Introduction. Aims. Keywords Mitral regurgitation Echocardiography Left ventricular dysfunction Surgery

Transcription

1 European Journal of Echocardiography (2011) 12, doi: /ejechocard/jer128 Predicting left ventricular dysfunction after valve repair for mitral regurgitation due to leaflet prolapse: additive value of left ventricular end-systolic dimension to ejection fraction Christophe Tribouilloy 1,2 *, Dan Rusinaru 3, Catherine Szymanski 1, Sonia Mezghani 1, Alexandre Fournier 1, Franck Lévy 1,3, Marcel Peltier 1, Ammar Ben Ammar 4, Doron Carmi 3, Jean-Paul Remadi 3, Thierry Caus 2,3, and Gilles Touati 3 1 Department of Cardiology, University Hospital Amiens, Avenue René Laënnec, Amiens Cedex 1, France; 2 INSERM, ERI 12, University of Picardie, Amiens, France; 3 Department of Cardiac Surgery, University Hospital Amiens, Amiens, France; and 4 Department of Anesthesiology, University Hospital Amiens, Amiens, France Received 8 May 2011; accepted after revision 6 July 2011; online publish-ahead-of-print 4 August 2011 Aims Left ventricular (LV) dysfunction is the first cause of late mortality after mitral valve surgery. In this retrospective analysis, we studied the association between preoperative echocardiographic LV measures and occurrence of LV dysfunction after mitral valve repair (MVR).... Methods Between 1991 and 2009, 335 consecutive patients underwent MVR for severe mitral regurgitation due to leaflet prolapse in our institution. Echocardiography was performed preoperatively and at 10.8 ( ) months after surgery and results in 303 patients who represented the study population. Cardiac events were recorded during follow-up. LV ejection fraction (EF) decreased from % before surgery to % post-operatively (P, 0.001). Preoperative EF,64% and LV end-systolic diameter (ESD) 37 mm were the best cut-off values for the prediction of post-operative LV dysfunction (EF, 50%). On the basis of a combined analysis, the occurrence of post-operative LV dysfunction was 9% when EF was 64% and LVESD,37 mm, 21% with EF,64% or LVESD 37 mm, and 33% with EF,64% and LVESD 37 mm (P for trend,0.001). The combined variable EF, 64% and LVESD 37 mm added incremental prognostic value to the multivariable regression model (P ¼ 0.001).... Conclusion Simple preoperative echocardiography measures allow the prediction of LV dysfunction after MVR in patients with leaflet prolapse. Patients with preoperative EF 64% and LVESD,37 mm incur relatively low risk of post-operative LV dysfunction Keywords Mitral regurgitation Echocardiography Left ventricular dysfunction Surgery Introduction Mitral valve repair (MVR) is the standard of care for severe organic non-rheumatic mitral regurgitation (MR). 1,2 MVR has excellent long-term results and is associated with lower operative mortality and better long-term survival than valve replacement. 3 5 Guidelines recommend surgery, preferably MVR, for patients with severe organic MR and severe symptoms or overt left ventricular (LV) dysfunction. 1,2 However, operated patients may still present high rates of adverse cardiac events after successful surgery. 6 8 Restoration of life expectancy after MVR is therefore based on early identification of patients at high risk under conservative treatment. Patients with severe MR and overt preoperative LV dysfunction have increased post-operative mortality. 7 Chronic volume overload induces progressive LV dysfunction which initially may be concealed behind a normal ejection fraction (EF). 7,9 Thus, elimination of volume overload after surgery may generate further post- * Corresponding author. Tel: ; fax: , tribouilloy.christophe@chu-amiens.fr Published on behalf of the European Society of Cardiology. All rights reserved. & The Author For permissions please journals.permissions@oup.com

2 Predicting LV dysfunction after valve repair for MR due to leaflet prolapse 703 operative reduction in EF, 10,11 which is generally thought to be lower with MVR compared with valve replacement. 12 Therefore, the best predictors of post-operative LV dysfunction are still subject of debate Preoperative EF and LV end-systolic diameter (ESD) have been previously reported as predictors of postoperative LV dysfunction in surgical series that were small 9,10,16 or composed entirely or partially of mitral valve replacement. 