- Email: [email protected]
Long Term Follow-Up After Aortic Valve Replacement (Ross Procedure): Echocardiographic Determinants of Ventricular Recovery
Heart, Lung and Circulation (2014) 23, 132–143 1443-9506/04/$36.00 http://dx.doi.org/10.1016/j.hlc.2013.07.008
ORIGINAL ARTICLE
Long Term Follow-Up After Aortic Valve Replacement (Ross Procedure): Echocardiographic Determinants of Ventricular Recovery William M. Wilson a*, Leeanne E. Grigg a, Alexandra Gorelik b, Peter Skillington c a
Department of Cardiology, Royal Melbourne Hospital, Parkville 3052, Australia Melbourne EpiCentre, Royal Melbourne Hospital, Parkville 3052, Australia c Department of Cardiothoracic Surgery, Royal Melbourne Hospital, Parkville 3052, Australia b
Received 10 November 2010; received in revised form 28 June 2013; accepted 10 July 2013; online published-ahead-of-print 17 August 2013
Aims and Methods
This prospective cohort study aimed to assess LV recovery post aortic valve replacement, stratified according to pre-operative valve lesion (aortic stenosis (AS), mixed disease (AS/AR) or aortic regurgitation (AR)), as well as define predictors of persistent LV dilatation post operation. We prospectively followed all patients post Ross procedure performed between 1992 and 2009 by a single surgeon. Echocardiography was performed pre-operatively, at approximately one year post operation then second yearly thereafter.
Results
265 patients were followed for a mean of 6.4 years (range 1–14 years, total 1702 patient-years). Seventy percent were male and mean age was 38.8 12.6 years. The indication for surgery was AS in 44.5% (118), AS/AR in 23.4% (62), and AR in 32.1% (85). Overall mortality was 1.8% and 80% of deaths were non-cardiac. Morbidity was low and the need for pacing was less than 1%. Ninety-nine percent of patients were NYHA class 1 at one year follow up. The indexed LV end diastolic diameter (LVEDDi) decreased significantly postoperation in the AR (3.34 0.39–2.66 0.32 cm/m2, p < 0.001) and mixed (2.85 0.38–2.65 0.30 cm/m2, p = 0.01) groups, whilst the indexed systolic LV dimension decreased significantly post-operation in the AR group (2.26 0.34–1.87 0.27, p < 0.001). At five years post operation, independent predictors for a larger LVEDDi were female gender, a pre-operative LVEDDi >3.0 cm/m2 and the presence of mild aortic regurgitation at one year post-operation. Pre-operative valve lesion was not a predictor. The only independent predictor of a lesser reduction of LVEDDi at five year follow-up was mild post-operative AR, whilst predictors of a lesser reduction in indexed left ventricular end systolic diameter (LVESDi) at five year follow-up included mild post-operative AR and a larger pre-operative LVEDDi. LV wall thickness decreased significantly the AS and AS/AR groups within one year post operation, whilst the neo-aortic root size remained stable throughout follow-up.
Conclusions
Recovery of LV size post Ross procedure is influenced predominantly by the pre-operative LV size, in particular the indexed LV end diastolic diameter. The pre-operative valve lesion was not predictive of larger ventricular dimensions post AVR, but independent predictors of a larger ventricular dimensions post operation included female gender, enlarged pre-operative LVEDDi and the presence of mild AR in the first post operative year. Those with mild post-operative AR did not have progressive LV enlargement, thus the clinical significance of this finding remains unclear.
Keywords
Aortic valve replacement Ross procedure Left ventricular size
* Corresponding author. Tel.: +1 6479977745; fax: +61 393423130., Email: [email protected] Crown Copyright © 2013 Published by Elsevier Inc on behalf of Australian and New Zealand Society of Cardiac and Thoracic Surgeons (ANZSCTS) and the Cardiac Society of Australia and New Zealand (CSANZ). All rights reserved.
133
Echocardiographic Predictors of LV Recovery Post AVR
Introduction Surgical treatment lowers the mortality rate for aortic valve disease [1]. Left ventricular (LV) dilatation normalises in the majority of patients with aortic regurgitation (AR) post aortic valve replacement (AVR) within the first few months post procedure, which is associated with an improvement in LV function [2]. Failure to normalise portends a worse long-term clinical outcome, presumably related to pre-operative irreversible myocardial damage [3]. Accordingly, incomplete regression of left ventricular hypertrophy (LVH) post AVR for aortic stenosis (AS) is associated with decreased survival [4]. Factors influencing normalisation of these parameters include reduced LV systolic function and larger pre-operative ventricular size. Significant (moderate-severe) postoperative AR is clearly associated with worse clinical outcomes [5,6], however recent investigations into paravalvular leaks associated with transcatheter aortic valve replacement have suggested that any degree of post procedure aortic regurgitation can adversely affect LV recovery and clinical outcomes [7,8]. Little published data exists with long term regular, prospective follow up post AVR incorporating echocardiographic evaluation of the impact on the left ventricle. Furthermore, few studies have assessed the differential effect of valve replacement according to pre-operative condition [9,10]. The aim of this study was to assess the effects of aortic valve replacement with the Ross procedure on LV size and function in a prospective fashion with serial echocardiography. We also aimed to define predictors of LV recovery post aortic valve replacement, in particular the impact of preoperative valve lesion (aortic stenosis (AS), mixed disease (AS/AR) or aortic regurgitation (AR)), pre-operative LV diameter and early post-operative mild aortic regurgitation.
Methods All patients with aortic valve disease who underwent aortic valve replacement with the Ross procedure at the Royal Melbourne Hospital, Melbourne Private Hospital or Epworth Hospital (Melbourne, Australia) between October 1992 and April 2009 were enrolled in the study. All procedures were performed by a single surgeon. The study group comprised 266 consecutive patients with aortic valve disease who were separated into three subgroups according to their pre-operative condition: AS (aortic stenosis only), Mixed (AS/AR i.e. mixed disease with grade 2 aortic regurgitation) or AR (grade 3 or 4 aortic regurgitation). The surgical procedure was a standard Ross procedure, with a root inclusion method in 98% of patients. All were performed with extracorporeal circulation and with moderate hypothermia. Follow up protocol, patients underwent a clinical review with echocardiography early post operation then serially thereafter (usually two-yearly).
Echocardiographic Methods Analysis was based on serial echocardiographic reports. Standard images were acquired as part of the Royal Melbourne Hospital echocardiography protocol in accordance with the recommendations of the American Society of Echocardiography (ASE). All echocardiographic examinations were performed with the patient at rest in the left lateral decubitus position breathing quietly. The examinations were performed using a commercially available ultrasound system (GE Medical Systems Vivid 7 Dimension (Horten, Norway) from 2005 and HP Philips Sonos 5500 prior to 2005 (Palo Alto, CA, USA). Two-dimensional images and standard measurements were acquired with MS4 phased array sector probe (frequency 1.5– 4 MHz). Spectral Doppler traces were acquired with a steerable piedoff probe (frequency 1.9 MHz). Images were acquired and stored digitally. LV chamber measurements were made in the parasternal long axis view using M mode, whilst left ventricular end systolic diameter (LVESD) and diastolic diameter (LVEDD) were indexed for body surface area. The definition for a normal indexed LVEDD was 3.2 cm/m2 in females and 3.1 cm/m2 in males [11]. M mode of the aortic root and valve was made in the anterior posterior plane at the end of diastole (onset of QRS complex on the ECG) and three beats were averaged. Two-dimensional assessment of the aortic valve was made in the long axis parasternal view, apical long axis view and apical fivechamber view. Two-dimensional images were optimised with the appropriate selection of frequency, frame rate and sector size. Image magnification to improve spatial resolution was achieved with the write zoom function. Semi-quantification of the presence of any regurgitant signal was assessed by colour Doppler. Frame rate, colour gain and colour maps were optimised. Colour across the aortic valve was assessed in the parasternal long axis view. If present, the regurgitant jet width to LVOT width ratio was determined as was the vena contracta and jet area. Colour Doppler was also used in the apical long axis view and five-chamber for the jet spray and jet direction. Spectral Doppler determined the velocities prior to and at the aortic valve with three measurements being averaged. LVOT flow was assessed by pulsed wave Doppler while CW Doppler was used to determine the forward flow signal and regurgitant signal. Sweep speed was 100 ms/s. Doppler gain and dynamic range was optimised to display the full and complete velocity spectrum.
Statistical Analysis The data was analysed using Stata 10 (StataCorp, TX, USA) Continuous data has been expressed as mean SD and was analysed using either paired t-test or ANOVA where appropriate. The categorical data was presented as N and % and was analysed using the Chi-squared test. Mortality rates were defined through linkage of this database to the National Death Index database.
134
W.M. Wilson et al.
A regression analysis was performed (using Multivariate Generalised Linear Mixed Model) to determine independent predictors of both LV size and change in LV size post surgery. Baseline variables included age, sex, pre-operative valve lesion, pre-operative fractional shortening, pre-operative indexed systolic and diastolic ventricular dimensions and degree of aortic regurgitation at 0.5–1 year post operation. The model fit was assessed using AIC (Akaike information criterion) and BIC (Bayesian information criterion). In addition, cross validation of the outcome predicted by the model and actual outcome was assessed using the correlation coefficient. For LVEDD, the correlation coefficient was 54% and for LVESD it was 59%. Level of significance of 0.05 applied for all tests.
Results The study group comprised 265 patients (one patient died three weeks post-operation) who were followed through a period of 1–14 years (mean 6.4 years, total 1702 patientyears). The mean age was 38.8 12.6 years and 69.8% were male. There were 118 patients (44.5%) with aortic stenosis only, 62 (23.4%) with mixed disease and 85 (32.1%) with predominant aortic regurgitation. There were some significant differences between the groups in terms of age, gender and echocardiographic results (Table 1). Patients in the AS group were older with a higher proportion of females. No patient had co-morbid ischaemic heart disease and only a small percentage (<5%) in each group had hypertension, which reflects the relatively young age of this study population. As expected, the AS patients had smaller ventricular dimensions with a smaller LVEDDi (indexed left ventricular end diastolic diameter) and LVESDi (indexed left ventricular end systolic diameter) compared to the other groups. There was no difference in LV posterior wall thickness and LV fractional shortening (FS) between the
AS and mixed groups but both parameters were significantly lower in the AR group. Male patients had a higher baseline LVESDi (1.88 0.44 cm/m2 vs 1.74 0.39 cm/m2, p = 0.016) and LV posterior wall thickness (1.17 0.17 cm vs 1.08 0.20 cm, p < 0.001). There also existed a significant correlation between age and baseline LVEDDi (r = 0.308, p < 0.001), LVESDi (r = 0.160, p = 0.011) and LV posterior wall thickness (r = 0.199, p = 0.002), however, there was no correlation between FS and age, nor FS and gender. There was also no correlation between baseline LVEDDi and gender.
Mortality and Morbidity One patient died within 30 days of operation (due to myocardial infarction). There have been four late deaths, all of which had a non-cardiac cause (all cancer-related deaths). Linkage of the database to the National Australian Death Index has corroborated this long-term survival rate of 98%. Two patients developed heart block requiring permanent pacemaker implantation in the early post-operative period. The need for permanent pacemaker insertion rate remains less than 1%.
