A P-value <0.05 was considered significant. Pearson’s correlation coefficient analysis was used to assess interobserver variability for TTE and MDCT measurements of the LVOT and ST junction diameters. All the analyses were performed using GraphPad Prism 6 (GraphPad Prism Software, San Diego, CA, USA) and SPSS 21 (IBM Corporation, Armonk, NY, USA).

Results

Baseline characteristics

Table 1 displays the baseline clinical characteristics and Table 2 the baseline echocardiographic findings of the study population.

Reclassification of AS by fusion AVAi

Comparison of LVOT diameters measured by TTE and MDCT demonstrated that echocardiographic measurements performed in the parasternal long-axis view were comparable to the minimal diameter measured in MDCT (mean values echo: 20.7 ± 1.2 mm; CT min: 20.4 ± 1.5 mm; P = NS; Figure 1A). However, both TTE and MDCT minimal diameter were consistently smaller than the maximal diameter in MDCT (mean value CT max: 27.5 ± 1.8 mm; P < 0.01 vs. echo and vs. CT min; Figure 1A). Calculation of the sphericity index defined as the ratio of minimal to maximal diameter in MDCT underscored the ovoid shape of the LVOT (sphericity index = 0.75 ± 0.1; Figure1B andC). Inclusion of calcifications resulted in significantly larger values for planimetered LVOT area (4.6 ± 0.33 vs. 4.1 ± 0.26 cm²; P < 0.05; data not shown). Exclusion of calcifications for the planimetry of the LVOT area was considered to represent the haemodynamically relevant LVOT area and was therefore used for the remainder of this study.

The LVOT area calculated from LVOT diameter measured in TTE was significantly smaller than the LVOT area planimetered in MDCT (echo 3.4 ± 0.12 vs. MDCT 4.5 ± 0.23 cm²; P < 0.001; Figure 1D). Bland-Altmann plot analysis demonstrated that TTE consistently underestimated LVOT area as compared to MDCT (Figure 1E).

Indexed aortic valve area determined by inserting the MDCT derived planimetric LVOT area in the continuity equation (fusion AVAi) was significantly larger in all the patients than the conventional AVAi value calculated using echo parameters only (AVAi 0.41 ± 0.1 vs. fusion AVAi 0.51 ± 0.1 cm²/m²; P < 0.01; Figure 2A). Under these conditions, 62 patients were reclassified because fusion AVAi was >0.6 cm²/m² and thus no longer consistent with severe AS (Figure 2B).

Reclassification of AS severity by fusion ELI

Similar to LVOT dimensions, the diameter and area of the ST junction were compared using both imaging modalities. The diameter of ST junction measured by TTE using the LTL method was similar to the mean ST junction diameter determined by MDCT (Figure 3A). The sphericity index was 0.94 ± 0.4 indicating an almost circular anatomy of the ST junction as opposed to the ovoid shape of the LVOT (Figure3B andC). There was no difference in ST junction area derived from LTL diameter as compared to the planimetric assessment of ST junction area by MDCT (echo 6.6 ± 0.32 vs. MDCT 6.2 ± 0.43 cm²; P = NS; Figure3D andE).

To evaluate the impact of pressure recovery on the classification of AS severity, ELI was calculated for both AVAi obtained from TTE and from MDCT data fusion. Each of these calculations was performed using ST junction area as determined by TTE and MDCT. Calculation of ELI using AVAi obtained from TTE resulted in a significantly larger AVA irrespective of whether the ST junction area was determined by TTE or MDCT [AVAi 0.41 ± 0.1 vs. ELI_LTL 0.47 ± 0.1 cm²/m² (P < 0.001) and vs. ELI_MDCT 0.48 ± 0.1 cm²/m² (P < 0.01); Figure 2C]. Aortic valve area was >0.6 cm²/m² in 31 (ST area from TTE) and 44 patients (ST area from MDCT), respectively (Figure 2D). Calculation of ELI based on fusion AVAi resulted in an even larger AVA, again irrespective of whether the ST junction area was determined by TTE or MDCT [fusion AVAi 0.51 ± 0.1 vs. fusion ELI_LTL 0.60 ± 0.1 cm²/m² (P < 0.01) and vs. fusion ELI_MDCT 0.61 ± 0.05 cm²/m² (P < 0.01); Figure 2E]. Aortic valve area was reclassified >0.6 cm²/m² in 83 (ST area from TTE) and 85 patients (ST area from MDCT), respectively (Figure 2F). Hence, a total of 85 patients (43%) initially considered to have severe AS had to be reclassified when ELI was calculated with the MDCT derived planimetric area for both LVOT and ST junction and according to currently recommended cut-off values (>0.6 cm²/m²). In a previous sub-study of the SEAS trial the authors reported that the best cut-off value to predict outcomes in patients with AS was larger with ELI (0.76 cm²/m²) than for AVAi (0.6 cm²/m²).¹⁸ If an adjusted cut-off value of 0.76 cm²/m² was applied, 56 patients (28.4%) were reclassified according to fusion ELI.

