Recovery of left ventricular and left atrial mechanics in various entities of aortic stenosis 12 months after TAVI

Transcatheter aortic valve implantation (TAVI) has been shown to improve prognosis of high-risk patients. Data, however, concerning the impact of TAVI on regional and global left atrial (LA) and left ventricular (LV) mechanics in varying entities of severe aortic stenosis (AS) are sparse, particularly in patients with paradoxical low-ﬂow (PLF) AS or with reduced LV ejection fraction (LVEF). This study evaluated the effects of TAVI on LA and LV mechanics in varying entities of AS 12 months after implantation. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Keywords no evidence of a 7 ascending aortic aneurysm with intramural and hypoplastic descending aorta ( ). Transthoracic echocardiography a large ascending aortic aneurysm measuring 74 mm at the aortic root with a dissection ﬂap (white and signiﬁcant thrombus in the false lumen (black ( , Supplementary ). associated with a bicuspid aortic ( Panel ), severe aortic regurgitation, and aortic coarctation , Supplementary data online, 3 ). echocardiography these and the presence of a severediscrete


Introduction
With increasing incidence, aortic stenosis (AS) is currently the most common indication for valve intervention 1 with relevant impact on mortality and morbidity. 2 Severe AS is defined as an effective orifice area (EOA) of ,1.0 cm 2 and a mean gradient of ≥40 mmHg. 3 However, varying subgroups of patients have been identified: i.e. patients with a reduced flow due to reduced left ventricular ejection fraction (LVEF) and patients with only moderate pressure gradients despite preserved LVEF. The latter is called paradoxical low-flow (PLF) AS 4 and may reflect an advanced condition associated with worse prognosis compared with normal-flow, high-gradient AS. 5 Transcatheter aortic valve implantation (TAVI) is a novel therapy for severe AS with pronounced benefit on mortality compared with medical therapy 2 and is non-inferior compared with surgical valve replacement in high-risk patients with a normal flow. 6 However, less is known about the effects of TAVI on patients with a reduced LVEF or with a PLF. There is, moreover, no data concerning the effects on myocardial mechanics or on left atrial (LA) function beyond short-term follow-up. 7,8 In clinical routine, LVEF is commonly used to assess systolic function-a parameter with relevant limitation in describing complex three-dimensional (3D) myocardial motion. Newer echocardiography methods such as speckle tracking echocardiography (STE) allow robust assessment of the longitudinal shortening that is predominantly influenced by subendocardial fibres and is more suited to detect subtle myocardial damage. 9 Recently, Adda et al. 10 reported markedly reduced LV longitudinal shortening in patients with PLF: a finding that may help explain the underlying pathophysiological mechanisms. In the application of STE analysis to the left atrium, three separate components 11 can be characterized: reservoir function, conduit function, and active LA contraction ( Figure 1). There is some evidence that LA function is impaired in severe AS, 12 and improvement in reservoir function after surgical valve replacement has been described, albeit only in low-risk patients. 13 Until now, it has been unclear whether recovery of LA mechanics takes place in high-risk patients undergoing TAVI.
The objective of this study was therefore to assess LA and LV strainderived parameters using speckle tracking in three subgroups of patients with severe AS, including patients with a normal LVEF and normal flow, with a PLF, and with a reduced LVEF 12 months after TAVI.

Study population
We prospectively enrolled 77 consecutive patients with symptomatic severe AS who had undergone baseline transthoracic echocardiography and transfemoral TAVI in our centre between July 2009 and March 2011. Of these, 11 patients died within 12 months. Two deaths were directly related to the TAVI procedure. In the other patients, echocardiography shortly before death showed normal aortic valve function. Valve dysfunction after TAVI was therefore able to be ruled out as the cause of death in these patients. In addition, four patients were excluded due to reduced acoustic windows, five who were referred from remote hospitals were not able to attend the follow-up visit, and three refused the follow-up visit. We accordingly included 54 patients with follow-up echocardiography in our study. Patients were treated with TAVI if the aortic valve area was ,1 cm 2 , if the European System for Cardiac Operative Risk Evaluation score (EuroSCORE; Supplementary data online, Appendix S1) 14 was .20%, or if ≥1 of the following criteria was met: contraindication for surgery, severely reduced pulmonary function, livercirrhosis, or metastatic cancer. Written informed consent was obtained from each patient. The Ethics Committee of the Charité Universitätsmedizin centre approved the study. The normal LVEF and normal flow were defined as an LVEF of ≥50% and stroke volume index (SVI) of ≥35 mL/m 2 (Group 1). PLF was defined as an LVEF of ≥50% and SVI of ,35 mL/m 2 4 (Group 2) and patients with a reduced LVEF, as an LVEF of ,50% (Group 3).