Functional Improvement All patients, except one (mixed group), were NYHA 1 at one year and remained so throughout follow up. Transvalvular gradients remained low throughout followup (see Fig. 1)
LV Dimensions and Fractional Shortening After aortic valve replacement, the indexed diastolic left ventricular dimension decreased significantly in the Mixed and AR groups when measured at four to five years (see Tables 2 and 3). Reduction in LVEDDi occurred predominantly in the first year post-operatively in these patients and remained stable throughout follow up (Figs. 2 and 3). Those
Table 1 Baseline clinical and echocardiographic data. AS (N = 118) (Mean W SD)
Mixed (N = 62) (Mean W SD)
AR (N = 85) (Mean W SD)
p value
AGE (years)
43.7 (12.2)
34.8 (13.4)
34.8 (10.1)
<0.001
Female, N (%)
49 (41.5%)
19 (30.7%)
12 (14.1%)
<0.001
NYHA class
2.2 (0.7)
2.5 (0.7)
1.5 (0.7)
<0.001
LVEDD (cm)
4.84 ( 0.56)
5.59 (0.69)
6.52 (0.54)
<0.001
LVEDDi (cm/m2)
2.58 (0.34)
2.92 (0.40)
3.32 (0.37)
<0.001
LVESD (cm)
2.98 (0.60)
3.46 (0.62)
4.33 (0.60)
<0.001
LVESDi (cm/m2)
1.59 (0.33)
1.80 (0.33)
2.20 (0.35)
<0.001
LV posterior wall thickness (cm) LV fractional shortening (%)
1.19 (0.19) 38.5 (8.3)
1.19 (0.17) 38.7 (7.7)
1.06 (0.15) 33.8 (6.9)
<0.001 <0.001
Aortic sinus (indexed) (cm/m2)
17.2 ( 2.3)
17.4 ( 2.5)
19.0 (2.5)
<0.001
Transvalvular aortic gradient
38.5 (8.1)
37 (4.2)
NS
135
Echocardiographic Predictors of LV Recovery Post AVR
Neo- Aortic Root The (indexed) size of the neo-aortic root was stable throughout follow-up with no significant differences observed between the three subgroups over time. There appears to be an increase in the root size in the mixed and AR groups at 12 and 13 years follow-up, however the error margin is large at this timepoint owing to small patient numbers (see Fig. 6).
Aortic Regurgitation Post Operation
Fig. 1 Mean AV gradient post AVR (error bars represent 95% CI for mean of total cohort).
with pre-operative AR were more likely to have a baseline enlarged LVEDDi and appeared to have persistently larger mean LVEDDi measurements post-operatively compared to the AS group, however this difference was not statistically significant after four to five year follow-up (Fig. 2). Patients with enlarged pre-operative LVEDDi tended to have a significantly larger LVEDDi post-operatively compared to those with a normal pre-operative LVEDDi (see Table 4, Fig. 4) LVESDi decreased significantly in the AR group only (see Table 2). Mean LV systolic function, as measured by fractional shortening was normal pre-operatively and did not increase in any group; in fact it was significantly lower in the AS and mixed groups at five years compared to baseline (however, remained within the normal range) (see Table 2). Wall thickness (posterior wall) reduced significantly in the AS and Mixed groups (Table 2) and did so rapidly within one year (Fig. 5).
The degree of aortic regurgitation remained low overall post procedure (Fig. 7). Approximately 80% of patients had nil or trivial AR during the entire follow-up period. Five patients developed moderate AR; all five underwent redo procedures with a mechanical AVR and remain well, however, one patient developed endocarditis, which necessitated a third operation (with insertion of a bioprosthetic AVR). The mechanism for AR in these patients who developed moderate AR was not clear but all had AR as their initial lesion. The presence of mild AR at time period 0.5–1 year appeared to significantly impact upon the size of the left ventricle over time. LVEDDi measurements according to degree of AR as measured at 0.5–1 year post operation are shown in Fig. 8. At four to five year follow-up, the mean LVEDDi in patients with mild AR was significantly larger than in those with niltrivial AR post-operatively (2.72 (0.33) cm/m2 versus 2.58 (0.32) cm/m2, p = 0.035). Possible confounders to consider for this finding include: a different proportion of pre-operative lesion within each group ("Nil trivial group’’ comprised: AS 48%, Mixed 22%, AR 30% versus "Mild AR’’: AS 31%, Mixed 31%, AR 38%), however a trend does exist within each subgroup (see Table 5); or, alternatively, an unrecognised difference in the baseline LVEDDi between the two groups (but no significant difference existed between the mean baseline LVEDDi: Nil-Trivial AR baseline LVEDDi 2.89 cm/m2 (95% CI for mean 2.82–2.96)) versus Mild AR baseline LVEDD 2.97 cm/cm2 (95% CI 2.84–3.11); p = 0.309. Despite having a slightly larger LVEDDi at four to five year followup, it does not appear that those with mild AR had progressive LV dilatation over time. The size of the neo-aortic root during follow-up was not associated with the degree of AR observed during follow up.
Table 2 Echocardiographic data before and 5 years post aortic valve replacement in patients with aortic stenosis (AS), Mixed disease (AS + AR) and aortic regurgitation (AR) who completed 5 years of follow-up. Echo parameter
LVEDDi (cm/m2) LVESDi (cm/m2) PW thickness (cm) FS (%)
AS
Mixed
AR
Pre op
Post op
p value
Pre op
2.55 (0.35) 1.55 (0.36)
2.55 (0.34) 1.63 (0.39)
0.94 0.11
2.85 (0.38) 1.77 (0.32)
2.65 (0.3) <0.001 1.74 (0.30) 0.63
1.19 (0.21)
0.97 (0.19) <0.001
1.22 (0.18)
0.95 (0.13) <0.001
40.84 (8.13) 37.59 (7.01)
0.016
Post op
38.54 (7.31) 34.89 (7.83)
p value
0.038
Pre op
Post op
p value
3.34 (0.39) 2.26 (0.34)
2.67 (0.32) <0.001 1.85 (0.27) <0.001
1.03 (0.13)
0.99 (0.12)
0.178
32.98 (5.99) 32.06 (6.13)
0.351
136
W.M. Wilson et al.
Table 3 Normalisation of LVEDD and LVH post operation (percentage of patients with normal LVEDDi (defined as 3.2 cm/m2 in females and 3.1 cm/m2 in males) and LV wall thickness 11 mm pre and 5 years post operation). AS
2
LVEDDi <3.2 cm/m LVESDi <2.0 cm/m2 LVH < = 11 mm
AS/AR
Pre
5y
p
Pre
91.4% (107/117) 88% (97/110)
97.1% (69/71) 86% (49/57)
0.14 0.81
68.3% (41/60) 66.6% (40/60)
<0.001
38.2% (21/55)
40.0% (46/115)
85.9% (55/64)
5y 95% (38/40) 78% (28/36)
AR
97.1% (33/34)
p 0.001 0.35 <0.001
ALL
Pre
5y
p
Pre
5y
p
LVEDDi <3.2 cm/m2
35.3.0% (30/85)
96.0% (48/50)
<0.001
69.4% (182/262)
96.2% (155/161)
<0.001
LVESDi <2.0 cm/m2
28.5% (24/84)
72.9 (35/48)
<0.001
63.3% (161/254)
79.4% (112/141)
0.001
LVH < = 11 mm
73.8% (59/80)
97.8% (45/46)
<0.001
50.4% (126/250)
92.4% (133/144)
<0.001
Fig. 2 LVEDDi over time according to pre-operative condition (error bars
95% CI for mean).
Fig. 3 Change in LVEDDi (cm/m2) over the time (according to pre-operative condition).
137
Echocardiographic Predictors of LV Recovery Post AVR
Table 4 LVEDDi pre operation and at 5 years post operation according to pre-operative LV size and stratified according to pre-operative condition and (Normal indexed LVEDD is defined as 3.2 cm/m2 in females and 3.1 cm/m2 in males). Normal LVEDDi N ALL
Enlarged LVEDDi
Pre
5 years
p value
N
Pre
5 years
3.51 (0.30)
2.79 (0.33)
3.5
2.55
3.39 (0.17) 3.54 (0.33)
2.93 (0.33) 2.77 (0.32)
115
2.63 (0.32)
2.54 (0.30)
0.003
43
AS
68
2.52 ( 0.31)
2.55 (0.34)
0.364
2
Mixed AR
29 18
2.69 (0.25) 2.98 (0.15)
2.56 (0.23) 2.48 (0.21)
0.014 <0.001
9 32
p value <0.001 NA 0.003 <0.001
Fig. 4 LVEDD post procedure according to pre-operative LVEDD (normal is 3.2 cm/m2 in females and 3.1 cm/m2 in males) (error bars 95% CI for mean).
Fig. 5 LV wall thickness post procedure according to pre-operative condition (error bars for total cohort only) (error bars 95% CI for mean).
138
W.M. Wilson et al.
Fig. 6 Neo-aortic size post Ross procedure (error bars represent 95% CI for mean and present for total cohort only).
Multivariate Analysis At time period four to five years post operation, independent predictors of a larger LVEDDi included: female gender (LVEDDi 4.9% larger than males (95% CI 1.0–8.8, p = 0.014)); baseline LVEDDi >3.0 cm/m2 (LVEDDi 10.2% larger than if baseline LVEDDi 3.0 cm/m2 (95% CI 5.9–14.7, p < 0.0001)); and degree of post-operative AR (LVEDDi 5.3% larger if mild AR at 0.5–1 year compared to nil-trivial AR at same timepoint (95% CI 1.1–9.6%, p = 0.014)). Pre-operative aortic regurgitation and baseline LVESDi >2.0 cm/m2 were not predictors of a larger LVEDDi at four to five years. Higher pre-operative
fractional shortening was a predictor of a smaller LVEDDi at follow-up (an increase in 1 point FS decreased the LVEDDi by 0.3% (95% CI 0.01–0.6, p = 0.012). Similarly, at five years follow-up, independent predictors of a larger LVESDi included: female gender (LVESDi 7.9% larger than males (95% CI 1.1–14.6 p = 0.023)), and baseline LVEDDi >3.0 cm/m2 (LVESDi 10.2% larger than if baseline LVEDDi 3.0 cm/m2 (95% CI 2.7–17.7, p = 0.008)). Neither pre-operative valve lesion, pre-operative LVESDi >2.0 cm/ m2 nor post-operative mild AR were predictors of LVESDi at four to five years. Higher pre-operative fractional shortening
Fig. 7 Degree of AR post operation over time (expressed as a proportion).
139
Echocardiographic Predictors of LV Recovery Post AVR
Fig. 8 LVEDDi according to degree of AR at 0.5–1 year post operation (error bars–95% CI for mean).
was a predictor of smaller LVESDi (LVESDi 0.8% less for each 1 point increase in FS (95% CI 0.4–1.2, p < 0.0001)). Furthermore, the multivariate regression model shows that at one year post-operation, the degree of reduction in LVEDDi from baseline (change in LVEDDi) depends on: gender (male patients exhibited 8.3% less reduction in LVEDDi compared to female patients (95% CI: 0.6–16.0%, p = 0.033)); degree of postoperative AR (patients with mild AR exhibited 11.3% less reduction in LVEDDi compared to those with nil-trivial AR (95% CI: 3.0–19.6%, p = 0.008)) and LVEDDi size at baseline (there is a 70.6% greater reduction in LVEDDi per 1 cm increase in the baseline LVEDDi (95% CI: 50.3–90.8, p < 0.001)). At five years post operation there was less reduction in LVEDDi from baseline if there existed: mild post-operative AR (those with mild post-operative AR had 14.2% less reduction compared to those with nil-trivial AR (95% CI 4.0%– 24.3%, p = 0.006)). There was a greater reduction in LVEDDi according to: baseline LVEDDi (there was a 83.4% greater reduction in LVEDDi per 1 cm increase in the baseline
Table 5 Mean (SD) LVEDDi (cm/m2) according to post-operative AR degree, stratified according to preoperative condition. p value
5 years
Nil-Trivial
Mild
ALL
2.58 (0.32)
2.72 (0.33)
0.035
AS
2.53 (0.33)
2.72 (0.37)
0.162
2.62 ( 0.27)
2.69 (0.36)
0.497
2.63 (0.34)
2.74 ( 0.31)
0.311
AS/AR AR
LVEDDi (95% CI: 55.7–111.2%, p < 0.001)) and the pre-operative valve lesion (patients in the AR group had a 36.0% greater reduction compared to those in the AS group when adjusted for all other variables (95% CI 21.9–50.2%, p < 0.001). The reduction in LVESDi from baseline to five years level was dependent on baseline LVESDi only with a greater reduction according to higher baseline LVESDi (each 1 cm increase in baseline LVESDi is associated with a 86.1% greater reduction in LVESDi at five years (95% CI 0.35.2– 1.37, p < 0.001)) when adjusted for all other factors. The only predictor for a LVEDDi 3.0 cm/m2 at four to five years was LVEDDi at baseline being greater than 3.0 cm/ m2 (OR 2.4 (95% CI 1.4–4.1, p = 0.001). The only predictor for a LVESDi 2.0 cm/m2 at four to five years was LVESDi at baseline being greater than 2.0 cm/m2 (OR 1.8 95% CI 1.03–3.2, p = 0.041). Age was not a predictor of LV size post operation.