7,9,10,13,15,16 Moreover, preoperative EF and LVESD cut-off values for the prediction of post-operative LV dysfunction are not well defined. In this single-centre study, we used our consecutive experience with MVR for severe MR due to leaflet prolapse to identify a subset of patients with low occurrence of post-operative LV dysfunction based on preoperative echocardiography measures. We hypothesized that preoperative EF and LVESD have additive effect and allow good prediction of post-operative LV dysfunction. Methods Study design and data collection Between 1 January 1991 and 31 December 2009, 335 patients underwent MVR for organic MR due to leaflet prolapse at our institution (Department of Cardiac Surgery, University Hospital Amiens, France). This study included all consecutive patients who underwent MVR for MR due to leaflet prolapse and had complete pre- and postoperative echocardiographic evaluation. Preoperative echocardiography was defined as echocardiography performed,3 months before surgery and post-operative echocardiography as echocardiography performed 9 12 months after surgery. Therefore, patients who died at surgery (n ¼ 9) and patients who died after hospital discharge, before the assessment of post-operative EF (n ¼ 2), were not included. Exclusion criteria were: (1) valve leakage caused by ischaemic heart disease or dilated cardiomyopathy, rheumatic heart disease, active infective endocarditis, congenital heart disease; (2) previous cardiac surgery; (3) concomitant aortic valve or aortic root surgery; (4) patients with history of myocardial infarction or recent acute coronary syndromes; and (5) patients with missing pre-/post-operative EF or LVESD data (n ¼ 21). We kept in the analysis patients who had associated coronary artery bypass graft surgery, tricuspid annuloplasty, and tricuspid valve plasty. Three hundred and three patients met the study s selection criteria and represented the study population. During the study period, 30 patients had mitral valve replacement for severe MR due to leaflet prolapse at our institution. Baseline clinical characteristics, comorbidity, symptoms, and operative data were collected from patients medical records. The standard EuroSCORE was calculated retrospectively for each patient using the calculator available online at 17 Atrial fibrillation (AF) at baseline was determined by electrocardiogram. We obtained institutional review board authorization prior to conducting the study. The study was conducted in accordance with institutional policies, national legal requirements, and the revised Helsinki declaration. Echocardiography All patients had baseline and post-operative Dopplerechocardiography using commercially available ultrasound systems. Time between preoperative echocardiography and surgery was days. Post-operative echocardiography was performed at 10.8 ( ) months after surgery. Diagnoses of leaflet prolapse and flail leaflets were identified according to the recommended criteria. 18,19 LV dimensions were assessed from parasternal long-axis views by two-dimensional (2D)-guided M-mode using the leading edge methodology at end-diastole and end-systole. EF and left atrium diameter measurements were guided by 2D echocardiography. 8,20,21 EF was estimated by the Simpson biplane method or visually. 20 LA diameter was measured in M-mode at end-systole in the parasternal long-axis view, according to the method proposed by the American Society of Echocardiography. 21 Severity of MR was assessed by Doppler echocardiography, 22,23 and grading was retrospectively established according to recent guidelines. 24 Pulmonary artery systolic pressure was estimated using the simplified Bernoulli equation (4 Tricuspid valve regurgitation velocity 2 ) adding the assumed right atrial pressure. Surgical procedures The most frequently observed valve lesion was posterior leaflet prolapse in the middle scallop caused by chordal elongation or rupture. Most cases of posterior leaflet prolapse were treated by triangular or quadrangular leaflet resection with suture repair of the involved portion of the posterior leaflet and prosthetic ring remodelling annuloplasty. Chordal shortening or transfer with annuloplasty was used for anterior leaflet prolapse. In the second decade of the study, we frequently used insertion of artificial Gore-Tex neochordae associated with ring annuloplasty. 25 Twelve per cent of patients (n ¼ 36) had associated coronary artery bypass graft surgery. End-points and follow-up The end-point of the study was the occurrence of post-operative LV dysfunction. Post-operative LV dysfunction was identified when postoperative EF was,50%. 13 We also analysed the prognostic value of post-operative LV dysfunction during follow-up, after the postoperative echocardiography. Cardiac events during follow-up were defined as cardiac death, hospitalization due to heart failure, MR moderate at follow-up, and redo cardiac surgery. Cardiac death was death due to LV dysfunction or sudden death. Mean overall duration of follow-up was 8.2 years. During follow-up, patients were monitored by their personal general practitioners or cardiologists. Events were ascertained by review of medical records, clinical interviews or by telephone calls to physicians, patients, and (if necessary) next of kin. Autopsy records and death certificates were consulted for attribution of causes of death. Statistical analysis Continuous variables were tested for normal distribution using the Kolmogorov Smirnov test. Variables not normally distributed were expressed as median (interquartile