Predictors of post-operative cardiovascular events, focused on atrial fibrillation, after valve surgery for primary mitral regurgitation

Predictors of post-operative cardiovascular events, focused on atrial fibrillation, after valve... Abstract Aims Primary mitral regurgitation (PMR) can be considered as a heterogeneous clinical disease. The optimal timing of valve surgery for severe PMR remains unknown. To determine whether unbiased clustering analysis using dense phenotypic data (phenomapping) could identify phenotypically distinct PMR categories of patients. Methods and results One hundred and twenty-two patients who underwent surgery were analysed, excluding patients with pre-operative permanent atrial fibrillation (AF), were prospectively included before surgery. They were given an extensive echocardiographic evaluation before surgery, and clinical data were collected. These phenotypic variables were grouped in clusters using hierarchical clustering analysis. Then, different groups were created using a dedicated phenomapping algorithm. Post-operative outcomes were compared between the groups. The primary endpoint was post-operative cardiovascular events (PCE), defined as a composite of: deaths, AF, stroke, and rehospitalization. The secondary endpoint was post-operative AF. Data from three phenogroups with different characteristics and prognoses were identified. Phenogroup-1 (67 patients) was the reference group. Phenogroup-2 (33 patients) included intermediate-risk male and smoker patients with heart remodelling. Phenogroup-3 (22 patients) included older female patients with comorbidities (chronic renal failure, paroxysmal AF) and diastolic dysfunction. They had a higher risk of developing both PCE [(hazard ratio) HR = 3.57(1.72–7.44), P < 0.001] and post-operative AF [HR = 4.75(2.03–11.10), P < 0.001]. Pre-operative paroxysmal AF was identified as an independent risk factor for PCE. Conclusion Classification of PMR can be improved using statistical learning algorithms to define therapeutically homogeneous patient subclasses. High-risk patients can be identified, and these patients should be carefully monitored and may even be treated earlier. mitral regurgitation , machine learning , echocardiography , atrial fibrillation Introduction The optimal timing of mitral valve surgery for severe primary mitral regurgitation (PMR) is still being debated even with the most recent guidelines on the management of valvular heart disease.1,2 Although some risk factors of developing events, including post-operative atrial fibrillation (AF), have been identified, it remains difficult to predict which patients will develop post-operative AF and events before sending the patient to surgery. Therefore, we hypothesized that PMR is a heterogeneous clinical disease that affects variable patient populations and, thus, different phenotypes of patients; available predictive parameters have low positive predictive values.3,4 Machine learning, which can be defined as a process of using data to learn relationships between objects,5 was used in this study to determine the different phenotypes of severe PMR and to assign patients into these groups. Machine learning uses statistical learning algorithms to realize an unbiased hierarchical clustering analysis of phenotypic data. This statistical method is usually performed to analyse genetic data, but the approach was recently used to improve characterization of another heterogeneous cardiovascular syndrome.6 In this study, unbiased clustering analysis using dense phenotypic data, called ‘phenomapping’, resulted in the characterization of three groups of patients with different characteristics, defining a novel HFpEF-phenogroup classification. The aims of this study were (i) to identify different phenogroups of patients with severe PMR, analysing clinical, demographic, electrocardiographic, and echocardiographic data, (ii) to search predictors of post-operative cardiovascular events (PCE), particularly AF, and (iii) to analyse outcomes in each phenogroup to improve management of this heterogeneous population. Methods Data collection and patients From June 2007 to December 2015, data from 201 patients suffering from PMR due to leaflet prolapse, who underwent a pre-operative check-up in our institution, were prospectively collected. Demographic data, patients’ medical backgrounds, haemodynamic status, treatments, symptoms, operative data, and in-hospital outcomes were collected from patients’ medical records. All patients underwent pre-operative transthoracic 2D echocardiography (Vivid 7 or 9, GE Healthcare, Horten, Norway) with Doppler and tissue Doppler imaging. Echo recordings were retrospectively reanalysed, and measurements were performed by experienced blinded cardiologists (EchoPAC BT12, General Electric, Horten, Norway). Post-operative events were collected in hospitalization reports and by phone-calls with referent physicians. Among these 201 patients, 151 underwent surgery (mitral valve repair or replacement) at the end of the follow-up (December 2015). Patients with permanent pre-operative AF were excluded to account for confounders, as were patients presenting a non-severe PMR, a missing follow-up or extreme echocardiographic values (i.e. values above or below three deviations around the mean) (see Figure 1 and Supplementary Data online). Finally, 122 patients who underwent mitral valve surgery for severe PMR were included in this study. This study was conducted in accordance with institutional policies, national legal requirements, and the revised Declaration of Helsinki. The study was approved by an ethics committee (Person Protection Committee West V) (PIME 08/16-675). Figure 1 View largeDownload slide Flowchart. Figure 1 View largeDownload slide Flowchart. Definitions In this study, the PMR aetiology was mitral valve prolapse (Barlow disease and degenerative mitral regurgitation), Type 2 of Carpentier’s classification, defined according to the recommended criteria.7 Rheumatic heart diseases, restrictive lesions (due to drugs or radiation therapy), and infective endocarditis were not included. Quantification of mitral regurgitation was assessed according to recommendations.8 Ventricular functions and volume measurements were also based on the recommendations.9 Global longitudinal strain (GLS) was measured from the three apical views using EchoPAC. Paroxysmal AF was defined according to European Society of Cardiology (ESC) guidelines.10 Renal failure was defined as an estimated glomerular filtration rate <60 mL/min. The standard logistic European System for Cardiac Operative Risk Evaluation (EuroSCOREs 1 and 2) were calculated (www.euroscore.org). Endpoints After identification of three phenogroups (details below), post-operative outcomes were compared between groups: post-operative immediate AF (occurrence before or at 30 days), post-operative long-term AF (after 30 days), all-cause mortality, cardiovascular mortality, stroke, and cardiovascular cause of hospitalization. The primary endpoint was survival free of cardiovascular events. The secondary endpoint was survival free of post-operative long-term AF. Statistical analysis Phenotypic domains and clustering of variables The phenotypic domains consisted of 64 variables, including clinical and demographic data, patients’ medical background (AF history, renal failure, and cardiovascular risk factors), EuroSCORE, treatments, symptoms, and echocardiographic parameters (indexed variables) (Supplementary Data online). As described above, we excluded patients presenting extreme values, defined as values below or above three standard deviations around the mean. In case of asymmetric distribution, we performed either log, square-root, inverse, or power transformation to improve normality of variables, which was graphically checked. We used the missForest algorithm11 to impute missing data before the clustering analysis. To remove redundancies, we performed ascending hierarchical variable clustering using the ClustVarLV R package.12 This algorithm is based on the aggregation of variables around latent components, with the capability to take into account the direction of correlations (i.e. position or negative association) between variables of mixed types. Clustering of patients The next step consisted of identifying clusters of similar patients from the previous latent components, which provided summarized information of the original variable. We used hierarchical clustering with the dissimilarity matrix given by Euclidean distance and the Ward’s minimum variance method for aggregation. The final clusters were determined by consensus across a set of criteria used to select the optimal number of clusters.13 Comparison of clinical characteristics and survival among phenogroups We compared the clinical, demographic, electrocardiographic, and echocardiographic characteristics between clusters using the Kruskal–Wallis test for continuous variables