Discussion Surgical treatment lowers mortality rate for aortic valve disease, and in our study the five years mortality rate was <1% [1]. The main benefit of AVR appears to be via haemodynamic effects with removal of after-load in AS and removal of pre-load in AR. Pre-operative LV size and function are clear predictors of long-term survival post aortic valve replacement [12–14] and early improvement in ejection fraction post surgery has been demonstrated to be an independent predictor of survival [15], as has a post-operative LV diastolic diameter [16] and post-operative reduction in end-diastolic dimension of more than 20% [12].
140
Left ventricular dilatation has previously been shown to normalise in most patients with AR post AVR within the first few months [16,17]. This was reflected in our study as a reduction in LV size to normal was evident in the majority of cases by one year, and this was persistent throughout follow-up. Interestingly, short-term and long-term improvement in left ventricular systolic function post operation appears related to the early reduction in left ventricular dilatation arising from correction of left ventricular volume overload and, furthermore, late improvements in ejection fraction seem to occur only in those with early improvements in LV function [2]. Poor functional LV recovery has also been shown to be predicted by pre-operative LV spherical shape (ratio of minor axis to major axis of LV) [18] and systolic radial strain rate [19]. It is recognised that a certain proportion of patients do not derive benefit from surgery in terms of reduction in LV size and improvement in function (presumably relating to preoperative irreversible myocardial damage) and these patients have worse long term clinical outcomes [20]. Indeed, post-operative LV diastolic diameter dimension has been shown to be a strong predictor of late death from congestive heart failure after AVR for aortic regurgitation [16]. A number of factors in our study were associated with a significantly larger LVEDDi post procedure, in particular: a larger pre-operative ventricular size; the degree of post-operative AR, and female gender.
Predictors of LV Recovery Indexed values for ventricular size appear to be better predictors of LV recovery and clinical outcomes post AVR than non-indexed values, especially in those with low body weight [6,21–23]. The degree of enlargement and normal cut-offs for indexed LV dimensions are not clearly defined but the most commonly quoted measurements are indexed LVESD values less than 2.5 cm/m2 and indexed LVEDD values of less than 3.5 cm/m2. An indexed baseline LVESD value of >2.5 cm/m2 was a marker of poor outcome in patients with moderate to severe AR managed conservatively in a study by Dujardin et al. [24]. In another study of 171 patients undergoing AVR for AR, the sensitivity and specificity in predicting normalisation of LV function were 88% and 92% for indexed LVESD values <3.53 cm/m2 and 71% and 86% for indexed LVEDD values <4.44 cm/m2, whilst in a study of 66 patients undergoing AVR for severe AS with LV dysfunction, a cut off value of pre-operative indexed LVESD 2.75 cm/m2 was 85% sensitive and 72% specific in predicting intermediate-term recovery of LV function after AVR [6]. Morgan et al., however, found that the best discrimination for survival post AVR was seen at lower indexed values (2.0 cm/m2 for indexed LVESD and 3.0 cm/m2 for indexed LVEDD) [25]. Other studies have suggested that baseline LV end-systolic indices are more predictive of poor LV recovery than LV end-diastolic indices [17,26]). In our study, a baseline LVEDDi >3.0 cm/m2 was an independent predictor of a larger LVEDDi and LVESDi at five years follow-up.
W.M. Wilson et al.
The presence of mild AR at follow-up echocardiographic assessment in the first post-operative year in this study appeared to have an impact on the size of the left ventricle during longer term follow-up; it was an independent predictor of a larger LVEDDi at five years, and an independent predictor of a lower degree of reduction of LVEDDi at one and five years. It is important to note, however, that the presence of mild AR did not seem to cause progressive LV dilatation over long-term follow-up. Moderate or severe residual AR post AVR generally mandates redo surgery given it is associated with poorer clinical outcomes. In a study by Sponga et al. of 3201 patients who had undergone aortic valve replacement in a single institution, the incidence of residual AR >1/4 (i.e. more than mild) was 4.2%, which was associated with an increased risk of need for reoperation as well as an increased risk of mortality at all time points during follow-up [27]. Mild residual AR, however, is often expected to improve over time and appears benign, although few studies have addressed this specifically; in a small study of 40 patients identified as having a small periprosthetic leak by Rallidis et al. [28], no deaths were recorded and echocardiographic parameters remained unchanged during five year follow-up, and no increase in mortality at six weeks post AVR was found by O’Rourke et al. in a study of 113 patients with trivial/mild AR post surgery [29]. The advent of transcatheter aortic valve replacement has led to a renewed interest in the impact of residual aortic regurgitation post valve replacement given the relatively high frequency of paravalvular leak post TAVI compared to surgical AVR [30]. Paravalvular leak >1/4 has been shown to be an independent predictor of mortality post TAVI in a number of trials in keeping with the data post surgical AVR [31,32], however, interestingly, two year outcome data from the PARTNER trial [30] suggest that the presence of mild, moderate or severe paravalvular or total aortic regurgitation (vs. none or trace) after TAVI is associated with increased late mortality. The effect of aortic regurgitation on mortality was certainly proportional to the severity of the regurgitation, however even mild aortic regurgitation was associated with an increased rate of late deaths in the PARTNER trial. It should be noted that this association was not analysed in a multivariate fashion and this finding was not corroborated by the core valve Advance trial of 1015 patients who underwent transcatheter valve insertion where six month mortality in those with mild post-procedure AR was the same as those with no AR [33]. Furthermore, in a multicentre Canadian study of 339 patients, the presence of mild AR did not impact upon LV diameter nor LV ejection fraction over time and no differences were observed between those patients with mild AR and those with nil-trace AR [34]. A potential limitation of drawing conclusions from TAVI trials regarding the impact of mild AR is that assessment of paravalvular leak severity is more difficult and less reproducible than that of valvar regurgitation severity. It has previously been reported that women have a poorer outcome after aortic valve replacement compared to men [35,36]. In a report by Klodas et al., female gender was an
141
Echocardiographic Predictors of LV Recovery Post AVR
independent predictor of lower late survival post AVR for AR; women in this study were more likely to have surgery on the basis of symptoms rather than LV dimensions which served to highlight the fact that generalisation of surgical criteria based on unadjusted LV diameters established in men may not be appropriate in women. The authors concluded that surgical correction of AR should be considered at an earlier stage in women [35]. Conversely, female sex was an independent predictor of greater incremental improvement in ejection fraction post AR in a study of 1012 patients post AVR by Morris et al. [15]. In our study, female gender was an independent predictor of larger indexed LVEDDi and LVESDi post-operatively. The impact of gender on LV mass regression post surgery for AS is unclear with multiple studies yielding conflicting results [37].
Differential Effects of Valve Replacement According to Pre-Operative Condition Few studies have assessed the differential effects of valve replacement according to pre-operative condition. Waszyrowski et al. [10] followed 143 patients post AVR for a mean of five years and analysed them according to preoperative valve lesion (AS, AR or mixed valvular disease). They noted LVEDD decreased significantly in all groups, however LVESD decreased only in the AR and mixed groups whilst LVEF increased in the mixed group only. As expected, the LV wall thickness reduced in the AS and mixed groups only. LVEDDi decreased significantly in the AR and mixed groups in our study such that no significant difference in LVEDDi existed between the AS/mixed/AR groups at five years. More recently, Lamb et al. [9] focussed on LV mass index and diastolic function post AVR, which is of interest given impaired diastolic function has implications for morbidity and mortality post AVR [38–40]. This MRI based assessment suggested that early after AVR, patients with aortic valve stenosis show a decrease in both LVMi (LV mass index) and LVMi/LVEDVi ratio (LV end diastolic volume index) with an improvement in diastolic filling. This was in clear contrast to patients with aortic regurgitation where LVMi decreases less rapidly than LVEDVi, causing concentric remodelling of the LV and deterioration in diastolic filling in these patients. In a similar echocardiographic based study by Iyem et al. [41], the timecourse for LV remodelling and improvement in EF was different depending on the pre-operative lesion; in the AS group, LV remodelling was a continuing process with significant improvements occurring in the second year of follow-up, whilst LV remodelling and improvement in EF did not change significantly in the AR group between six months and two years post-operatively. In our cohort, LVEDDi normalised by 0.5–1 year follow-up (if dilated pre-operatively) and remained stable thereafter. A disproportionately high relative wall thickness before AVR is associated with excessive perioperative risk. Reverse LV remodelling after aortic valve replacement with regression of myocardial hypertrophy occurs mostly within the
first few years [42] and has been shown to result in improvement of LV longitudinal myocardial strain and decrease of Ntpro-BNP plasma levels [43], whilst failure of regression of LVH post AVR for AS is associated with worse longterm outcomes [4,44,45]. Mean LV wall thickness reduced significantly post Ross procedure in the majority (90%) of patients in our study and did so within the first two to three years. This impressive reduction in LVH is in contrast to the study by Waszyrowski et al. [10], where normalisation of left ventricular hypertrophy was observed in less than half of the study population. An increase in LV systolic function was not seen in our study, which likely reflects near normal pre-operative function relating to careful patient selection and proper surgery timing in a relatively young cohort. Nonetheless, it was evident on multivariate analysis that a higher baseline fractional shortening was a predictor of smaller ventricular dimensions (both LVEDDi and LVESDi) at five years.
Study Limitations Previous studies regarding the impact of AVR on left ventricular remodelling have been limited in particular by heterogeneity of the surgical procedure undertaken. A relative strength of our study is that each operation was performed by the same surgeon in one of three institutions, thus limiting confounding due to differences in surgical techniques. Inherent limitations exist in echocardiographic assessment of LV dimensions and wall thickness as well as in the evaluation of AR. Variability in the performance of the echocardiograms over more than a 10 year period is an unavoidable limitation, however, all the studies were predominantly performed at one institution where strict protocols for echocardiographic performance exist. Not all previous images were available, hence the data for this study have been obtained directly from echocardiogram reports and echocardiographic images were not independently assessed in a controlled and blinded fashion at a core lab, which represents another limitation of this study. It would be interesting to know whether those with persistent LV enlargement fare worse clinically; the vast majority of patients in the study were NYHA class 1 at long term follow-up so there appears to be no gross difference in symptomatology between patients. Objective assessments of functional capacity (through cardiopulmonary or stress testing for example) were not available for many patients to allow us to define subtle differences in exercise capacity between those with normal versus larger ventricles postoperation. We have confirmed the quoted mortality rate, however, through linkage with the Australian death index. The Ross procedure has a lower post-operative gradient than other forms of AVR and this represents a younger cohort than other studies, hence these findings may not necessarily be extrapolated to all patients undergoing AVR.