range). Normally distributed continuous variables were expressed as the mean + 1 SD. Categorical variables were summarized as frequency percentages and absolute numbers. The comparison of continuous variables between patients with and without post-operative LV dysfunction was carried out using the Mann Whitney U-test. Frequencies of categorical variables were compared with the x 2 test or Fisher s test, as appropriate. Receiver operating characteristic (ROC) curve analysis was used to identify the best cut-off values for preoperative EF and preoperative LVESD for predicting post-operative LV dysfunction. A new combined ordinal variable LVESD-EF was created to take into account the presence of one or both of the two preoperative echocardiographic parameters (EF and LVESD) according to the identified cut-off values. The relationship of baseline variables with the occurrence of post-operative LV

3 704 C. Tribouilloy et al. dysfunction was assessed using univariate and multivariable logistic regression analysis. An epidemiological approach was taken and factors thought to be important for the endpoints were entered in multivariable analyses: age, gender, New York Heart Association (NYHA) class, EuroSCORE, baseline AF, preoperative EF, preoperative LVESD, decade of surgery, and extracorporeal circulation time. Euro- SCORE and extracorporeal circulation time were normalized by log 10 transformation. The combined ordinal variable LVESD-EF was the last that was added to the multivariable regression model to assess its incremental predictive power. Model discrimination was assessed with ROC analysis by computing the area under the ROC curve (AUC). Comparisons between ROC curves were based on the method of Hanley and McNeil. 26 Calibration of the model was tested with the Hosmer Lemeshow test. Overall model performance was also assessed with the likelihood ratio. The likelihood ratio indicates the degree to which the pretest probability of an event is altered by information provided by a predictive model. The higher the model likelihood ratio, the greater the probability of accurately predicting events. Multivariable analysis of time to events according to post-operative EF was performed using a Cox proportional hazards model. Patients who died of non-cardiac causes were censored as non-events at the time of death. The proportional hazards assumption was confirmed using statistics and graphs based on the Schoenfeld residuals. A significance level of 0.05 (two-sided) was assumed for all statistical tests. Data were analysed with SPSS 13.0 (SPSS Inc., Chicago, IL, USA) and STATA (version 10, StataCorp LP, College Station, TX, USA). Results Baseline characteristics and surgery Characteristics of the study population, overall and according to post-operative LV dysfunction, are presented in Table 1. Mean age was 65 years and 70% (n ¼ 212) of patients were men. Symptoms were observed at baseline in more than 85% of the population and 22% (n ¼ 67) were in AF. A history of angina was present in only 3% of patients (n ¼ 9) and no patient had previous myocardial infarction. MR was severe in 91% of patients (n ¼ 276) and moderate to severe in 9% of patients (n ¼ 27). Mean preoperative EF and LVESD were % and mm, respectively. EF was assessed using the Simpson method in 266 patients (88%) and by visual estimation in 37 (12%). Angiotensin-converting enzyme inhibitors, b-blockers, and diuretics were prescribed in 178 (59%), 77 (25%), and 84 (27%) patients, respectively. Patients with post-operative LV dysfunction had a significantly higher frequency of AF on baseline electrocardiogram than those with post-operative EF 50% but no difference in EuroSCORE or preoperative NYHA class (Table 1). Posterior leaflet prolapse was the most frequent valvular abnormality (65% of patients, n ¼ 196). Ten per cent of patients (n ¼ 29) had anterior leaflet prolapse and 25% (n ¼ 78) prolapse of both leaflets. Patients with isolated posterior leaflet prolapse (65%) had leaflet resection with suture repair and prosthetic ring annuloplasty, while more complex techniques (chordal shortening, transfer, and use of neo-chordae) were used for anterior and bivalvular prolapse (35%). Ring annuloplasty was used in 94% of patients (n ¼ 285). Patients who underwent surgery in the 2000s had more often post-operative EF 50% (Table 1). Changes in ejection fraction and left ventricular dimension EF decreased significantly after surgery compared with preoperative EF ( vs %; P, 0.001). This was due to an important reduction in LV end-diastolic diameter ( vs mm; P, 0.001) with a small but not significant reduction in LVESD ( vs mm; P ¼ 0.12). Left atrial diameter and systolic pulmonary artery pressure decreased significantly compared with preoperative values ( vs , P, and vs , P, 0.001, respectively) due to correction of volume overload. Compared with patients with post-operative EF 50%, those with post-operative LV dysfunction