and the χ2 test (or Fisher’s exact test when appropriate) for categorical variables. Statistical significance was considered as a two-sided P-value <0.05. In this case, we also computed pairwise comparisons using the Conover test for continuous variables and the Fisher’s exact test for qualitative variables. P-values were adjusted with the Bonferroni correction. We used Cox regression models to calculate between-group differences in PCEs and post-operative AF. Group 1 was considered as the reference group for survival analysis. We analysed Schoenfeld residuals to test the assumption of proportional hazards and used the Kaplan–Meier method to calculate survival curves. We used R statistical software, version 3.3.3, for all analyses.14 Results Patients and description of phenogroups Performing ascending hierarchical variable clustering, the 64 phenotypic variables were grouped in six clusters of variables (Supplementary Data online). Then, we identified three groups of patients with the biclustering procedure. The baseline characteristics of patients are depicted in Table 1. Data from 122 patients (among 151 operated, Figure 1) were analysed (median age 63 years old; 68% male). Mitral valve repair was performed in 105 (86%) patients, whereas 17 patients received mitral valve replacements (14%). Table 1 Baseline patient characteristics Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Quantitative data are expressed as means and standard deviations. Categorical variables are expressed as numbers (%). AF, atrial fibrillation; BMI, body mass index; BSA, body surface area; CABG, coronary artery bypass graft; GFR, glomerular filtration rate; NYHA, New York Heart Association. Table 1 Baseline patient characteristics Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Quantitative data are expressed as means and standard deviations. Categorical variables are expressed as numbers (%). AF, atrial fibrillation; BMI, body mass index; BSA, body surface area; CABG, coronary artery bypass graft; GFR, glomerular filtration rate; NYHA, New York Heart Association. Phenogroup-1 included 67 (median age 61 years old; 70% male) patients. Phenogroup-2 included 33 (median age 61 years old; 94% male) patients, whereas phenogroup-3 included 22 (median age 73 years old; 23% male) patients. The baseline characteristics were imbalanced between groups. Phenogroups-1 and 2 were predominantly male (70% and 94% of patients, respectively), whereas phenogroup-3 was mostly composed of women (77% of patients, P < 0.001) (Table 1 and Supplementary data online, Table S1). Patients in Group 3 were older than in the others (73 ± 5 years old vs. 61 ± 10 years old in Group 2 and 61 ± 11 in Group 1; P < 0.0001 after pairwise comparison). Regarding cardiovascular risk factors (besides age and overweight, mentioned above), patients in phenogroup-3 were more likely to have high blood pressure than those of the first phenogroup: 64% of patients in Group 3 had hypertension vs. 29% in Group 1 (P = 0.015) (Supplementary data online, Table S1). Patients in Group 3 were also more likely to have diabetes mellitus (Table 1). Regarding comorbidities, chronic obstructive pulmonary disease (COPD) was more frequent in Group 2 (see Supplementary data online, Table S1). Before surgery, the predicted operative mortality (estimated using EuroSCORE) in phenogroup-3 was higher than in other groups: the EuroSCORE 1 was 8.8% in Group 3 vs. 4.2% in Group 2 (P < 0.0001) and 3.3% in the Group 1 (P < 0.0001). The EuroSCORE 2 was 2.3% in Group 3 vs. 1.1% in Group 2 (P < 0.0001) and 1.0% in the Group 1 (P < 0.0001). Patients in Group 3 were more likely to have pre-operative paroxysmal AF: 45% of patients in Group 3 had a history of AF vs. 18% in Group 2 and only 13% in the Group 1 (P = 0.009) (Table 1). Pre-operative echocardiographic characteristics Regarding left ventricular (LV) systolic function: LV ejection fraction (LVEF) was comparable between groups, with a mean of 67 ± 7% (P = 0.127) (Table 2). There was a trend towards a lower GLS in phenogroup-2, but this difference was weakly significant: GLS was −19 ± 3.1% in Group 2 vs. −20 ± 3.0% in Group 3 and −21 ± 3.1% in the Group 1, P = 0.049 (see Supplementary data online, Table S2). In phenogroup-2, the left ventricle was more dilated than in the Group 1. Table 2 Pre-operative echocardiographic data Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Quantitative data are expressed as means and standard deviations. EROA, effective