Even though the number of reclassified patients tended to be higher when the ELI calculation was based on ST junction area derived from MDCT as compared to TTE, differences between groups did not reach statistical significance (Figure2C andE and Supplementary data online, Figure S4). In contrast, ELI derived from TTE was consistently smaller than the one derived from MDCT data fusion demonstrating that calculation of AVAi and ELI are both affected by a systematic error and that data fusion should be considered for either parameter (Supplementary data online, Figure S4).

Morphological, haemodynamic, and clinical characteristics of patients reclassified by fusion ELI

Analysis of echocardiographic, baseline clinical, and procedural characteristics between patients with true severe AS and those reclassified according to fusion ELI are displayed in Tables 3–5. The prevalence of pressure recovery expressed as the difference between fusion ELI and fusion AVAi was higher in the patients with smaller ST junction area (Supplementary data online, Figure S5).

No difference in stroke volume index determined by TTE was observed between reclassified patients and those with true severe AS. In contrast, fusion SVI calculated using the planimetered LVOT area from MDCT increased significantly as compared to the TTE value and was markedly larger in reclassified patients (55 ± 14 vs. 35 ± 11 mL/m²; P < 0.01). While 76% (n = 150) of patients exhibited a low-flow severe AS when SVI was determined by TTE, the majority of these patients was reclassified into the normal-flow category when multimodality imaging was used for calculation of fusion SVI (Figure 4). Under these conditions, 82% (n = 48) of the patients presenting with a normal-flow low-gradient constellation exhibited a fusion ELI >0.6 cm²/m² (Figure 4) suggesting that this constellation is a hallmark of moderate AS.

Clinical outcomes of patients reclassified by fusion ELI

Follow-up was available for all the patients. Functional status after 1 year determined by NYHA classification had improved in both patients with true severe AS (fusion ELI <0.6 cm²/m²) and those who had been reclassified (fusion ELI >0.6 cm²/m²; Figure 5A). Survival 3 years after TAVR was lower in patients with true severe AS as compared to those who had been reclassified by fusion ELI (fusion ELI >0.6 cm²/m²; 67% vs. 78.8%; P = 0.01; Figure 5B). In patients reclassified according to fusion AVAi only survival rates at 3 years did not differ between the groups (Supplementary data online, Figure S6).

Discussion

The current study highlights the following aspects:

1. multimodality imaging derived ELI reclassifies AS severity

2. multimodality imaging derived ELI predicts outcome after TAVR

3. benefit of TAVR in symptomatic patients with reclassified moderate AS

Assessment of LVOT and ST junction dimensions

In current clinical practice, the LVOT diameter is measured in 2D TTE in the parasternal long axis, whereby the diameter is determined 2–10 mm below the aortic annulus.⁶ The current study confirms previous observations demonstrating that the LVOT exhibits an elliptic rather than circular anatomy.¹³ This appears to be particularly relevant in patients with severe AS in whom increased fibrocalcific remodelling enhances the stiffness of the LVOT leading to a pronounced elliptic shape during systole.^19,20 As a consequence, the LVOT area calculated with the LVOT diameter measured in TTE consistently underestimates the true anatomical LVOT area as compared to the planimetered area in MDCT. In contrast to the LVOT, the ST junction displays a circular anatomy as reflected by a sphericity index approximating 1. Consistent with this, no significant difference was observed between ST junction diameters measured in TTE according to the LTL method and the minimum, maximum, and mean diameters determined in MDTC.