had lower preoperative EF (P ¼ 0.003, Table 1 and Figure 1A) and greater LVESD (P ¼ 0.002, Table 1 and Figure 1B) and LV end-diastolic diameter (P ¼ 0.025, Table 1). The correlations between preoperative EF and post-operative EF and between preoperative LVESD and post-operative EF are displayed in Figure 2A and B. Echocardiography prediction of post-operative left ventricular dysfunction Both preoperative EF [OR: 1.05 ( ) per 1% decrement; P ¼ 0.011] and preoperative LVESD [OR: 1.07 ( ) per 1 mm increment; P ¼ 0.004] were associated with post-operative EF,50% on univariate logistic analysis. Further univariate predictors of post-operative LV dysfunction were: baseline AF, surgery in the 1990s, and LV end-diastolic diameter. On multivariable analysis adjusted for age, gender, NYHA class, EuroSCORE, baseline AF, decade of surgery, and extracorporeal circulation time, preoperative EF [OR: 1.02 ( ) per 1% decrement; P ¼ 0.04] and preoperative LVESD [OR: 1.03 ( ) per 1 mm increment; P ¼ 0.03] were independently predictive of post-operative LV dysfunction. When LV end-diastolic diameter was forced into this model, it was not independently associated with post-operative LV dysfunction [OR: 1.01 ( ); P ¼ 0.32]. To examine whether preoperative EF and preoperative LVESD allow identification of a high-risk subgroup of patients, these two variables were dichotomized according to the optimal cut-off values for prediction of post-operative LV dysfunction as identified by ROC analysis. These cut-off values were 64% for preoperative EF (Figure 3A) and 37 mm for preoperative LVESD (Figure 3B). Results of the multivariable logistic regression that included age, gender, NYHA class, EuroSCORE, baseline AF, decade of surgery, and extracorporeal circulation time as covariates are presented in Table 2 (Model 1). To further elucidate the incremental value of echocardiography variables, we used two combined ordinal variables that were added into multivariable logistic models on the top of variables in Model 1. The first variable took into account the presence of either preoperative EF,64% or preoperative LVESD 37 mm (Model 2), and the second one combined preoperative EF,64% and preoperative LVESD 37 mm (Model 3). When added to Model 1, the variable LVEF, 64% or LVESD 37 mm was an independent predictor of post-operative LV dysfunction [OR: 2.63 ( ); P ¼ 0.008] and added incremental value to the model (Model 2, Table 2). Finally, when added to Model 1, the variable LVEF,

4 Predicting LV dysfunction after valve repair for MR due to leaflet prolapse 705 Table 1 Preoperative and operative characteristics of the population, overall and according to post-operative left ventricular dysfunction (post-operative EF <50%) Characteristic Overall study group Post-operative EF < 50% Post-operative EF 50% P (n 5 303) (n 5 48) (n 5 255)... Clinical data Age (years) Male sex [% (n)] 70 (212) 69 (33) 70 (179) 0.84 NYHA class [% (n)] I 13 (41) 15 (7) 13 (34) 0.46 II 35 (106) 25 (12) 37 (94) III 40 (120) 45 (22) 39 (98) IV 12 (36) 15 (7) 11 (29) Overt symptoms 52% (156) 60% (29) 50% (127) 0.18 Heart rate (bpm) 73 (65 82) 78 (67 88) 65 (72 81) 0.26 Systolic blood pressure (mmhg) 130 ( ) 130 ( ) 130 ( ) 0.84 Diastolic blood pressure (mmhg) 80 (70 80) 80 (70 80) 80 (70 80) 0.95 Cardiac history and risk factors Angina [% (n)] 3 (9) 2 (1) 3 (8) 1.00 Hypertension [% (n)] 38 (114) 35 (17) 38 (97) 0.81 Diabetes mellitus [% (n)] 7 (22) 6 (3) 7 (19) 0.89 Standard EuroSCORE 2.07 ( ) 2.18 ( ) 2.07 ( ) 0.45 Atrial fibrillation [% (n)] 22 (67) 38 (18) 19 (49) Echocardiography data Prolapse [% (n)] Posterior leaflet 65 (196) 58 (28) 66 (168) 0.21 Anterior leaflet 10 (29) 6 (3) 10 (26) Both leaflets 25 (78) 36 (17) 24 (61) LV end-diastolic diameter (mm) LV end-systolic diameter (mm) LVEF (%) Left atrial diameter (mm) spap (mmhg) 45 (35 57) 50 (37 58) 45 (35 56) 0.61 Severe MR [% (n)] 91 (276) 94 (45) 91 (231) 0.56 Medical therapy [% (n)] ACE inhibitors 59 (178) 67 (32) 57 (146) 0.22 Angiotensin receptor blockers 8 (24) 6 (3) 8 (21) 0.64 b-blockers 25 (77) 25 (12) 26 (65) 0.94 Diuretics 27 (84) 25 (12) 28 (72) 0.65 Warfarin 23 (71) 31 (15) 22 (56) 0.16 Antiplatelet agents 18 (56) 10 (5) 20 (51) 0.12 Calcium channel blockers 21 (63) 25 (12) 20 (51) 0.43 Digoxin 20 (59) 25 (12) 18 (47) 0.31 Nitrates 53 (160) 42 (20) 55 (140) 0.43 Operative data Decade of surgery % (126) 58% (28) 39% (98) % (177) 42% (20) 61% (157) ECC time (min) 126 ( ) 136 ( ) 125 ( ) 0.05 Cross-clamp time (min) 90 (77 106) 95 (83 108) 89 (77 106) 0.11 Associated CABG surgery 12 (36) 15 (7) 11 (29) 0.53 Discharge residual MR moderate [% (n)] 1.7 (5) 2 (1) 1.6 (4) 0.79 ACE, angiotensin converting enzyme; CABG, coronary artery bypass graft; ECC, extracorporeal circulation; EuroSCORE, European System for Cardiac Operative Risk Evaluation; LV, left ventricular; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association; spap, systolic pulmonary artery pressure.