regurgitant orifice area; GLS, global longitudinal strain; LA, left atrial; LV, left ventricular; LVEDD, left ventricular end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; LVESV, left ventricular end-systolic volume; PASP, pulmonary artery systolic pressure; RA, right atrial; RV, right ventricular; TAPSE, tricuspid annular plane systolic excursion; VTI, velocity time integral. Table 2 Pre-operative echocardiographic data Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Quantitative data are expressed as means and standard deviations. EROA, effective regurgitant orifice area; GLS, global longitudinal strain; LA, left atrial; LV, left ventricular; LVEDD, left ventricular end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; LVESV, left ventricular end-systolic volume; PASP, pulmonary artery systolic pressure; RA, right atrial; RV, right ventricular; TAPSE, tricuspid annular plane systolic excursion; VTI, velocity time integral. The left atrium (LA) was larger in Group 2: the indexed LA volume was 66 ± 25 mL/m2 (which represent a severe LA dilatation) vs. 58 ± 19 mL/m2 in Group 3 and 44 ± 19 mL/m2 in the Group 1 (P < 0.001). The pairwise comparison highlighted LA enlargements in both Groups 2 and 3 compared with the Group 1, but there was no longer any statistically significant difference between Groups 2 and 3 (see Supplementary data online, Table S2). The right atrium was also enlarged in Group 2 (Table 2). Differences between groups regarding quantitative evaluation of mitral regurgitation were observed. Indeed, a larger effective regurgitant orifice area (EROA) and a larger regurgitation volume were assessed in Group 2: the average EROA was 69 mm2 vs. 55 mm2 in Group 3 and 56 mm2 in Group 1 (P = 0.010) (Table 2). Regarding the right heart, pulmonary artery systolic pressure was estimated at ∼48 ± 13 mmHg in Group 3 vs. ∼42 ± 17 mmHg in Group 2 and ∼35 ± 9.7 mmHg in the Group 1 (P < 0.001) (Table 2). The right ventricle (RV) was more dilated in Group 2 (RV end-diastolic area = 22 ± 3.6 cm2 vs. 17 ± 3.1 cm2 in Group 3 and 15 ± 2.7 cm2 in the Group 1; P < 0.001). Post-operative outcomes and endpoints Regarding the primary endpoint (Table 3 and Figure 2) compared with patients in the Group 1, patients in Group 3 were more likely to develop PCE. Table 3 Association of phenogroups with post-operative outcomes on a univariate Cox model: hazard ratio with (95% confidence interval) and P-value Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Table 3 Association of phenogroups with post-operative outcomes on a univariate Cox model: hazard ratio with (95% confidence interval) and P-value Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Figure 2 View largeDownload slide Survival free of cardiovascular events, stratified by phenogroup. Figure 2 View largeDownload slide Survival free of cardiovascular events, stratified by phenogroup. Post-operative outcomes are detailed in Table 4. Regarding the secondary endpoint, 29 (24%) patients have developed post-operative long-term AF in the overall population. Post-operative long-term AF occurred significantly more often in phenogroup-3 than in phenogroup-1. There was no statistical difference between the Groups 1 and 2 or between the Groups 2 and 3 (Table 4 and Supplementary data online, Table S3). The risk of developing post-operative long-term AF was significantly increased in Group 3 (Table 3 and Figure 3). Table 4 Post-operative outcomes Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Categorical variables are expressed as numbers (%). Table 4 Post-operative outcomes Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Categorical variables are expressed as numbers (%). Figure 3 View largeDownload slide Survival free of post-operative long-term AF, stratified by phenogroup. Figure 3 View largeDownload slide Survival free of post-operative long-term AF, stratified by phenogroup. Regarding mortality, only four patients died during the follow-up, including two patients in Groups 1 and 2 patients in Group 3. There was no difference between groups (P = 0.153). The incidence of cardiovascular causes of hospitalization was significantly higher in phenogroup-2 compared with the first group. Other between-groups differences were not statistically significant. Paroxysmal pre-operative AF was confirmed as a risk factor of developing both PCE and post-operative AF (Supplementary data online, Figures S1 and S2). The risk of developing post-operative AF (including