Conclusion A number of factors impact upon the degree of reduction in left ventricular size post AVR, in particular pre-operative LV
142
size. Female gender and mild AR early post AVR were also predictors of larger LV dimensions in this study. Patients with mild post-operative AR did not have progressive LV enlargement, however, suggesting this finding has limited clinical significance.
W.M. Wilson et al.
[19]
[20]
References [1] Scognamiglio R, Negut C, Palisi M, Fasoli G, Dalla Volta S. Long-term survival and functional results after aortic valve replacement in asymptomatic patients with chronic severe aortic regurgitation. J Am Coll Cardiol 2005;45:1025–30. [2] Bonow RO, Dodd JT, Maron BJ, O’Gara PT, White GG, McIntosh CL, et al. Long-term serial changes in left ventricular function and reversal of ventricular dilatation after valve replacement for aortic regurgitation. Circulation 1988;78:1108–20. [3] Carroll JD, Gaasch WH, Zile MR, Levine HJ. Serial changes in left ventricular function after correction of chronic aortic regurgitation: dependence on early changes in preload and subsequent regression of hypertrophy. Am J Cardiol 1983;51:476–82. [4] Bech-Hanssen O, Wallentin I, Houltz E, Suurkula MB, Larsson S, Caidahl K. Gender differences in patients with severe aortic stenosis: impact on preoperative left ventricular geometry and function, as well as early postoperative morbidity and mortality. Eur J Cardiothorac Surg 1999;15:24–30. [5] Jindani A, Neville EM, Venn G, Williams BT. Paraprosthetic leak: a complication of cardiac valve replacement. J Cardiovasc Surg 1991;32: 503–8. [6] Cho SH, Byun CS, Kim KW, Chang BC, Yoo KJ, Lee S. Preoperative indexed left ventricular dimensions to predict early recovery of left ventricular function after aortic valve replacement for chronic aortic regurgitation. Circ J 2010 Nov;74:2340–5. [7] Lerakis S, Hayek S, Douglas P. Paravalvular aortic leak after transcatheter aortic valve replacement current knowledge. Circulation 2013;127:397–407. [8] Kumpuris AG, Quinones MA, Waggoner AD, Kanon DJ, Nelson JG, Miller RR. Importance of preoperative hypertrophy, wall stress and endsystolic dimension as echocardiographic predictors of normalization of left ventricular dilatation after valve replacement in chronic aortic insufficiency. Am J Cardiol 1982;49:1091–100. [9] Lamb H, Beyerbacht HP, de Roos A, van der Laarse A, Vliegen HW, Leujes F, et al. Left ventricular remodeling early after aortic valve replacement: differential effects on diastolic function in aortic valve stenosis and aortic regurgitation. J Am Coll Card 2002;40(12):2182–8. [10] Waszyrowski T, Kasprzak JD, Krzemihka-Pakula M, Droz J, Dziatkowiak A, Zaslonka J. Regression of left ventricular dilatation and hypertrophy after aortic valve replacement. Int J Cardiol 1996;57:217–25. [11] Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, et al. Recommendations for chamber quantification. Eur J Echocardiogr 2006;7:79–108. [12] Corti R, Binggeli C, Turina M, Jenni R, Luscher TF, Turina J. Predictors of long-term survival after valve replacement for chronic aortic regurgitation is Mmode echocardiography sufficient? Eur Heart J 2001;22:866. [13] Tornos P, Sambola A, Permayer-Miralda G, Evangelista A, Gomez Z, Soler-Soler J. Long-term outcome of surgically treated aortic regurgitation: influence of guideline adherence toward early surgery. J Am Coll Cardiol 2006;47:1012. [14] Turina J, Milincic J, Seifert B, Turina M. Valve replacement in chronic aortic regurgitation: true predictors of survival after extended follow-up. Circulation 1998;98:100II–6II. [15] Morris JJ, Schaff HV, Mullany CJ, Rastogi A, McGregor CGA, Daly RC, et al. Determinants of survival and recovery of left ventricular function after aortic valve replacement. Ann Thor Surg 1993;56((1) July):22–9. [16] Henry WL, Bonow RO, Borer JS, Ware JH, Kent KM, Redwood DR, et al. Observations on the optimum time for operative intervention for aortic regurgitation. I. Evaluation of the results of aortic valve replacement in symptomatic patients. Circulation 1980;61:471–83. [17] Kumpuris AG, Quinones MA, Waggoner AD, Kanon DJ, Nelson JG, Miller RR. Importance of preoperative hypertrophy, wall stress and endsystolic dimension as echocardiographic predictors of normalization of left ventricular dilatation after valve replacement in chronic aortic insufficiency. Am J Cardiol 1982;49:1091–100 [8]. [18] Matsumura Y, Gillinov AM, Toyono M, Wada N, Yamano T, Thomas JD, et al. Usefulness of left ventricular shape to predict the early recovery
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33] [34]
[35]
[36]
[37]
[38]
[39]
of left ventricular function after isolated aortic valve replacement for aortic valve stenosis. Am J Cardiol 2008;11:1530–4. Onishi T, Kawai H, Tatsumi K, Kataoka T, Sugiyama D, Tanaka H, et al. Preoperative systolic strain rate predicts postoperative left ventricular dysfunction in patients with chronic aortic regurgitation. Circ Cardiovasc Imag 2010;3:134–41. Bonow RO, Picone AL, McIntosh CL, Jones M, Rosing DR, Maron BJ, et al. Survival and functional results after valve replacement for aortic regurgitation from 1976 to 1983: impact of preoperative left ventricular function. Circulation 1985;72:1244–56. Brown ML, Schaff HV, Suri RM, Zhuo L, Sundt TM, Dearani JA, et al. Indexed left ventricular dimensions best predict survival after aortic valve replacement in patients with aortic valve regurgitation. Ann Thor Surg 2009;87:1170–2117. Sambola A, Tornos P, Ferreira-Gonzalez I, Evangelista A. Prognostic value of preoperative indexed end-systolic left ventricle diameter in the outcome after surgery in patients with chronic aortic regurgitation. Am Heart J 2008;155:1114–20. Ding W, Lam Y, Kaya MG, Li W, Chung R, Pepper JR, et al. Echocardiographic predictors of left ventricular functional recovery following valve replacement surgery for severe aortic stenosis. Int Heart J 2008;128:178–84. Dujardin KS, Enriquez-Sarano M, Schaff HV, Bailey KR, Seward JB, Tajik AJ. Mortality and morbidity of aortic regurgitation in clinical practice: a long-term follow-up study. Circulation 1999;99:1851–7. Brown ML, Schaff HV, Suri RM, Zhuo L, Sundt TM, Dearani JA, et al. Indexed left ventricular dimensions best predict survival after aortic valve replacement in patients with aortic valve regurgitation. Ann Thorac Surg 2009;87:1170–6. Cox DA, Walton K, Bartz PJ, Tweddell JS, Frommelt PC, Earing MG. Predicting left ventricular recovery after replacement of a regurgitant aortic valve in pediatric and young adult patients: is it ever too late? Pediatr Cardiol 2012 [Epub ahead of print]. Sponga S, Perron J, Dagenais F, Mohammadi S, Baillot R, Doyle D, et al. Impact of residual regurgitation after aortic valve replacement. Eur J Cardiothor Surg 2012;42:486–92. Rallidis LS, Moyssakis IE, Ikonomidis I, Nihoyannopoulos P. Natural history of early aortic paraprosthetic regurgitation: a five-year follow-up. Am Heart J 1999;138:351–7. O’Rourke DJ, Palac RT, Malenka DJ, Marrin CA, Arbuckle BE, Plehn JF. Outcome of mild periprosthetic regurgitation detected by intraoperative transesophageal echocardiography. J Am Coll Cardiol 2001;38:163–6. Kodali SK, Williams MR, Smith CR, Svensson LG, Webb JG, Makkar RR, et al. Leon MB for the PARTNER trial investigators. Two year outcomes after transcatheter or surgical aortic valve replacement. N Engl J Med 2012;366:1686–95. Tamburino C, Capodanno D, Ramondo A, Pertronio AS, Ettori F, Santoro G, et al. Incidence and predictors of early and late mortality after transcatheter aortic valve implantation in 663 patients with severe aortic stenosis. Circulation 2011;125:299–308. Abdel-Wahab M, Zahn R, Sherif M, Schneider S, Gerckens U, Schuler G, et al. In-hospital outcome of aortic regurgitation after TAVI: results from the prospective multicentre German TAVI registry. Eur Heart J 2010;31:160. Linke A, on behalf of the ADVANCE Steering Committee and Investigators. The ADVANCE TAVR registry. In: Presented at TCT 2012; 2012. Rode´s-Cabau J, Webb JG, Cheung A, Ye J, Dumont E, Osten M, et al. Long-term outcomes after transcatheter aortic valve implantation insights on prognostic factors and valve durability from the canadian multicenter experience. J Am Coll Cardiol 2012;60:1864–75. Klodas E, Enriquez-Sarano M, Tajik AJ, Mullany CJ, Bailey KR, Seward JB. Surgery for aortic regurgitation in women. Contrasting indications and outcomes compared with men. Circulation 1996;94:2472. Flores-Marı´n A, Go´mez-Doblas JJ, Caballero-Borrego J, Cabrera-Bueno F, Rodrı´guez-Bailo´n I, Meleroa JM, et al. Long-term predictors of mortality and functional recovery after aortic valve replacement for severe aortic stenosis with left ventricular dysfunction. Rev Esp Cardiol 2010;63:36–45. Villa E, Troise G, Cirillo M, Brunelli F, Tomba MD, Mhagna Z, et al. Factors affecting left ventricular remodeling after valve replacement for aortic stenosis. An overview. Cardiovasc ultrasound 2006;4:25. Lund O, Flø C, Jensen FT, Emmertsen K, Nielsen TT, Rasmussen BS, et al. Left ventricular systolic and diastolic function in aortic stenosis. Eur Heart J 1997;18:1977–87. Lund O, Nielsen TT, Emmertsen K, Flø C, Rasmussen BS, Jensen FT, et al. Mortality and worsening of prognostic profile during waiting time
Echocardiographic Predictors of LV Recovery Post AVR
for valve replacement in aortic stenosis. Thorac Cardiovasc Surg 1996;44: 289–95. [40] Villari B, Vassalli G, Monrad ES, Chiariello M, Turina M, Hess OM. Normalization of diastolic dysfunction in aortic stenosis late after valve replacement. Circulation 1995;91:2353–8. [41] Iyem H, Sekuri C, Tavli M, Bu¨ket S. Left ventricular hypertrophy and remodeling after aortic valve replacement. Asian Cardiovasc Thorac Ann 2007;15:459–62. [42] Lim E, Ali A, Theodorou P, Sousa I, Ashrafian H, Chamageorgakis T, et al. Longitudinal study of the profile and predictors of left ventricular mass regression after stentless aortic valve replacement. Ann Thorac Surg 2008;85:2026–9.
143
[43] Poulsen SH, Søgaard P, Nielsen-Kudsk JE, Egeblad H. Recovery of left ventricular systolic longitudinal strain after valve replacement in aortic stenosis and relation to natriuretic peptides. J Am Soc Echocardiogr 2007 Jul;20:877–84. [44] Aurigemma GP, Silver KH, McLaughlin M, Mauser J, Gaasch WH. Impact of chamber geometry and gender on left ventricular systolic function in patients >60 years of age with aortic stenosis. Am J Cardiol 1994;74:794–8. [45] Gaasch WH, Carroll JD, Levine HJ, Criscitiello MG. Chronic aortic regurgitation: prognostic value of left ventricular end-systolic dimension and end-diastolic radius/thickness ratio. J Am Coll Cardiol 1983;1: 775–82.