5 706 C. Tribouilloy et al. Figure 1 Preoperative LVEF (A) and preoperative LVESD (B) in patients with and without post-operative left ventricular dysfunction. LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter. Figure 2 Correlation between post- and pre-operative LVEF (A) and between post-operative LVEF and preoperative LVESD (B). LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; NYHA, New York Heart Association. 64% and LVESD 37 mm was associated with a 3.8-fold higher risk of post-operative LV dysfunction [OR: 3.78 ( ); P, 0.001] and significantly improved the model (Model 3, Table 2). Calibration was good for both Model 2 and Model 3, as shown in Table 2. The AUC (C statistic) of Model 3 (0.83) exceeded that of Model 1 (0.64; P ¼ 0.015) and Model 2 (0.74; P ¼ 0.002). Further adjustment for the preoperative severity of MR and the presence of residual post-operative MR did not essentially change the results. The variable LVEF, 64% and LVESD 37 mm still showed a stronger association with post-operative LV dysfunction [OR: 3.66 ( ); P, 0.001] compared with the variable LVEF,64% or LVESD 37 mm [OR: 2.54 ( ); P ¼ 0.009] and significantly improved the multivariable model (log likelihood: vs ; AUC: 0.86 vs. 0.79). The frequency of post-operative LV dysfunction was 23% when preoperative EF was,64% and 20% when preoperative LVESD was 37 mm. On the basis of the combined analysis, the occurrence of post-operative LV dysfunction was 9% when EF was 64% and LVESD,37 mm, 21% with EF,64% or LVESD 37 mm, and 33% with EF,64% and LVESD 37 mm (P for trend,0.001; Figure 4). In the 147 patients without overt symptoms at baseline (NYHA classes I and II), the combined variable LVEF, 64% and LVESD 37 mm was still associated with increased risk of post-operative LV dysfunction [OR: 2.66 ( ); P ¼ 0.06], although the statistical significance of the relationship was borderline. Seventy deaths occurred during a median follow-up 8.2 ( ) years. Twenty-nine deaths (41%) were of cardiac causes. There were 27 hospital admissions for heart failure. Residual or new-onset MR moderate was identified at follow-up in 55 patients (18.2%). MR was severe in 5 patients (1.7%), moderate to severe in 9 (3%) and mild to moderate in 41 (13.5%). Nineteen patients (6.3%) underwent redo cardiac surgery during follow-up (14 for significant MR, 3 for infective endocarditis, and 2 for coronary artery disease). Post-operative LV dysfunction

6 Predicting LV dysfunction after valve repair for MR due to leaflet prolapse 707 Figure 3 Receiver operating characteristic curve analysis of preoperative LVEF (A) and preoperative LVESD (B) for the prediction of postoperative left ventricular dysfunction. AUC, area under the curve; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter. Table 2 Multivariable logistic regression models for the prediction of post-operative left ventricular dysfunction independent predictive factors and model performance OR (95% CI) P-value Log AUC (95% CI) Hosmer Hosmer likelihood Lemeshow Lemeshow statistic P-value... Model ( ) Baseline atrial fibrillation (yes vs. no) 2.19 ( ) 0.02 Surgery period (1990s vs. 2000s) 1.51 ( ) 0.03 Model ( ) LVEF, 64% or LVESD 37 mm (yes vs ( ) no) Baseline atrial fibrillation (yes vs. no) 1.89 ( ) 0.07 Surgery period (1990s vs. 2000s) 1.54 ( ) 0.02 Model ( ) LVEF, 64% and LVESD 37 mm (yes vs ( ),0.001 no) Baseline atrial fibrillation (yes vs. no) 1.95 ( ) 0.05 Surgery period (1990s vs. 2000s) 1.56 ( ) 0.01 AUC, area under the receiver operating characteristic curve; CI, confidence interval; OR, odds ratio; LVEF, preoperative left ventricular ejection fraction; LVESD, preoperative left ventricular end-systolic diameter. Model 1 was adjusted for age, gender, New York Heart Association class, EuroSCORE, baseline atrial fibrillation, decade of surgery, and extracorporeal circulation time. Model 2 included LVEF, 64% or LVESD 37 mm (yes vs. no) on top of variables in Model 1. Model 3 included LVEF, 64% and LVESD 37 mm (yes vs. no) on top of variables in Model 1. was associated with more than 2.5-fold increase in the risk of cardiac death or heart failure during follow-up [hazard ratio (HR) 2.67 ( ); P ¼ 0.003; Figure 5]. Patients with postoperative EF,50% also had higher risk of cardiac events (cardiac death, hospitalization due to heart failure, residual or newonset MR moderate, and redo cardiac surgery) during follow-up [HR: 1.91 ( ); P ¼ 0.027]. Discussion Our findings show that classical echocardiography LV measures performed preoperatively are associated with LV dysfunction after MVR for MR due to leaflet prolapse. Post-operative LV dysfunction is observed in 16% of patients who undergo MVR and negatively affects long-term prognosis. We demonstrate the