Echocardiography and Doppler measurements
Standard echocardiographic parameters were obtained according to the guidelines of the American Society of Echocardiography (ASE) 15 -17 on a Vivid 7 Dimension (GE Vingmed, Horton, Norway, M4S 1.5-4.0 MHz transducer). LV mass (LVM) was assessed using the Devereux formula 18 and indexed to the body surface area (BSA) calculated by the Mosteller formula. 19 Stroke volume (SV) and SV index (SVI) were obtained using pulsed wave (PW) Doppler-derived velocity time integral (VTI) in the LV outflow tract (LVOT), LVOT diameter, and BSA. In general, LVOT diameter was quantified by transoesophageal echocardiography. Mitral regurgitation (MR) was assessed according to the recommendations of the European Association of Cardiovascular Imaging. 20 For grading aortic regurgitation (AR), we used an integrative approach with emphasis on the flow profile in the thoracic and abdominal aorta, since holodiastolic flow reversal in the descending thoracic aorta indicates at least moderate AR. 15 LV diastolic function was assessed using PW Doppler and PW tissue Doppler imaging (TDI) recordings based on the ASE. 21 Atrial fraction was calculated as the ratio of the VTI of the A wave and that of the transmitral inflow in diastole. 22 2D speckle tracking strain analysis of LV and LA For the assessment of longitudinal speckle tracking, we recorded the following with a frame rate between 60 and 80 frames per second: strain and strain rate of the left ventricle, standard two-dimensional (2D) ultrasound images from the apical long-axis, and two-and four-chamber views. We stored these recordings digitally for offline analysis (EchoPac PC, GE Vingmed, Horton, Norway) as previously described. 23,24 In short, we used a semi-automatic algorithm for tracking the LV myocardial wall, which was divided into 18 segments to obtain the global longitudinal peak systolic strain (GLS) and strain rate (GLSR).
Similarly, the longitudinal strain of the LA septal basal segment was analysed in the four-chamber view, with location of the trigger at the onset of the QRS complex. Offline analysis determined peak positive strain

Inter-and intraobserver variability analyses
Two echocardiographers, blinded to previously obtained data, separately measured longitudinal LV and LA strain (R LA , LA conduit, and contractile function) and data from 13 random patients for interobserver variability analysis. Additionally, an experienced observer calculated strain values twice on two consecutive days for analysis of intraobserver variability.
We employed inter-and intraobserver variabilities to determine the interclass coefficient.

Statistics and figures
All results are expressed as mean + standard deviation (SD). Statistics were calculated using SPSS 19.0 (SPSS, Inc., Chicago, IL, USA). The Mann-Whitney non-parametric test was employed to compare echocardiographic data with baseline and follow-up values. For nominal data analysis, McNemar's test was used. P-values of ,0.05 were considered statistically significant.
The echo templates for Figure