both immediate and long-term AF) was significantly increased in patients with pre-operative AF: hazard ratio = 3.19(1.49–6.82) (P = 0.00284). Discussion In this study, using machine learning and hierarchical clustering analysis applied on dense phenotypic data, the main findings were as follows (i) PMR is a heterogeneous disease, (ii) using a phenomapping dedicated algorithm, three phenogroups of patients were identified, despite PMR heterogeneity, (iii) these phenogroups had different prognoses, suggesting different risk profiles, and (iv) pre-operative paroxysmal AF is an important risk factor of PCE. A previous study, ‘Phenomapping for Novel Classification of Heart Failure With Preserved Ejection Fraction’,6 provided a novel classification of a cardiovascular disorder using phenomapping and was ‘the first study that applies machine learning techniques to resolve heterogeneity in a cardiovascular syndrome using dense phenotypic data’. Using a similar algorithm, our study demonstrates that it is also possible to identify different phenogroups in PMR disease. Indeed, all patients had criteria for severe PMR, but phenomapping showed that PMR is a heterogeneous disease. A novel classification of PMR patients was defined using correlations between phenotypic variables to find patterns in dense data. Once the three groups were identified, the differences between them were striking regarding clinical and echocardiographic characteristics. The three phenogroups represented three archetypes of PMR according to their demographic, clinical, and echocardiographic characteristics: Phenogroup-1 represented low-risk patients (less risk factors, less dilatation of the ventricle…) but their main characteristics did not differ from those of the overall population. Phenogroup-2, which represented intermediate-risk patients, included predominantly male patients, aged approximately 61 years old, with large body surface areas, who were smokers and with histories of COPD. Their heart remodelling appeared to be more pronounced. Phenogroup-3, which represented high-risk patients, included predominantly women who were older, thinner, and smaller, with hypertension and comorbidities such as renal failure or a pre-operative history of AF. Their EuroSCORE was higher than those of the other groups. Few studies have explored gender influence on valvular disease. However, women seemed to have worse outcomes than men after mitral surgery,15 although this difference can be partially driven by the rate of mitral valve replacement in women. The heart remodelling in phenogroup-3 appeared to be surprisingly less marked than that in Group 2. Regarding echocardiographic characteristics: LA peak systolic strain is confirmed as a predictor of cardiovascular events in mitral regurgitation16 and is associated with elevated filling pressures17. E-wave velocity was also the highest in Group 3 (150 ± 44 cm/s): it may be explained by both an important mitral regurgitation (increasing stroke volume) and a more severe diastolic dysfunction in these patients, as described previously.18 LV diastolic dysfunction in PMR is known to result in poor outcomes19 and seemed to predict better cardiovascular outcomes than LVEF (mean pre-operative LVEF was 67%, with no difference between groups). LVEF may remain in the normal range for a long period of time, whereas subclinical myocardial LV dysfunction may be present.20 Therefore, it is necessary to look for other echocardiographic predictors of outcomes, such as increased LA volume,21 abnormal LV GLS,4 pulmonary hypertension,22 increased E/e’ ratio,23 or abnormal LA peak systolic strain.16 Most of these parameters were altered in phenogroup-3 (except LV GLS). Of note, regurgitant volume and EROA were not good predictors of post-operative outcomes. Phenogroup-2 had global cardiac chamber enlargement (LA and right atrial volumes, LV volume, and RV size) without obvious diastolic dysfunction or even elevated filling pressures. Although cardiac chamber enlargement suggests myocardial remodelling, it did not seem to result in an increased risk of PCE. The enlargement in the right chambers and the lowest RV global strain in phenogroup-2 can also be explained by the important number of patients with COPD (36% of patients had a history of COPD in phenogroup-2).24 The risk of developing PCE was not significantly increased in phenogroup-2. We hypothesized that phenogroup-2 likely included intermediate-risk patients. Pre-operative