ORIGINAL ARTICLE
Long Term Follow-Up After Aortic Valve Replacement (Ross Procedure): Echocardiographic Determinants of Ventricular Recovery William M. Wilson a*, Leeanne E. Grigg a, Alexandra Gorelik b, Peter Skillington c a
Department of Cardiology, Royal Melbourne Hospital, Parkville 3052, Australia Melbourne EpiCentre, Royal Melbourne Hospital, Parkville 3052, Australia c Department of Cardiothoracic Surgery, Royal Melbourne Hospital, Parkville 3052, Australia b
Received 10 November 2010; received in revised form 28 June 2013; accepted 10 July 2013; online published-ahead-of-print 17 August 2013
Aims and Methods
This prospective cohort study aimed to assess LV recovery post aortic valve replacement, stratified according to pre-operative valve lesion (aortic stenosis (AS), mixed disease (AS/AR) or aortic regurgitation (AR)), as well as define predictors of persistent LV dilatation post operation. We prospectively followed all patients post Ross procedure performed between 1992 and 2009 by a single surgeon. Echocardiography was performed pre-operatively, at approximately one year post operation then second yearly thereafter.
Results
265 patients were followed for a mean of 6.4 years (range 1–14 years, total 1702 patient-years). Seventy percent were male and mean age was 38.8 12.6 years. The indication for surgery was AS in 44.5% (118), AS/AR in 23.4% (62), and AR in 32.1% (85). Overall mortality was 1.8% and 80% of deaths were non-cardiac. Morbidity was low and the need for pacing was less than 1%. Ninety-nine percent of patients were NYHA class 1 at one year follow up. The indexed LV end diastolic diameter (LVEDDi) decreased significantly postoperation in the AR (3.34 0.39–2.66 0.32 cm/m2, p < 0.001) and mixed (2.85 0.38–2.65 0.30 cm/m2, p = 0.01) groups, whilst the indexed systolic LV dimension decreased significantly post-operation in the AR group (2.26 0.34–1.87 0.27, p < 0.001). At five years post operation, independent predictors for a larger LVEDDi were female gender, a pre-operative LVEDDi >3.0 cm/m2 and the presence of mild aortic regurgitation at one year post-operation. Pre-operative valve lesion was not a predictor. The only independent predictor of a lesser reduction of LVEDDi at five year follow-up was mild post-operative AR, whilst predictors of a lesser reduction in indexed left ventricular end systolic diameter (LVESDi) at five year follow-up included mild post-operative AR and a larger pre-operative LVEDDi. LV wall thickness decreased significantly the AS and AS/AR groups within one year post operation, whilst the neo-aortic root size remained stable throughout follow-up.
Conclusions
Recovery of LV size post Ross procedure is influenced predominantly by the pre-operative LV size, in particular the indexed LV end diastolic diameter. The pre-operative valve lesion was not predictive of larger ventricular dimensions post AVR, but independent predictors of a larger ventricular dimensions post operation included female gender, enlarged pre-operative LVEDDi and the presence of mild AR in the first post operative year. Those with mild post-operative AR did not have progressive LV enlargement, thus the clinical significance of this finding remains unclear.
Keywords
Aortic valve replacement Ross procedure Left ventricular size
* Corresponding author. Tel.: +1 6479977745; fax: +61 393423130., Email: [email protected] Crown Copyright © 2013 Published by Elsevier Inc on behalf of Australian and New Zealand Society of Cardiac and Thoracic Surgeons (ANZSCTS) and the Cardiac Society of Australia and New Zealand (CSANZ). All rights reserved.
133
Echocardiographic Predictors of LV Recovery Post AVR
Introduction Surgical treatment lowers the mortality rate for aortic valve disease [1]. Left ventricular (LV) dilatation normalises in the majority of patients with aortic regurgitation (AR) post aortic valve replacement (AVR) within the first few months post procedure, which is associated with an improvement in LV function [2]. Failure to normalise portends a worse long-term clinical outcome, presumably related to pre-operative irreversible myocardial damage [3]. Accordingly, incomplete regression of left ventricular hypertrophy (LVH) post AVR for aortic stenosis (AS) is associated with decreased survival [4]. Factors influencing normalisation of these parameters include reduced LV systolic function and larger pre-operative ventricular size. Significant (moderate-severe) postoperative AR is clearly associated with worse clinical outcomes [5,6], however recent investigations into paravalvular leaks associated with transcatheter aortic valve replacement have suggested that any degree of post procedure aortic regurgitation can adversely affect LV recovery and clinical outcomes [7,8]. Little published data exists with long term regular, prospective follow up post AVR incorporating echocardiographic evaluation of the impact on the left ventricle. Furthermore, few studies have assessed the differential effect of valve replacement according to pre-operative condition [9,10]. The aim of this study was to assess the effects of aortic valve replacement with the Ross procedure on LV size and function in a prospective fashion with serial echocardiography. We also aimed to define predictors of LV recovery post aortic valve replacement, in particular the impact of preoperative valve lesion (aortic stenosis (AS), mixed disease (AS/AR) or aortic regurgitation (AR)), pre-operative LV diameter and early post-operative mild aortic regurgitation.
Methods All patients with aortic valve disease who underwent aortic valve replacement with the Ross procedure at the Royal Melbourne Hospital, Melbourne Private Hospital or Epworth Hospital (Melbourne, Australia) between October 1992 and April 2009 were enrolled in the study. All procedures were performed by a single surgeon. The study group comprised 266 consecutive patients with aortic valve disease who were separated into three subgroups according to their pre-operative condition: AS (aortic stenosis only), Mixed (AS/AR i.e. mixed disease with grade 2 aortic regurgitation) or AR (grade 3 or 4 aortic regurgitation). The surgical procedure was a standard Ross procedure, with a root inclusion method in 98% of patients. All were performed with extracorporeal circulation and with moderate hypothermia. Follow up protocol, patients underwent a clinical review with echocardiography early post operation then serially thereafter (usually two-yearly).
Echocardiographic Methods Analysis was based on serial echocardiographic reports. Standard images were acquired as part of the Royal Melbourne Hospital echocardiography protocol in accordance with the recommendations of the American Society of Echocardiography (ASE). All echocardiographic examinations were performed with the patient at rest in the left lateral decubitus position breathing quietly. The examinations were performed using a commercially available ultrasound system (GE Medical Systems Vivid 7 Dimension (Horten, Norway) from 2005 and HP Philips Sonos 5500 prior to 2005 (Palo Alto, CA, USA). Two-dimensional images and standard measurements were acquired with MS4 phased array sector probe (frequency 1.5– 4 MHz). Spectral Doppler traces were acquired with a steerable piedoff probe (frequency 1.9 MHz). Images were acquired and stored digitally. LV chamber measurements were made in the parasternal long axis view using M mode, whilst left ventricular end systolic diameter (LVESD) and diastolic diameter (LVEDD) were indexed for body surface area. The definition for a normal indexed LVEDD was 3.2 cm/m2 in females and 3.1 cm/m2 in males [11]. M mode of the aortic root and valve was made in the anterior posterior plane at the end of diastole (onset of QRS complex on the ECG) and three beats were averaged. Two-dimensional assessment of the aortic valve was made in the long axis parasternal view, apical long axis view and apical fivechamber view. Two-dimensional images were optimised with the appropriate selection of frequency, frame rate and sector size. Image magnification to improve spatial resolution was achieved with the write zoom function. Semi-quantification of the presence of any regurgitant signal was assessed by colour Doppler. Frame rate, colour gain and colour maps were optimised. Colour across the aortic valve was assessed in the parasternal long axis view. If present, the regurgitant jet width to LVOT width ratio was determined as was the vena contracta and jet area. Colour Doppler was also used in the apical long axis view and five-chamber for the jet spray and jet direction. Spectral Doppler determined the velocities prior to and at the aortic valve with three measurements being averaged. LVOT flow was assessed by pulsed wave Doppler while CW Doppler was used to determine the forward flow signal and regurgitant signal. Sweep speed was 100 ms/s. Doppler gain and dynamic range was optimised to display the full and complete velocity spectrum.
Statistical Analysis The data was analysed using Stata 10 (StataCorp, TX, USA) Continuous data has been expressed as mean SD and was analysed using either paired t-test or ANOVA where appropriate. The categorical data was presented as N and % and was analysed using the Chi-squared test. Mortality rates were defined through linkage of this database to the National Death Index database.
134
W.M. Wilson et al.
A regression analysis was performed (using Multivariate Generalised Linear Mixed Model) to determine independent predictors of both LV size and change in LV size post surgery. Baseline variables included age, sex, pre-operative valve lesion, pre-operative fractional shortening, pre-operative indexed systolic and diastolic ventricular dimensions and degree of aortic regurgitation at 0.5–1 year post operation. The model fit was assessed using AIC (Akaike information criterion) and BIC (Bayesian information criterion). In addition, cross validation of the outcome predicted by the model and actual outcome was assessed using the correlation coefficient. For LVEDD, the correlation coefficient was 54% and for LVESD it was 59%. Level of significance of 0.05 applied for all tests.
Results The study group comprised 265 patients (one patient died three weeks post-operation) who were followed through a period of 1–14 years (mean 6.4 years, total 1702 patientyears). The mean age was 38.8 12.6 years and 69.8% were male. There were 118 patients (44.5%) with aortic stenosis only, 62 (23.4%) with mixed disease and 85 (32.1%) with predominant aortic regurgitation. There were some significant differences between the groups in terms of age, gender and echocardiographic results (Table 1). Patients in the AS group were older with a higher proportion of females. No patient had co-morbid ischaemic heart disease and only a small percentage (<5%) in each group had hypertension, which reflects the relatively young age of this study population. As expected, the AS patients had smaller ventricular dimensions with a smaller LVEDDi (indexed left ventricular end diastolic diameter) and LVESDi (indexed left ventricular end systolic diameter) compared to the other groups. There was no difference in LV posterior wall thickness and LV fractional shortening (FS) between the
AS and mixed groups but both parameters were significantly lower in the AR group. Male patients had a higher baseline LVESDi (1.88 0.44 cm/m2 vs 1.74 0.39 cm/m2, p = 0.016) and LV posterior wall thickness (1.17 0.17 cm vs 1.08 0.20 cm, p < 0.001). There also existed a significant correlation between age and baseline LVEDDi (r = 0.308, p < 0.001), LVESDi (r = 0.160, p = 0.011) and LV posterior wall thickness (r = 0.199, p = 0.002), however, there was no correlation between FS and age, nor FS and gender. There was also no correlation between baseline LVEDDi and gender.
Mortality and Morbidity One patient died within 30 days of operation (due to myocardial infarction). There have been four late deaths, all of which had a non-cardiac cause (all cancer-related deaths). Linkage of the database to the National Australian Death Index has corroborated this long-term survival rate of 98%. Two patients developed heart block requiring permanent pacemaker implantation in the early post-operative period. The need for permanent pacemaker insertion rate remains less than 1%.