Reclassification of AS patients by multimodality imaging

Fusion of data derived from TTE and MDCT results in larger LVOT areas and consequently larger AVA as compared to measurements derived from TTE alone.^10,11 The present study introduces for the first time a combined multimodality imaging for assessment of both LVOT and ST junction dimensions to quantify the severity of AS in patients undergoing TAVR. Integration of planimetered areas for LVOT and ST junction with correction for pressure recovery by calculation of fusion ELI leads to the reclassification of 43% of AS. Of note, reclassification occurred mostly in patients categorized as normal-flow low-gradient AS, suggesting that multimodality data fusion for improved estimation of AS severity is particularly important in this patient population. Even though at a lesser extent, reclassification also occurred in patients displaying a mean Doppler-derived pressure gradient >40 mmHg (normal-flow high-gradient). This discrepancy is most likely explained by the fact that the pressure gradient determined by Doppler at the vena contracta overestimates the net pressure gradient after pressure recovery in the proximal aorta.¹⁶

Assessment of aortic valve calcium by MDCT has been proposed to differentiate patients with true severe AS from those with inconsistently graded severe AS as defined by an AVA <1 cm² and a low transvalvular pressure gradient.²¹ Analyses of isolated aortic valve specimens in patients undergoing surgical aortic valve replacement demonstrated that the weight of excised aortic valves was lower in patients with paradoxical low-flow low-gradient AS than in patients with high-gradient severe AS.²² Our observation that patients reclassified according to fusion ELI display less aortic valve calcification and have less advanced aortic valve disease is in line with these data.

Impact of AS reclassification on clinical decision-making: is it truly moderate AS?

The current study investigates for the first time the effect of fusion imaging for calculation of pressure recovery and true aortic valve effective orifice area in patients diagnosed with severe symptomatic AS undergoing TAVR. Bahlmann et al.¹⁸ demonstrated that ELI was superior to AVAi in predicting adverse clinical events in patients with moderate-to-severe AS and identified a value of 0.76 cm²/m² as the best predictive cut-off value in this population. Hence, depending on the cut-off value (0.6 or 0.76 cm²/m²) applied, fusion ELI reclassified either 43% or 28% of patients diagnosed with severe AS upon TTE examination. Whether adjustment of AVAi for both the elliptic shape of the LVOT and pressure recovery based on fusion of echocardiography and MDCT data provide incremental prognostic information in comparison to standard parameters needs to be determined in large-scale studies. Considering the lack of prospective long-term data, current guidelines do not recommend routine assessment of pressure recovery and fusion of echocardiography and MDCT data for AS grading.⁶

It is not known whether AVR should be recommended in all patients with moderate AS irrespective of the presence of additional conditions such as left ventricular dysfunction.² AS and related symptoms do not always progress in parallel. On the one hand, patients with severe AS may be asymptomatic, on the other hand, patients with moderate AS may display symptoms related to impaired systolic and/or diastolic left ventricular function.^23–25 Since concomitant significant CAD was more frequent in patients reclassified to moderate AS, we cannot exclude that coronary revascularization may have contributed to the functional improvement observed at 1-year follow-up.

Patients reclassified to moderate AS by fusion imaging appeared to display less advanced aortic valve disease at baseline as determined by less aortic valve calcification, lower transvalvular pressure gradients, lower LV mass, lower NT-proBNP level, and a better functional status than those with true severe AS. Nevertheless, TAVR resulted in a symptomatic improvement after 1 year in these patients as well. The excellent device success rates and the lower mortality 3 years after TAVR compared to true severe AS patients confirm the feasibility, safety, and efficacy of TAVR in patients reclassified to moderate AS, which exhibit less advanced aortic valve disease. On the other hand, reclassification of AS severity by multimodality imaging identifies a subset of patients with true severe AS at a higher risk which may benefit from a tighter follow-up after TAVR.

In summary, ELI corrected for true LVOT and ST junction area reclassified a large number of patients initially diagnosed with severe AS. While TAVR resulted in a similar functional improvement of both reclassified patients and those with true severe AS, the reclassified group displayed lower rates of all-cause mortality at 3-year follow-up, suggesting that TAVR may a valuable therapeutic strategy also in symptomatic patients with moderate or borderline moderate-to-severe AS. Large-scale randomized trials are warranted to confirm the clinical relevance of fusion ELI in the assessment of AS severity, along with the role of aortic valve replacement in the management of symptomatic moderate AS.

Study limitations

Some limitations merit consideration. The first limitation lays in the retrospective nature of the study. Furthermore, a certain selection bias cannot be excluded as only patients with on-site TTE and MDCT data were included in the analysis. As for previous studies and in line with current guidelines, right heart catheterization and retrograde crossing of the aortic valve was not routinely performed in this patient cohort and thus, confirmation of the estimated pressure recovery by invasive assessment of haemodynamics was precluded.

Acknowledgements

All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.

Funding

The study was supported by a research grant from the Swiss Heart Foundation and a research grant from the Walter and Gertrud Siegenthaler Foundation, both to E.W.H.

Conflict of interest: none declared.

References

Author notes