AF is known as an important predictor of cardiovascular outcomes,25 which was confirmed in this study, where 45% of phenogroup-3 patients, who presented the highest risk of PCE occurrence, had pre-operative paroxysmal AF. The high incidences of pre-operative AF in this group can be explained by several factors. LA dilatation is considered as an important risk factor for the occurrence of AF,21 and indexed LA volumes in phenogroup-3 (mean 58 ± 19 mL/m2) were close to the cut-off of 60 mL/m2.21 Nevertheless, left atrium was also dilated in Group 2, although these patients were less likely to have post-operative long-term AF than those in Group 3. Therefore, an enlarged LA is not sufficient to explain the highest prevalence of AF in Group 3. Phenogroup-3 patients were older10 and were also more likely to have hypertension and chronic renal failure.10 LA peak systolic strain was reduced in phenogroup-3, reflecting an LA dysfunction (reduced reservoir function).26 Finally, pre-operative AF was a predictor of post-operative AF (Supplementary data online, Figure S2).27 Perspectives Determining optimal timing for surgery in severe PMR is still problematic, and is currently based on symptoms, severity of the regurgitation and its impact on LV volume and systolic function. Predictors of outcome currently considered in regurgitant valvular diseases are LV increased volumes and reduced LVEF, indicating an alteration of myocardial contractility. However, reduced LVEF is often a late consequence of valve dysfunction and may even imply irreversible myocardial injury. Machine learning identified three phenogroups with different prognoses. Therefore, the management of patients with severe PMR could be improved, as high-risk patients can now be identified earlier. We suggest that high-risk patients should be carefully monitored (i.e. by more frequent visits to their usual cardiologist). Sinus rhythm should be maintained as long as possible, and paroxysmal AF can be detected by Holter monitoring. Pulmonary vein isolation could be performed during valve surgery in patients with pre-operative paroxysmal AF to prevent post-operative AF.10 Mitral surgery may be performed early in (asymptomatic) phenogroup-3 individuals to prevent the AF-occurrence risk and can be encouraged by early detection of the cardiac consequences of PMR by performing exercise echocardiography.28 Study limitations We acknowledge several limitations. Hierarchical clustering led to an imbalanced number of patients between groups. This imbalance impeded definite conclusions about the value of phenomapping in PMR and resulted in a lack of statistical power, particularly because of the small sample size. The retrospective design of our study also implies that we had to perform an imputation procedure to deal with missing values and to exclude several patients having extreme values due to mistyping in electronic health records. Therefore, our results should be considered as hypothesis generating and interpreted with caution. To confirm the reproducibility of our results (i.e. the identification of phenogroups, outcomes, and risk-levels), we will have to further control our results with a validation cohort. Conclusion In this study, PMR was confirmed as a heterogeneous clinical disease, but this heterogeneity may be resolved using a dedicated algorithm to perform phenomapping analysis. Phenotypic data can be grouped in clusters using machine learning to identify three phenogroups of patients with different prognoses. Supplementary data Supplementary data are available at European Heart Journal - Cardiovascular Imaging online. Acknowledgements The authors thank the study nurses from the CIC-IT1414 but also the CORECT programmes and the scientific committee of the CHU-Rennes for their support. Conflict of interest: None declared. References 1 Baumgartner H , Falk V , Bax JJ , De Bonis M , Hamm C , Holm PJ et al. 2017 ESC/EACTS Guidelines for the management of valvular heart disease: the Task Force for the Management of Valvular Heart Disease of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS) . Eur Heart J 2017 ; 38 : 2739 – 91 . Google Scholar CrossRef Search ADS PubMed 2 Nishimura RA , Otto CM , Bonow RO , Carabello BA , Erwin JP , Fleisher LA et al. 2017 AHA/ACC Focused Update of the 2014 AHA/ACC guideline for the management of patients with valvular heart disease: a Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines . J Am Coll Cardiol 2017 ; 70 : 252 – 89 . Google Scholar CrossRef Search ADS PubMed 3 Tribouilloy C , Rusinaru D , Szymanski C , Mezghani S , Fournier A , Lévy F et al. 