Functional Improvement All patients, except one (mixed group), were NYHA 1 at one year and remained so throughout follow up. Transvalvular gradients remained low throughout followup (see Fig. 1)
LV Dimensions and Fractional Shortening After aortic valve replacement, the indexed diastolic left ventricular dimension decreased significantly in the Mixed and AR groups when measured at four to five years (see Tables 2 and 3). Reduction in LVEDDi occurred predominantly in the first year post-operatively in these patients and remained stable throughout follow up (Figs. 2 and 3). Those
Table 1 Baseline clinical and echocardiographic data. AS (N = 118) (Mean W SD)
Mixed (N = 62) (Mean W SD)
AR (N = 85) (Mean W SD)
p value
AGE (years)
43.7 (12.2)
34.8 (13.4)
34.8 (10.1)
<0.001
Female, N (%)
49 (41.5%)
19 (30.7%)
12 (14.1%)
<0.001
NYHA class
2.2 (0.7)
2.5 (0.7)
1.5 (0.7)
<0.001
LVEDD (cm)
4.84 ( 0.56)
5.59 (0.69)
6.52 (0.54)
<0.001
LVEDDi (cm/m2)
2.58 (0.34)
2.92 (0.40)
3.32 (0.37)
<0.001
LVESD (cm)
2.98 (0.60)
3.46 (0.62)
4.33 (0.60)
<0.001
LVESDi (cm/m2)
1.59 (0.33)
1.80 (0.33)
2.20 (0.35)
<0.001
LV posterior wall thickness (cm) LV fractional shortening (%)
1.19 (0.19) 38.5 (8.3)
1.19 (0.17) 38.7 (7.7)
1.06 (0.15) 33.8 (6.9)
<0.001 <0.001
Aortic sinus (indexed) (cm/m2)
17.2 ( 2.3)
17.4 ( 2.5)
19.0 (2.5)
<0.001
Transvalvular aortic gradient
38.5 (8.1)
37 (4.2)
NS
135
Echocardiographic Predictors of LV Recovery Post AVR
Neo- Aortic Root The (indexed) size of the neo-aortic root was stable throughout follow-up with no significant differences observed between the three subgroups over time. There appears to be an increase in the root size in the mixed and AR groups at 12 and 13 years follow-up, however the error margin is large at this timepoint owing to small patient numbers (see Fig. 6).
Aortic Regurgitation Post Operation
Fig. 1 Mean AV gradient post AVR (error bars represent 95% CI for mean of total cohort).
with pre-operative AR were more likely to have a baseline enlarged LVEDDi and appeared to have persistently larger mean LVEDDi measurements post-operatively compared to the AS group, however this difference was not statistically significant after four to five year follow-up (Fig. 2). Patients with enlarged pre-operative LVEDDi tended to have a significantly larger LVEDDi post-operatively compared to those with a normal pre-operative LVEDDi (see Table 4, Fig. 4) LVESDi decreased significantly in the AR group only (see Table 2). Mean LV systolic function, as measured by fractional shortening was normal pre-operatively and did not increase in any group; in fact it was significantly lower in the AS and mixed groups at five years compared to baseline (however, remained within the normal range) (see Table 2). Wall thickness (posterior wall) reduced significantly in the AS and Mixed groups (Table 2) and did so rapidly within one year (Fig. 5).
The degree of aortic regurgitation remained low overall post procedure (Fig. 7). Approximately 80% of patients had nil or trivial AR during the entire follow-up period. Five patients developed moderate AR; all five underwent redo procedures with a mechanical AVR and remain well, however, one patient developed endocarditis, which necessitated a third operation (with insertion of a bioprosthetic AVR). The mechanism for AR in these patients who developed moderate AR was not clear but all had AR as their initial lesion. The presence of mild AR at time period 0.5–1 year appeared to significantly impact upon the size of the left ventricle over time. LVEDDi measurements according to degree of AR as measured at 0.5–1 year post operation are shown in Fig. 8. At four to five year follow-up, the mean LVEDDi in patients with mild AR was significantly larger than in those with niltrivial AR post-operatively (2.72 (0.33) cm/m2 versus 2.58 (0.32) cm/m2, p = 0.035). Possible confounders to consider for this finding include: a different proportion of pre-operative lesion within each group ("Nil trivial group’’ comprised: AS 48%, Mixed 22%, AR 30% versus "Mild AR’’: AS 31%, Mixed 31%, AR 38%), however a trend does exist within each subgroup (see Table 5); or, alternatively, an unrecognised difference in the baseline LVEDDi between the two groups (but no significant difference existed between the mean baseline LVEDDi: Nil-Trivial AR baseline LVEDDi 2.89 cm/m2 (95% CI for mean 2.82–2.96)) versus Mild AR baseline LVEDD 2.97 cm/cm2 (95% CI 2.84–3.11); p = 0.309. Despite having a slightly larger LVEDDi at four to five year followup, it does not appear that those with mild AR had progressive LV dilatation over time. The size of the neo-aortic root during follow-up was not associated with the degree of AR observed during follow up.
Table 2 Echocardiographic data before and 5 years post aortic valve replacement in patients with aortic stenosis (AS), Mixed disease (AS + AR) and aortic regurgitation (AR) who completed 5 years of follow-up. Echo parameter
LVEDDi (cm/m2) LVESDi (cm/m2) PW thickness (cm) FS (%)
AS
Mixed
AR
Pre op
Post op
p value
Pre op
2.55 (0.35) 1.55 (0.36)
2.55 (0.34) 1.63 (0.39)
0.94 0.11
2.85 (0.38) 1.77 (0.32)
2.65 (0.3) <0.001 1.74 (0.30) 0.63
1.19 (0.21)
0.97 (0.19) <0.001
1.22 (0.18)
0.95 (0.13) <0.001
40.84 (8.13) 37.59 (7.01)
0.016
Post op
38.54 (7.31) 34.89 (7.83)
p value
0.038
Pre op
Post op
p value
3.34 (0.39) 2.26 (0.34)
2.67 (0.32) <0.001 1.85 (0.27) <0.001
1.03 (0.13)
0.99 (0.12)
0.178
32.98 (5.99) 32.06 (6.13)
0.351
136
W.M. Wilson et al.
Table 3 Normalisation of LVEDD and LVH post operation (percentage of patients with normal LVEDDi (defined as 3.2 cm/m2 in females and 3.1 cm/m2 in males) and LV wall thickness 11 mm pre and 5 years post operation). AS
2
LVEDDi <3.2 cm/m LVESDi <2.0 cm/m2 LVH < = 11 mm
AS/AR
Pre
5y
p
Pre
91.4% (107/117) 88% (97/110)
97.1% (69/71) 86% (49/57)
0.14 0.81
68.3% (41/60) 66.6% (40/60)
<0.001
38.2% (21/55)
40.0% (46/115)
85.9% (55/64)
5y 95% (38/40) 78% (28/36)
AR
97.1% (33/34)
p 0.001 0.35 <0.001
ALL
Pre
5y
p
Pre
5y
p
LVEDDi <3.2 cm/m2
35.3.0% (30/85)
96.0% (48/50)
<0.001
69.4% (182/262)
96.2% (155/161)
<0.001
LVESDi <2.0 cm/m2
28.5% (24/84)
72.9 (35/48)
<0.001
63.3% (161/254)
79.4% (112/141)
0.001
LVH < = 11 mm
73.8% (59/80)
97.8% (45/46)
<0.001
50.4% (126/250)
92.4% (133/144)
<0.001
Fig. 2 LVEDDi over time according to pre-operative condition (error bars
95% CI for mean).
Fig. 3 Change in LVEDDi (cm/m2) over the time (according to pre-operative condition).
137
Echocardiographic Predictors of LV Recovery Post AVR
Table 4 LVEDDi pre operation and at 5 years post operation according to pre-operative LV size and stratified according to pre-operative condition and (Normal indexed LVEDD is defined as 3.2 cm/m2 in females and 3.1 cm/m2 in males). Normal LVEDDi N ALL
Enlarged LVEDDi
Pre
5 years
p value
N
Pre
5 years
3.51 (0.30)
2.79 (0.33)
3.5
2.55
3.39 (0.17) 3.54 (0.33)
2.93 (0.33) 2.77 (0.32)
115
2.63 (0.32)
2.54 (0.30)
0.003
43
AS
68
2.52 ( 0.31)
2.55 (0.34)
0.364
2
Mixed AR
29 18
2.69 (0.25) 2.98 (0.15)
2.56 (0.23) 2.48 (0.21)
0.014 <0.001
9 32
p value <0.001 NA 0.003 <0.001
Fig. 4 LVEDD post procedure according to pre-operative LVEDD (normal is 3.2 cm/m2 in females and 3.1 cm/m2 in males) (error bars 95% CI for mean).
Fig. 5 LV wall thickness post procedure according to pre-operative condition (error bars for total cohort only) (error bars 95% CI for mean).
138
W.M. Wilson et al.
Fig. 6 Neo-aortic size post Ross procedure (error bars represent 95% CI for mean and present for total cohort only).
Multivariate Analysis At time period four to five years post operation, independent predictors of a larger LVEDDi included: female gender (LVEDDi 4.9% larger than males (95% CI 1.0–8.8, p = 0.014)); baseline LVEDDi >3.0 cm/m2 (LVEDDi 10.2% larger than if baseline LVEDDi 3.0 cm/m2 (95% CI 5.9–14.7, p < 0.0001)); and degree of post-operative AR (LVEDDi 5.3% larger if mild AR at 0.5–1 year compared to nil-trivial AR at same timepoint (95% CI 1.1–9.6%, p = 0.014)). Pre-operative aortic regurgitation and baseline LVESDi >2.0 cm/m2 were not predictors of a larger LVEDDi at four to five years. Higher pre-operative
fractional shortening was a predictor of a smaller LVEDDi at follow-up (an increase in 1 point FS decreased the LVEDDi by 0.3% (95% CI 0.01–0.6, p = 0.012). Similarly, at five years follow-up, independent predictors of a larger LVESDi included: female gender (LVESDi 7.9% larger than males (95% CI 1.1–14.6 p = 0.023)), and baseline LVEDDi >3.0 cm/m2 (LVESDi 10.2% larger than if baseline LVEDDi 3.0 cm/m2 (95% CI 2.7–17.7, p = 0.008)). Neither pre-operative valve lesion, pre-operative LVESDi >2.0 cm/ m2 nor post-operative mild AR were predictors of LVESDi at four to five years. Higher pre-operative fractional shortening
Fig. 7 Degree of AR post operation over time (expressed as a proportion).
139
Echocardiographic Predictors of LV Recovery Post AVR
Fig. 8 LVEDDi according to degree of AR at 0.5–1 year post operation (error bars–95% CI for mean).
was a predictor of smaller LVESDi (LVESDi 0.8% less for each 1 point increase in FS (95% CI 0.4–1.2, p < 0.0001)). Furthermore, the multivariate regression model shows that at one year post-operation, the degree of reduction in LVEDDi from baseline (change in LVEDDi) depends on: gender (male patients exhibited 8.3% less reduction in LVEDDi compared to female patients (95% CI: 0.6–16.0%, p = 0.033)); degree of postoperative AR (patients with mild AR exhibited 11.3% less reduction in LVEDDi compared to those with nil-trivial AR (95% CI: 3.0–19.6%, p = 0.008)) and LVEDDi size at baseline (there is a 70.6% greater reduction in LVEDDi per 1 cm increase in the baseline LVEDDi (95% CI: 50.3–90.8, p < 0.001)). At five years post operation there was less reduction in LVEDDi from baseline if there existed: mild post-operative AR (those with mild post-operative AR had 14.2% less reduction compared to those with nil-trivial AR (95% CI 4.0%– 24.3%, p = 0.006)). There was a greater reduction in LVEDDi according to: baseline LVEDDi (there was a 83.4% greater reduction in LVEDDi per 1 cm increase in the baseline
Table 5 Mean (SD) LVEDDi (cm/m2) according to post-operative AR degree, stratified according to preoperative condition. p value
5 years
Nil-Trivial
Mild
ALL
2.58 (0.32)
2.72 (0.33)
0.035
AS
2.53 (0.33)
2.72 (0.37)
0.162
2.62 ( 0.27)
2.69 (0.36)
0.497
2.63 (0.34)
2.74 ( 0.31)
0.311
AS/AR AR
LVEDDi (95% CI: 55.7–111.2%, p < 0.001)) and the pre-operative valve lesion (patients in the AR group had a 36.0% greater reduction compared to those in the AS group when adjusted for all other variables (95% CI 21.9–50.2%, p < 0.001). The reduction in LVESDi from baseline to five years level was dependent on baseline LVESDi only with a greater reduction according to higher baseline LVESDi (each 1 cm increase in baseline LVESDi is associated with a 86.1% greater reduction in LVESDi at five years (95% CI 0.35.2– 1.37, p < 0.001)) when adjusted for all other factors. The only predictor for a LVEDDi 3.0 cm/m2 at four to five years was LVEDDi at baseline being greater than 3.0 cm/ m2 (OR 2.4 (95% CI 1.4–4.1, p = 0.001). The only predictor for a LVESDi 2.0 cm/m2 at four to five years was LVESDi at baseline being greater than 2.0 cm/m2 (OR 1.8 95% CI 1.03–3.2, p = 0.041). Age was not a predictor of LV size post operation.