Conventional echocardiography
All standard echocardiographic findings are presented in Table 2. The median interval between baseline echocardiography and valve implantation was 34 + 47 days. The median time from TAVI to the follow-up echocardiographic examination was 359 + 38 days. Between baseline and follow-up examination after TAVI, there were no statistically significant differences regarding rhythm and heart rate. After valve implantation, peak transaortic velocity, mean transaortic systolic gradient, and EOA improved significantly. The incidence of severe prosthesis-patient mismatch (PPM; ,0.65 cm 2 /m 2 ) 6 was low: 2 (6%) patients. After TAVI, no or only mild AR was present  in 17 (31.5%) and 26 patients (48.1%), respectively. In 11 (20.4%) patients, moderate AR occurred. Moderate AR was due to isolated paravalvular regurgitation in 7 patients, isolated transvalvular regurgitation in 1, and a combination of para-and transvalvular regurgitation in 3. No severe AR was detected. The LVEF and the TDI-derived peak systolic myocardial velocity (S ′ ) improved significantly. Interestingly, the grade of MR improved, as shown in Table 2. Moreover, the LVM index decreased significantly in both men and women after TAVI, but still remained on the average above the upper normal limit. 25 LV longitudinal strain and strain rate Global longitudinal peak systolic strain (GLS) improved significantly in our study population from baseline to 1 year after TAVI (214.1 + 3.9 vs. 216.5 + 4.0%, P , 0.001). This enhancement was also present in each apical view ( Table 3 and Figure 2). With regard to the regional analyses, we found significant improvement in basal, medial, and apical segments ( Table 3). In addition, the longitudinal peak systolic strain rate was enhanced significantly after TAVI in every view, as was the global longitudinal strain rate (GLSR) ( Table 3). Accordingly, the number of patients with GLS and GLSR values in the range of normal (cut-off 218% and 21.1/s 21 , respectively) increased after TAVI: 9 (16.7%) vs. 21 (38.9%) and 10 (18.5%) vs. 31 (57.4%), respectively.

LV performance in various entities of AS
As summarized in Table 4, patients with a reduced LVEF had the most reduced longitudinal strain and strain rate. Notably, even patients with a preserved LVEF but low-flow AS presented with a reduced longitudinal strain and strain rate compared with those with a normal LVEF and normal flow ( Table 4). After TAVI, all the three groups showed a significant improvement in global longitudinal strain ( Figure 3) and global longitudinal strain rate. Particularly, in patients with PLF AS, valve implantation had the most benefit and resulted in near normalization of SVI and longitudinal function.

Diastolic LV function
For the entire study population, all standard diastolic parameters prior to and after TAVI are presented in Table 5. E ′ , indicating early diastolic LV filling, increased significantly. However, the isovolumetric relaxation time and the E/E ′ ratio did not change significantly. Parameters of atrial contraction did not improve: A ′ decreased significantly, whereas A-wave VTI and atrial fraction remained unchanged. Six (12.0%) patients had no diastolic dysfunction after TAVI, compared with none at baseline. In 21 (41.2%) patients, diastolic dysfunction improved by at least one grade.
The diastolic LA volume and the diastolic LA volume index did not change. We found no differences concerning LA mechanics between the various entities of AS.

Systolic function
Beneficial mid-and long-term effects of surgical aortic valve replacement in severe AS have been described, 26 but little is known about changes in LV deformation and LA mechanics after TAVI beyond 3 months in high-risk patients. 7,8 Our study demonstrated that high-risk patients with severe AS have remarkably reduced longitudinal strain and strain rates, which illustrates subtle systolic LV dysfunction despite nearnormal LVEF. Importantly, 12 months after aortic valve implantation, significant and clinically relevant improvement in these parameters was observed-whereas LVEF was enhanced to a much lesser extent. These data underscore the value of strain imaging compared with mere LVEF assessment for the estimation of global systolic function. Apparently, this commonly used routine parameter has relevant limitations-particularly in patients with AS. 27 Evolving echocardiography techniques-especially involving STE-derived strain and strain rate-appear more suitable to detect subtle myocardial damage and to characterize improvement in myocardial deformation after TAVI. 28 As previously published, 29 we also found in our patients that reduced longitudinal function was more pronounced in basal and medial segments. Acutely after TAVI, only the longitudinal strain of basal and medial segments improved 30 . However, significant amelioration also occurred in the apical segments after 12 months, which indicates constant long-term recovery of systolic function from the basis to the apex. 31 In addition, we detected regression of LVM, which further indicates favourable reverse remodelling. 31