7 708 C. Tribouilloy et al. Figure 4 Frequency of post-operative left ventricular dysfunction according to preoperative LVEF and LVESD. Cut-off values were identified by receiver operating characteristic curve analysis. LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter. Figure 5 Cumulative hazard of cardiac death and hospital admission for heart failure during follow-up according to postoperative left ventricular dysfunction. Results of the Cox proportional analysis with post-operative EF,50% and 50%. EF, left ventricular ejection fraction. additive value of preoperative LVESD to preoperative EF for the prediction of post-operative EF,50%. Patients with preoperative EF 64% and LVESD,37 mm represent a group at relatively low risk of post-operative LV dysfunction. Conversely, with preoperative EF,64% and LVESD 37 mm, the frequency of LV dysfunction after MVR is high (33%). The natural history of chronic MR is characterized in its early stages by a compensated phase followed by a progressive remodelling leading to irreversible LV dysfunction. Progressive impairment in LV contractility is initially compensated by modified loading conditions and EF is kept in the normal range for a long time. Low EF (,50%) usually implies irreversible LV dysfunction and carries a grim prognosis. 27 Surgical correction of chronic MR in a timely manner should theoretically restore normal LV function and life expectancy. Current guidelines recommend surgery, preferably MVR, for patients with severe organic MR and severe symptoms or overt LV dysfunction. 1,2 However, operated patients may still present high rates of adverse cardiac events after successful surgery. This is mainly due to irreversible structural and functional LV remodelling secondary to chronic volume overload. Early postoperative LV dysfunction was associated with excess long-term mortality in a series of patients who underwent valve repair or replacement in the 1980s for organic MR of different aetiologies, including ischaemic MR. 13 In our series of patients with MVR for leaflet prolapse, post-operative LV dysfunction was associated with more than 2.5-fold increase in the risk of cardiac death or heart failure during long-term follow-up. Elimination of volume overload after surgery may sometimes generate further 10 15% early reduction in EF, 10,11 which is generally thought to be lower with MVR compared with valve replacement. 12 After this initial decline, in most patients, EF shows a sustained improvement during the months following surgery. 15 Therefore, we performed post-operative echocardiography at 9 12 months after surgery because EF may improve after the early post-operative period. In the present series, EF and LV end-diastolic diameter decreased significantly after MVR, while changes in LVESD were not significant. These results are in accordance with previous studies reporting EF decline 15,28 and LV end-diastolic diameter 15 reduction after MVR. EF, 60% and LVESD 40 mm are associated with excess mortality under medical management and after mitral surgery in organic MR. 7,8,27 The prediction of post-operative LV dysfunction using preoperative parameters of LV size and function was previously studied in surgical series that were small 9,10,16 or composed entirely or partially of mitral valve replacement. 7,9,10,13,15,16 Lee et al. 29 reported increased risk of LV dysfunction after mitral surgery (replacement or repair) with preoperative EF,40%. Japanese patients in the series by Matsumura et al. 14 had increased risk of LV dysfunction after MVR when preoperative EF was,55% and preoperative LVESD was 40 mm. In a cohort of patients undergoing mitral surgery for MR caused by leaflet prolapse, EF recovery detected by echocardiography at years post-operatively was predicted by a preoperative EF 65% and an LVESD,36 mm. 15 We undertook an analysis of post-operative EF at a definite time point after surgery in a homogenous cohort of patients who underwent MVR for severe MR due to leaflet prolapse. Thus, we tried to eliminate bias associated with MR aetiology, surgical techniques (repair vs. replacement), and early post-operative EF recovery over time. Cut-off values for preoperative EF and preoperative LVESD were tested in logistic models, and the incremental value of each parameter was assessed. The frequency of LV dysfunction was high in patients with preoperative EF,64% and LVESD 37 mm, intermediate in patients presenting one of these two features, and low in patients with EF 64% and LVESD,37 mm. The combined analysis suggests that best preservation of LV function is achieved when MVR is performed before

8 Predicting LV dysfunction after valve repair for MR due to leaflet prolapse 709 these thresholds are reached. However, some patients with normal preoperative LV size and EF may still develop postoperative LV dysfunction, illustrating the limits of classical preoperative echo measures of LV size and function to predict post-operative LV dysfunction. MVR is currently the procedure of choice for the treatment of organic MR. This approach is supported by the low-risk and excellent long-term results of valve repair. The sustained improvement of surgical techniques and post-operative intensive care has lead to a dramatic reduction of operative risk. Under these circumstances, an optimal post-operative result in terms of LV function and survival is closely related to prompt preoperative identification of subgroups of patients at high risk. The 64% and 37 mm thresholds for preoperative EF and, respectively, preoperative LVESD allow relatively good prediction of LV dysfunction after MVR and might be considered for clinical decision-making. This study was a retrospective analysis of patient record and therefore had the inherent limitations of such analyses. Quantitation of MR was not available in all patients and therefore not included in the multivariable analyses. The visual method for EF assessment used in some patients has probably insufficient precision