Discussion Surgical treatment lowers mortality rate for aortic valve disease, and in our study the five years mortality rate was <1% [1]. The main benefit of AVR appears to be via haemodynamic effects with removal of after-load in AS and removal of pre-load in AR. Pre-operative LV size and function are clear predictors of long-term survival post aortic valve replacement [12–14] and early improvement in ejection fraction post surgery has been demonstrated to be an independent predictor of survival [15], as has a post-operative LV diastolic diameter [16] and post-operative reduction in end-diastolic dimension of more than 20% [12].
140
Left ventricular dilatation has previously been shown to normalise in most patients with AR post AVR within the first few months [16,17]. This was reflected in our study as a reduction in LV size to normal was evident in the majority of cases by one year, and this was persistent throughout follow-up. Interestingly, short-term and long-term improvement in left ventricular systolic function post operation appears related to the early reduction in left ventricular dilatation arising from correction of left ventricular volume overload and, furthermore, late improvements in ejection fraction seem to occur only in those with early improvements in LV function [2]. Poor functional LV recovery has also been shown to be predicted by pre-operative LV spherical shape (ratio of minor axis to major axis of LV) [18] and systolic radial strain rate [19]. It is recognised that a certain proportion of patients do not derive benefit from surgery in terms of reduction in LV size and improvement in function (presumably relating to preoperative irreversible myocardial damage) and these patients have worse long term clinical outcomes [20]. Indeed, post-operative LV diastolic diameter dimension has been shown to be a strong predictor of late death from congestive heart failure after AVR for aortic regurgitation [16]. A number of factors in our study were associated with a significantly larger LVEDDi post procedure, in particular: a larger pre-operative ventricular size; the degree of post-operative AR, and female gender.
Predictors of LV Recovery Indexed values for ventricular size appear to be better predictors of LV recovery and clinical outcomes post AVR than non-indexed values, especially in those with low body weight [6,21–23]. The degree of enlargement and normal cut-offs for indexed LV dimensions are not clearly defined but the most commonly quoted measurements are indexed LVESD values less than 2.5 cm/m2 and indexed LVEDD values of less than 3.5 cm/m2. An indexed baseline LVESD value of >2.5 cm/m2 was a marker of poor outcome in patients with moderate to severe AR managed conservatively in a study by Dujardin et al. [24]. In another study of 171 patients undergoing AVR for AR, the sensitivity and specificity in predicting normalisation of LV function were 88% and 92% for indexed LVESD values <3.53 cm/m2 and 71% and 86% for indexed LVEDD values <4.44 cm/m2, whilst in a study of 66 patients undergoing AVR for severe AS with LV dysfunction, a cut off value of pre-operative indexed LVESD 2.75 cm/m2 was 85% sensitive and 72% specific in predicting intermediate-term recovery of LV function after AVR [6]. Morgan et al., however, found that the best discrimination for survival post AVR was seen at lower indexed values (2.0 cm/m2 for indexed LVESD and 3.0 cm/m2 for indexed LVEDD) [25]. Other studies have suggested that baseline LV end-systolic indices are more predictive of poor LV recovery than LV end-diastolic indices [17,26]). In our study, a baseline LVEDDi >3.0 cm/m2 was an independent predictor of a larger LVEDDi and LVESDi at five years follow-up.
W.M. Wilson et al.
The presence of mild AR at follow-up echocardiographic assessment in the first post-operative year in this study appeared to have an impact on the size of the left ventricle during longer term follow-up; it was an independent predictor of a larger LVEDDi at five years, and an independent predictor of a lower degree of reduction of LVEDDi at one and five years. It is important to note, however, that the presence of mild AR did not seem to cause progressive LV dilatation over long-term follow-up. Moderate or severe residual AR post AVR generally mandates redo surgery given it is associated with poorer clinical outcomes. In a study by Sponga et al. of 3201 patients who had undergone aortic valve replacement in a single institution, the incidence of residual AR >1/4 (i.e. more than mild) was 4.2%, which was associated with an increased risk of need for reoperation as well as an increased risk of mortality at all time points during follow-up [27]. Mild residual AR, however, is often expected to improve over time and appears benign, although few studies have addressed this specifically; in a small study of 40 patients identified as having a small periprosthetic leak by Rallidis et al. [28], no deaths were recorded and echocardiographic parameters remained unchanged during five year follow-up, and no increase in mortality at six weeks post AVR was found by O’Rourke et al. in a study of 113 patients with trivial/mild AR post surgery [29]. The advent of transcatheter aortic valve replacement has led to a renewed interest in the impact of residual aortic regurgitation post valve replacement given the relatively high frequency of paravalvular leak post TAVI compared to surgical AVR [30]. Paravalvular leak >1/4 has been shown to be an independent predictor of mortality post TAVI in a number of trials in keeping with the data post surgical AVR [31,32], however, interestingly, two year outcome data from the PARTNER trial [30] suggest that the presence of mild, moderate or severe paravalvular or total aortic regurgitation (vs. none or trace) after TAVI is associated with increased late mortality. The effect of aortic regurgitation on mortality was certainly proportional to the severity of the regurgitation, however even mild aortic regurgitation was associated with an increased rate of late deaths in the PARTNER trial. It should be noted that this association was not analysed in a multivariate fashion and this finding was not corroborated by the core valve Advance trial of 1015 patients who underwent transcatheter valve insertion where six month mortality in those with mild post-procedure AR was the same as those with no AR [33]. Furthermore, in a multicentre Canadian study of 339 patients, the presence of mild AR did not impact upon LV diameter nor LV ejection fraction over time and no differences were observed between those patients with mild AR and those with nil-trace AR [34]. A potential limitation of drawing conclusions from TAVI trials regarding the impact of mild AR is that assessment of paravalvular leak severity is more difficult and less reproducible than that of valvar regurgitation severity. It has previously been reported that women have a poorer outcome after aortic valve replacement compared to men [35,36]. In a report by Klodas et al., female gender was an
141
Echocardiographic Predictors of LV Recovery Post AVR
independent predictor of lower late survival post AVR for AR; women in this study were more likely to have surgery on the basis of symptoms rather than LV dimensions which served to highlight the fact that generalisation of surgical criteria based on unadjusted LV diameters established in men may not be appropriate in women. The authors concluded that surgical correction of AR should be considered at an earlier stage in women [35]. Conversely, female sex was an independent predictor of greater incremental improvement in ejection fraction post AR in a study of 1012 patients post AVR by Morris et al. [15]. In our study, female gender was an independent predictor of larger indexed LVEDDi and LVESDi post-operatively. The impact of gender on LV mass regression post surgery for AS is unclear with multiple studies yielding conflicting results [37].
Differential Effects of Valve Replacement According to Pre-Operative Condition Few studies have assessed the differential effects of valve replacement according to pre-operative condition. Waszyrowski et al. [10] followed 143 patients post AVR for a mean of five years and analysed them according to preoperative valve lesion (AS, AR or mixed valvular disease). They noted LVEDD decreased significantly in all groups, however LVESD decreased only in the AR and mixed groups whilst LVEF increased in the mixed group only. As expected, the LV wall thickness reduced in the AS and mixed groups only. LVEDDi decreased significantly in the AR and mixed groups in our study such that no significant difference in LVEDDi existed between the AS/mixed/AR groups at five years. More recently, Lamb et al. [9] focussed on LV mass index and diastolic function post AVR, which is of interest given impaired diastolic function has implications for morbidity and mortality post AVR [38–40]. This MRI based assessment suggested that early after AVR, patients with aortic valve stenosis show a decrease in both LVMi (LV mass index) and LVMi/LVEDVi ratio (LV end diastolic volume index) with an improvement in diastolic filling. This was in clear contrast to patients with aortic regurgitation where LVMi decreases less rapidly than LVEDVi, causing concentric remodelling of the LV and deterioration in diastolic filling in these patients. In a similar echocardiographic based study by Iyem et al. [41], the timecourse for LV remodelling and improvement in EF was different depending on the pre-operative lesion; in the AS group, LV remodelling was a continuing process with significant improvements occurring in the second year of follow-up, whilst LV remodelling and improvement in EF did not change significantly in the AR group between six months and two years post-operatively. In our cohort, LVEDDi normalised by 0.5–1 year follow-up (if dilated pre-operatively) and remained stable thereafter. A disproportionately high relative wall thickness before AVR is associated with excessive perioperative risk. Reverse LV remodelling after aortic valve replacement with regression of myocardial hypertrophy occurs mostly within the
first few years [42] and has been shown to result in improvement of LV longitudinal myocardial strain and decrease of Ntpro-BNP plasma levels [43], whilst failure of regression of LVH post AVR for AS is associated with worse longterm outcomes [4,44,45]. Mean LV wall thickness reduced significantly post Ross procedure in the majority (90%) of patients in our study and did so within the first two to three years. This impressive reduction in LVH is in contrast to the study by Waszyrowski et al. [10], where normalisation of left ventricular hypertrophy was observed in less than half of the study population. An increase in LV systolic function was not seen in our study, which likely reflects near normal pre-operative function relating to careful patient selection and proper surgery timing in a relatively young cohort. Nonetheless, it was evident on multivariate analysis that a higher baseline fractional shortening was a predictor of smaller ventricular dimensions (both LVEDDi and LVESDi) at five years.
Study Limitations Previous studies regarding the impact of AVR on left ventricular remodelling have been limited in particular by heterogeneity of the surgical procedure undertaken. A relative strength of our study is that each operation was performed by the same surgeon in one of three institutions, thus limiting confounding due to differences in surgical techniques. Inherent limitations exist in echocardiographic assessment of LV dimensions and wall thickness as well as in the evaluation of AR. Variability in the performance of the echocardiograms over more than a 10 year period is an unavoidable limitation, however, all the studies were predominantly performed at one institution where strict protocols for echocardiographic performance exist. Not all previous images were available, hence the data for this study have been obtained directly from echocardiogram reports and echocardiographic images were not independently assessed in a controlled and blinded fashion at a core lab, which represents another limitation of this study. It would be interesting to know whether those with persistent LV enlargement fare worse clinically; the vast majority of patients in the study were NYHA class 1 at long term follow-up so there appears to be no gross difference in symptomatology between patients. Objective assessments of functional capacity (through cardiopulmonary or stress testing for example) were not available for many patients to allow us to define subtle differences in exercise capacity between those with normal versus larger ventricles postoperation. We have confirmed the quoted mortality rate, however, through linkage with the Australian death index. The Ross procedure has a lower post-operative gradient than other forms of AVR and this represents a younger cohort than other studies, hence these findings may not necessarily be extrapolated to all patients undergoing AVR.