TAVI in paradoxical low-flow aortic stenosis
About one-third of our patients exhibited paradoxical low-flow AS, reflecting advanced disease and indicating poor prognosis. 5 At baseline, patients with PLF experienced pronounced reduction in longitudinal function compared with those with a preserved LVEF and normal flow. Interestingly, patients with PLF particularly benefit exceptionally from TAVI: SVI, longitudinal strain, and strain rate approach near-normal values.
Despite the fact that the outcome of patients with PLF can improve after surgical valve replacement, 4 a relevant number of these patients are often refused by the surgeon due to relevant co-morbidities and perioperative risk. 32 Furthermore, in 20-70% of patients after surgical valve replacement, PPM becomes apparent, 33 with grievous impact on long-term survival. 34 The prevention of PPM-particularly in patients with PLF-is challenging and of increasing importance. 32 In contrast, PPM is unlikely after TAVI. 35 Our data also indicate excellent haemodynamic performance in PLF, with a mean gradient of 8.3 mmHg and a mean EOA index of 1.08 cm 2 /m 2 . PPM occurred in only 1 patient with PLF. TAVI should consequently be considered, particularly for patients with PLF AS.

Diastolic LV function
In TAVI patients with longstanding LV pressure overload, the question arises whether sustained recovery of LV diastolic function can be achieved. In our patients, E ′ increased significantly, which indicates improvement in LV relaxation. This enhancement occurs immediately after TAVI, 36 and our study showed that this recovery was sustained up to 12 months after TAVI. Prior to valve implantation, all patients demonstrated LV diastolic dysfunction. 21 Interestingly, nearly half of these (41.2%) improved by at least one grade, and in 6 (12%) patients, diastolic function even normalized after TAVI. This indicates remarkable beneficial effects of TAVI on diastolic LV function in these high-risk patients.

LA function
Three phases of LA mechanics (i.e. reservoir function, conduit function, and atrial contraction) can be described in strain analysis 37 (Figure 1). Twelve months after valve implantation, we found Table 4 Differences of systolic LV function in patients with a normal LVEF (LVEF ≥50% and SVI ≥35 mL/m 2 ) and normal flow compared with patients with paradoxical low-flow, low-gradient AS (LVEF ≥50% and SVI <35 mL/m 2 ) and compared with those with a reduced LVEF (LVEF <50%) Group  improvement in reservoir and conduit function, but not in active LA contraction. Like E ′ , reservoir and conduit function are primarily related to LV filling pressures and LV relaxation, respectively. 38 In contrast, the E/E ′ ratio, which is an established parameter for the assessment of LV filling pressure, 21 did not improve in our patients. Since Cameli et al. 39 reported that invasively measured LV filling pressures correlated poorly with the E/E ′ ratio, but well with LA longitudinal deformation analysis, we speculate that LA strain may be more appropriate for estimating LV filling pressures after TAVI. In contrast, our patients' atrial contraction did not improve. Moreover, LA volume remained unchanged. These findings are in contrast to Lisi et al. 13 who recently described a reduction in LA volumes after surgical aortic valve replacement in younger low-risk patients. The differences may arise from less-advanced atrial remodelling with better ability to recover among surgically low-risk patients than found in our population. In summary, the observed improvement in reservoir and conduit function may more correctly be explained by an improvement in LV function than by reverse remodelling of LA. Effects of TAVI on various entities of aortic stenosis

Study limitations
Our study is a single-centre study with only a small number of patients. It, however, does focus on detailed characterization of LV and LA function by using advanced echocardiographic methods. Unfortunately, our patients did not undergo cardiac magnetic resonance imaging, which may have enabled insights into the degree of myocardial fibrosis in the left ventricle, particularly in patients with PLF. Since our study excluded 11 patients who died within 12 months after the procedure, 4 with reduced acoustic windows, and 8 without follow-up visits, our results may be distorted to some extent by a selection bias.
With regard to the various groups with AS, our study demonstrated age and gender preference with older patients in comparison to patients with surgical valve replacement. There were more men in the group with a reduced LVEF and more women in the group with PLF AS.
Finally, in our study, the software algorithm used for the assessment of LA mechanics was originally developed for LV studies and requires further validation. 11

Conclusions
Regardless of the underlying AS entity, TAVI improves global and regional LV and LA mechanics within 12 months. Patients with PLF particularly experience the most pronounced benefit, which suggests that TAVI represents an interesting treatment option. In addition, 2D speckle tracking is a robust method for assessing changes in LV and LA function after TAVI.