considering the proposed cut-off value (64%) for preoperative EF. The present study included a significant number of symptomatic patients and therefore our findings cannot be extrapolated to suggest earlier surgery in asymptomatic patients with severe organic MR. From a clinical point of view, we acknowledge that there is an overlap of values and standard deviations of preoperative EF and preoperative LVESD between the two groups (with and without post-operative LV dysfunction), limiting the application of the proposed cut-off values in the case of the single patient. Finally, newer estimates of LV function (2D strain) were not available in this population and therefore, not investigated. Conclusion Our study demonstrates that simple preoperative echocardiography measures allow prediction of LV dysfunction after MVR for MR due to leaflet prolapse. Preoperative EF and preoperative LVESD are equally important and have additive value for the prediction of post-operative LV dysfunction. The lowest frequency of post-operative LV dysfunction was observed in patients with preoperative EF 64% and preoperative LVESD,37 mm. These findings stress out the importance of echocardiography for preoperative LV function quantification and need to be confirmed by further studies conducted in asymptomatic patients with severe organic MR. Conflict of interest: none declared. References 1. Bonow RO, Carabello BA, Chatterjee K, de Leon AC Jr, Faxon DP, Freed MD et al. AHA 2006 guidelines for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (writing Committee to Revise the 1998 guidelines for the management of patients with valvular heart disease) developed in collaboration with the Society of Cardiovascular Anesthesiologists endorsed by the Society for Cardiovascular Angiography and Interventions and the Society of Thoracic Surgeons. J Am Coll Cardiol 2006;48:e Vahanian A, Baumgartner H, Bax J, Butchart E, Dion R, Filippatos G et al. Guidelines on the management of valvular heart disease: The Task Force on the Management of Valvular Heart Disease of the European Society of Cardiology. Eur Heart J 2007;28: Mohty D, Orszulak TA, Schaff HV, Avierinos JF, Tajik JA, Enriquez-Sarano M. Very long-term survival and durability of mitral valve repair for mitral valve prolapse. Circulation 2001;104(12 Suppl. 1):I Braunberger E, Deloche A, Berrebi A, Abdallah F, Celestin JA, Meimoun P et al. Very long-term results (more than 20 years) of valve repair with Carpentier s techniques in nonrheumatic mitral valve insufficiency. Circulation 2001;104(12 Suppl. 1):I David TE, Ivanov J, Armstrong S, Rakowski H. Late outcomes of mitral valve repair for floppy valves: implications for asymptomatic patients. J Thorac Cardiovasc Surg 2003;125: Tribouilloy CM, Enriquez-Sarano M, Schaff HV, Orszulak TA, Bailey KR, Tajik AJ et al. Impact of preoperative symptoms on survival after surgical correction of organic mitral regurgitation: rationale for optimizing surgical indications. Circulation 1999;99: Enriquez-Sarano M, Tajik AJ, Schaff HV, Orszulak TA, Bailey KR, Frye RL. Echocardiographic prediction of survival after surgical correction of organic mitral regurgitation. Circulation 1994;90: Tribouilloy C, Grigioni F, Avierinos JF, Barbieri A, Rusinaru D, Szymanski C et al. Survival implication of left ventricular end-systolic diameter in mitral regurgitation due to flail leaflets a long-term follow-up multicenter study. J Am Coll Cardiol 2009; 54: Flemming MA, Oral H, Rothman ED, Briesmiester K, Petrusha JA, Starling MR. Echocardiographic markers for mitral valve surgery to preserve left ventricular performance in mitral regurgitation. Am Heart J 2000;140: Crawford MH, Souchek J, Oprian CA, Miller DC, Rahimtoola S, Giacomini JC et al. Determinants of survival and left ventricular performance after mitral valve replacement. Department of Veterans Affairs Cooperative Study on Valvular Heart Disease. Circulation 1990;81: Starling MR, Kirsh MM, Montgomery DG, Gross MD. Impaired left ventricular contractile function in patients with long-term mitral regurgitation and normal ejection fraction. J Am Coll Cardiol 1993;22: Enriquez-Sarano M, Schaff HV, Orszulak TA, Tajik AJ, Bailey KR, Frye RL. Valve repair improves the outcome of surgery for mitral regurgitation. A multivariate analysis. Circulation 1995;91: Enriquez-Sarano M, Tajik AJ, Schaff HV, Orszulak TA, McGoon MD, Bailey KR et al. Echocardiographic prediction of left ventricular function after correction of mitral regurgitation: results and clinical implications. J Am Coll Cardiol 1994;24: Matsumura T, Ohtaki E, Tanaka K, Misu K, Tobaru T, Asano R et al. Echocardiographic prediction of left ventricular dysfunction after mitral valve repair for mitral regurgitation as an indicator to decide the optimal timing of repair. J Am Coll Cardiol 2003;42: Suri RM, Schaff HV, Dearani JA, Sundt TM, Daly RC, Mullany CJ et al. Recovery of left ventricular function after surgical correction of mitral regurgitation caused by leaflet prolapse. J Thorac Cardiovasc Surg 2009;137: Wisenbaugh T, Skudicky D, Sareli P. Prediction of outcome after valve replacement for rheumatic mitral regurgitation in the era of chordal preservation. Circulation 1994;89: European System for Cardiac Operative Risk Evaluation. org ( February 2011). 