Conclusion A number of factors impact upon the degree of reduction in left ventricular size post AVR, in particular pre-operative LV
142
size. Female gender and mild AR early post AVR were also predictors of larger LV dimensions in this study. Patients with mild post-operative AR did not have progressive LV enlargement, however, suggesting this finding has limited clinical significance.
W.M. Wilson et al.
[19]
[20]
References [1] Scognamiglio R, Negut C, Palisi M, Fasoli G, Dalla Volta S. Long-term survival and functional results after aortic valve replacement in asymptomatic patients with chronic severe aortic regurgitation. J Am Coll Cardiol 2005;45:1025–30. [2] Bonow RO, Dodd JT, Maron BJ, O’Gara PT, White GG, McIntosh CL, et al. Long-term serial changes in left ventricular function and reversal of ventricular dilatation after valve replacement for aortic regurgitation. Circulation 1988;78:1108–20. [3] Carroll JD, Gaasch WH, Zile MR, Levine HJ. Serial changes in left ventricular function after correction of chronic aortic regurgitation: dependence on early changes in preload and subsequent regression of hypertrophy. Am J Cardiol 1983;51:476–82. [4] Bech-Hanssen O, Wallentin I, Houltz E, Suurkula MB, Larsson S, Caidahl K. Gender differences in patients with severe aortic stenosis: impact on preoperative left ventricular geometry and function, as well as early postoperative morbidity and mortality. Eur J Cardiothorac Surg 1999;15:24–30. [5] Jindani A, Neville EM, Venn G, Williams BT. Paraprosthetic leak: a complication of cardiac valve replacement. J Cardiovasc Surg 1991;32: 503–8. [6] Cho SH, Byun CS, Kim KW, Chang BC, Yoo KJ, Lee S. Preoperative indexed left ventricular dimensions to predict early recovery of left ventricular function after aortic valve replacement for chronic aortic regurgitation. Circ J 2010 Nov;74:2340–5. [7] Lerakis S, Hayek S, Douglas P. Paravalvular aortic leak after transcatheter aortic valve replacement current knowledge. Circulation 2013;127:397–407. [8] Kumpuris AG, Quinones MA, Waggoner AD, Kanon DJ, Nelson JG, Miller RR. Importance of preoperative hypertrophy, wall stress and endsystolic dimension as echocardiographic predictors of normalization of left ventricular dilatation after valve replacement in chronic aortic insufficiency. Am J Cardiol 1982;49:1091–100. [9] Lamb H, Beyerbacht HP, de Roos A, van der Laarse A, Vliegen HW, Leujes F, et al. Left ventricular remodeling early after aortic valve replacement: differential effects on diastolic function in aortic valve stenosis and aortic regurgitation. J Am Coll Card 2002;40(12):2182–8. [10] Waszyrowski T, Kasprzak JD, Krzemihka-Pakula M, Droz J, Dziatkowiak A, Zaslonka J. Regression of left ventricular dilatation and hypertrophy after aortic valve replacement. Int J Cardiol 1996;57:217–25. [11] Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, et al. Recommendations for chamber quantification. Eur J Echocardiogr 2006;7:79–108. [12] Corti R, Binggeli C, Turina M, Jenni R, Luscher TF, Turina J. Predictors of long-term survival after valve replacement for chronic aortic regurgitation is Mmode echocardiography sufficient? Eur Heart J 2001;22:866. [13] Tornos P, Sambola A, Permayer-Miralda G, Evangelista A, Gomez Z, Soler-Soler J. Long-term outcome of surgically treated aortic regurgitation: influence of guideline adherence toward early surgery. J Am Coll Cardiol 2006;47:1012. [14] Turina J, Milincic J, Seifert B, Turina M. Valve replacement in chronic aortic regurgitation: true predictors of survival after extended follow-up. Circulation 1998;98:100II–6II. [15] Morris JJ, Schaff HV, Mullany CJ, Rastogi A, McGregor CGA, Daly RC, et al. Determinants of survival and recovery of left ventricular function after aortic valve replacement. Ann Thor Surg 1993;56((1) July):22–9. [16] Henry WL, Bonow RO, Borer JS, Ware JH, Kent KM, Redwood DR, et al. Observations on the optimum time for operative intervention for aortic regurgitation. I. Evaluation of the results of aortic valve replacement in symptomatic patients. Circulation 1980;61:471–83. [17] Kumpuris AG, Quinones MA, Waggoner AD, Kanon DJ, Nelson JG, Miller RR. Importance of preoperative hypertrophy, wall stress and endsystolic dimension as echocardiographic predictors of normalization of left ventricular dilatation after valve replacement in chronic aortic insufficiency. Am J Cardiol 1982;49:1091–100 [8]. [18] Matsumura Y, Gillinov AM, Toyono M, Wada N, Yamano T, Thomas JD, et al. Usefulness of left ventricular shape to predict the early recovery
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33] [34]
[35]
[36]
[37]
[38]
[39]
of left ventricular function after isolated aortic valve replacement for aortic valve stenosis. Am J Cardiol 2008;11:1530–4. Onishi T, Kawai H, Tatsumi K, Kataoka T, Sugiyama D, Tanaka H, et al. Preoperative systolic strain rate predicts postoperative left ventricular dysfunction in patients with chronic aortic regurgitation. Circ Cardiovasc Imag 2010;3:134–41. Bonow RO, Picone AL, McIntosh CL, Jones M, Rosing DR, Maron BJ, et al. Survival and functional results after valve replacement for aortic regurgitation from 1976 to 1983: impact of preoperative left ventricular function. Circulation 1985;72:1244–56. Brown ML, Schaff HV, Suri RM, Zhuo L, Sundt TM, Dearani JA, et al. Indexed left ventricular dimensions best predict survival after aortic valve replacement in patients with aortic valve regurgitation. Ann Thor Surg 2009;87:1170–2117. Sambola A, Tornos P, Ferreira-Gonzalez I, Evangelista A. Prognostic value of preoperative indexed end-systolic left ventricle diameter in the outcome after surgery in patients with chronic aortic regurgitation. Am Heart J 2008;155:1114–20. Ding W, Lam Y, Kaya MG, Li W, Chung R, Pepper JR, et al. Echocardiographic predictors of left ventricular functional recovery following valve replacement surgery for severe aortic stenosis. Int Heart J 2008;128:178–84. Dujardin KS, Enriquez-Sarano M, Schaff HV, Bailey KR, Seward JB, Tajik AJ. Mortality and morbidity of aortic regurgitation in clinical practice: a long-term follow-up study. Circulation 1999;99:1851–7. Brown ML, Schaff HV, Suri RM, Zhuo L, Sundt TM, Dearani JA, et al. Indexed left ventricular dimensions best predict survival after aortic valve replacement in patients with aortic valve regurgitation. Ann Thorac Surg 2009;87:1170–6. Cox DA, Walton K, Bartz PJ, Tweddell JS, Frommelt PC, Earing MG. Predicting left ventricular recovery after replacement of a regurgitant aortic valve in pediatric and young adult patients: is it ever too late? Pediatr Cardiol 2012 [Epub ahead of print]. Sponga S, Perron J, Dagenais F, Mohammadi S, Baillot R, Doyle D, et al. Impact of residual regurgitation after aortic valve replacement. Eur J Cardiothor Surg 2012;42:486–92. Rallidis LS, Moyssakis IE, Ikonomidis I, Nihoyannopoulos P. Natural history of early aortic paraprosthetic regurgitation: a five-year follow-up. Am Heart J 1999;138:351–7. O’Rourke DJ, Palac RT, Malenka DJ, Marrin CA, Arbuckle BE, Plehn JF. Outcome of mild periprosthetic regurgitation detected by intraoperative transesophageal echocardiography. J Am Coll Cardiol 2001;38:163–6. Kodali SK, Williams MR, Smith CR, Svensson LG, Webb JG, Makkar RR, et al. Leon MB for the PARTNER trial investigators. Two year outcomes after transcatheter or surgical aortic valve replacement. N Engl J Med 2012;366:1686–95. Tamburino C, Capodanno D, Ramondo A, Pertronio AS, Ettori F, Santoro G, et al. Incidence and predictors of early and late mortality after transcatheter aortic valve implantation in 663 patients with severe aortic stenosis. Circulation 2011;125:299–308. Abdel-Wahab M, Zahn R, Sherif M, Schneider S, Gerckens U, Schuler G, et al. In-hospital outcome of aortic regurgitation after TAVI: results from the prospective multicentre German TAVI registry. Eur Heart J 2010;31:160. Linke A, on behalf of the ADVANCE Steering Committee and Investigators. The ADVANCE TAVR registry. In: Presented at TCT 2012; 2012. Rode´s-Cabau J, Webb JG, Cheung A, Ye J, Dumont E, Osten M, et al. Long-term outcomes after transcatheter aortic valve implantation insights on prognostic factors and valve durability from the canadian multicenter experience. J Am Coll Cardiol 2012;60:1864–75. Klodas E, Enriquez-Sarano M, Tajik AJ, Mullany CJ, Bailey KR, Seward JB. Surgery for aortic regurgitation in women. Contrasting indications and outcomes compared with men. Circulation 1996;94:2472. Flores-Marı´n A, Go´mez-Doblas JJ, Caballero-Borrego J, Cabrera-Bueno F, Rodrı´guez-Bailo´n I, Meleroa JM, et al. Long-term predictors of mortality and functional recovery after aortic valve replacement for severe aortic stenosis with left ventricular dysfunction. Rev Esp Cardiol 2010;63:36–45. Villa E, Troise G, Cirillo M, Brunelli F, Tomba MD, Mhagna Z, et al. Factors affecting left ventricular remodeling after valve replacement for aortic stenosis. An overview. Cardiovasc ultrasound 2006;4:25. Lund O, Flø C, Jensen FT, Emmertsen K, Nielsen TT, Rasmussen BS, et al. Left ventricular systolic and diastolic function in aortic stenosis. Eur Heart J 1997;18:1977–87. Lund O, Nielsen TT, Emmertsen K, Flø C, Rasmussen BS, Jensen FT, et al. Mortality and worsening of prognostic profile during waiting time
Echocardiographic Predictors of LV Recovery Post AVR
for valve replacement in aortic stenosis. Thorac Cardiovasc Surg 1996;44: 289–95. [40] Villari B, Vassalli G, Monrad ES, Chiariello M, Turina M, Hess OM. Normalization of diastolic dysfunction in aortic stenosis late after valve replacement. Circulation 1995;91:2353–8. [41] Iyem H, Sekuri C, Tavli M, Bu¨ket S. Left ventricular hypertrophy and remodeling after aortic valve replacement. Asian Cardiovasc Thorac Ann 2007;15:459–62. [42] Lim E, Ali A, Theodorou P, Sousa I, Ashrafian H, Chamageorgakis T, et al. Longitudinal study of the profile and predictors of left ventricular mass regression after stentless aortic valve replacement. Ann Thorac Surg 2008;85:2026–9.
143
[43] Poulsen SH, Søgaard P, Nielsen-Kudsk JE, Egeblad H. Recovery of left ventricular systolic longitudinal strain after valve replacement in aortic stenosis and relation to natriuretic peptides. J Am Soc Echocardiogr 2007 Jul;20:877–84. [44] Aurigemma GP, Silver KH, McLaughlin M, Mauser J, Gaasch WH. Impact of chamber geometry and gender on left ventricular systolic function in patients >60 years of age with aortic stenosis. Am J Cardiol 1994;74:794–8. [45] Gaasch WH, Carroll JD, Levine HJ, Criscitiello MG. Chronic aortic regurgitation: prognostic value of left ventricular end-systolic dimension and end-diastolic radius/thickness ratio. J Am Coll Cardiol 1983;1: 775–82.