18. Levine RA, Stathogiannis E, Newell JB, Harrigan P, Weyman AE. Reconsideration of echographic standards for mitral valve prolapse: lack of association between leaflet displacement isolated to the apical four-chamber view and independent echocardiographic evidence of abnormality. J Am Coll Cardiol 1988;11: Mintz GS, Kotler MN, Segal BL, Parry WR. Two-dimensional echocardiographic recognition of ruptured chordae tendineae. Circulation 1978;57: Quinones MA, Waggoner AD, Reduto LA, Nelson JG, Young JB, Winters WL Jr. et al. A new, simplified and accurate method for determining ejection fraction with two-dimensional echocardiography. Circulation 1981;64: Sahn D, DeMaria A, Kisslo J, Weyman A. Recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation 1978;58: Grigioni F, Tribouilloy C, Avierinos JF, Barbieri A, Ferlito M, Trojette F et al. Outcomes in mitral regurgitation due to flail leaflets. J Am Coll Cardiol Img 2008;1: Enriquez-Sarano M, Avierinos JF, Messika-Zeitoun D, Detaint D, Capps M, Nkomo V et al. Quantitative determinants of the outcome of asymptomatic mitral regurgitation. N Engl J Med 2005;352: Lancellotti P, Moura L, Pierard LA, Agricola E, Popescu BA, Tribouilloy C et al. European Association of Echocardiography recommendations for the assessment

9 710 C. Tribouilloy et al. of valvular regurgitation. Part 2: mitral and tricuspid regurgitation (native valve disease). Eur J Echocardiogr 2010;11: David T. Artificial chordae. Semin Thorac Cardiovasc Surg 2004;16: Hanley JA, McNeil BJ. A method of comparing the areas under receiving operating characteristic curves derived from some cases. Radiology 1983;148: Ling LH, Enriquez-Sarano M, Seward JB, Tajik AJ, Schaff HV, Bailey KR et al. Clinical outcome of mitral regurgitation due to flail leaflet. N Engl J Med 1996;335: Leung DY, Griffin BP, Stewart WJ, Cosgrove DM III, Thomas JD Marwick TH. Left ventricular function after valve repair for chronic mitral regurgitation: predictive value of preoperative assessment of contractile reserve by exercise echocardiography. J Am Coll Cardiol 1996;28: Lee EM, Shapiro LM, Wells FC. Superiority of mitral valve repair in surgery for degenerative mitral regurgitation. Eur Heart J 1997;18: IMAGE FOCUS doi: /ejechocard/jer062 Online publish-ahead-of-print 23 July Active infective prosthetic endocarditis after percutaneous edge-to-edge mitral valve repair Jens-Gerrit Kluge*, Andreas Hagendorff, Dietrich Pfeiffer, Daniel Jurisch, and Adrienn Tarr Division of Cardiology and Angiology, Department of Internal Medicine, Neurology und Dermatology, University Hospital Leipzig, Liebigstraße 20, Leipzig, Germany * Corresponding author. Tel: ; fax: , Jens-Gerrit.Kluge@medizin.uni-leipzig.de A 57-year-old man with ischaemic cardiomyopathy was admitted to our institution with fever. The patient had undergone percutaneous mitral valve repair abroad 2 weeks before admission. Transthoracic and transoesophageal echocardiography (TTE/TEE) were performed, showing a mobile lesion (red arrows) originating from the implanted MitraClip mitral valve repair system (*). Staphylococcus aureus was detected in all six serial blood cultures. Therefore, the diagnosis of infective endocarditis was definite. Antibiotic therapy (Vancomycin, Gentamicin, Rifampicin) was initiated immediately after echocardiography and obtaining blood cultures. Owing to clinical deterioration and increasing vegetation (initial size: 8 mm, size after 1 week: 12 mm), despite predicted high perioperative mortality (EuroScore: 18, mortality 79.3%), surgical treatment was considered. Mitral valve replacement using a biological prosthesis (St Jude Epic Mitral 31 mm) and tricuspid valve reconstruction were performed 8 days after admission. Intraoperative surgical and histological findings of the mitral valve and the attached MitraClip w confirmed the diagnosis of infective prosthetic endocarditis. Fortunately, no major postoperative complications occurred. Antibiotic therapy was continued for 6 weeks. Twelve weeks after mitral valve replacement, the patient was readmitted with fever. In echocardiography, an oscillating remnant of the lateral papillary muscle with no change in morphology compared with the post-operative status was seen. Nevertheless, because of renal failure due to suspected embolic renal infarction, relapse of infective endocarditis was assumed. Antibiotic therapy (Vancomycin, Gentamicin, Rifampicin, and Tazobactam) was again initiated for 6 weeks. In the following months, the patient had been hospitalized several times for chronic heart failure. Nine months after mitral valve replacement, endocarditis caused by S. aureus reoccurred. Biological prosthesis replacement using a biological prosthesis (St Jude Epic Mitral 29 mm) was performed. The patient is currently recuperating from surgery. In the early stages of healing after percutaneous mitral valve repair the MitraClipw is covered with platelet and fibrin deposition. Therefore, endocarditis caused by S. aureus might be more prone to occur. Because no other access point of the infection could be determined, it seems possible that the infection could have been caused iatrogenically. To our knowledge, this is the first case of active infective prosthetic endocarditis after percutaneous edge-to-edge mitral valve repair in humans. Published on behalf of the European Society of Cardiology. All rights reserved. & The Author For permissions please journals.permissions@oup.com