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 . Eur J Echocardiogr 2011 ; 12 : 702 – 10 . Google Scholar CrossRef Search ADS PubMed 4 Mascle S , Schnell F , Thebault C , Corbineau H , Laurent M , Hamonic S et al. Predictive value of global longitudinal strain in a surgical population of organic mitral regurgitation . J Am Soc Echocardiogr 2012 ; 25 : 766 – 72 . Google Scholar CrossRef Search ADS PubMed 5 Hastie T , Tibshirani R , Friedman J. Elements of Statistical Learning: Data Mining, Inference, and Prediction , 2nd ed . Springer Ed ; 2009 . https://statweb.stanford.edu/∼tibs/ElemStatLearn/ (12 September 2017, date last accessed). 6 Shah SJ , Katz DH , Selvaraj S , Burke MA , Yancy CW , Gheorghiade M et al. Phenomapping for novel classification of heart failure with preserved ejection fraction . Circulation 2015 ; 131 : 269 – 79 . Google Scholar CrossRef Search ADS PubMed 7 Zoghbi WA , Adams D , Bonow RO , Enriquez-Sarano M , Foster E , Grayburn PA et al. Recommendations for noninvasive evaluation of native valvular regurgitation: a report from the American Society of Echocardiography Developed in Collaboration with the Society for Cardiovascular Magnetic Resonance . J Am Soc Echocardiogr 2017 ; 30 : 303 – 71 . Google Scholar CrossRef Search ADS PubMed 8 Lancellotti P , Tribouilloy C , Hagendorff A , Popescu BA , Edvardsen T , Pierard LA et al. ; Scientific Document Committee of the European Association of Cardiovascular Imaging . Recommendations for the echocardiographic assessment of native valvular regurgitation: an executive summary from the European Association of Cardiovascular Imaging . Eur Heart J Cardiovasc Imaging 2013 ; 14 : 611 – 44 . Google Scholar CrossRef Search ADS PubMed 9 Lang RM , Badano LP , Mor-Avi V , Afilalo J , Armstrong A , Ernande L et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging . Eur Heart J Cardiovasc Imaging 2015 ; 16 : 233 – 71 . Google Scholar CrossRef Search ADS PubMed 10 Kirchhof P , Benussi S , Kotecha D , Ahlsson A , Atar D , Casadei B et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS . Eur Heart J 2016 ; 37 : 2893 – 962 . Google Scholar CrossRef Search ADS PubMed 11 Stekhoven DJ , Bühlmann P. MissForest—non-parametric missing value imputation for mixed-type data . Bioinformatics 2012 ; 28 : 112 – 8 . Google Scholar CrossRef Search ADS PubMed 12 ClustVarLV: Clustering of Variables Around Latent Variables Version 1.5.1 from CRAN. https://rdrr.io/cran/ClustVarLV (12 September 2017, date last accessed). 13 Charrad. NbClust: An R Package for determining the relevant number of clusters in a data set. J Stat Softwe. https://www.jstatsoft.org/article/view/v061i06 (12 September 2017, date last accessed). 14 R Core Team . R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2013. https://www.r-project.org (12 September 2017, date last accessed). 15 Seeburger J , Eifert S , Pfannmüller B , Garbade J , Vollroth M , Misfeld M et al. Gender differences in mitral valve surgery . Thorac Cardiovasc Surg 2013 ; 61 : 42 – 6 . Google Scholar PubMed 16 Debonnaire P , Leong DP , Witkowski TG , Al Amri I , Joyce E , Katsanos S et al. Left atrial function by two-dimensional speckle-tracking echocardiography in patients with severe organic mitral regurgitation: association with guidelines-based surgical indication and postoperative (long-term) survival . J Am Soc Echocardiogr 2013 ; 26 : 1053 – 62 . Google Scholar CrossRef Search ADS PubMed 17 Wakami K , Ohte N , Asada K , Fukuta H , Goto T , Mukai S et al. Correlation between left ventricular end-diastolic pressure and peak left atrial wall strain during left ventricular systole . J Am Soc Echocardiogr 2009 ; 22 : 847 – 51 . Google Scholar CrossRef Search ADS PubMed 18 Nishimura RA , Tajik AJ. Evaluation of diastolic filling of left ventricle in health and disease: Doppler echocardiography is the clinician’s Rosetta Stone . J Am Coll Cardiol 1997 ; 30 : 8 – 18 . Google Scholar CrossRef Search ADS PubMed 19 Magne J , Mahjoub H , Pierard LA , O'Connor K , Pirlet C , Pibarot P et al. Prognostic importance of brain natriuretic peptide and left ventricular longitudinal function in asymptomatic degenerative mitral regurgitation . Heart 2012 ; 98 : 584 – 91 . Google Scholar CrossRef Search ADS PubMed 20 Magne J , Szymanski C , Fournier A , Malaquin D , Avierinos JF , Tribouilloy C. Clinical and prognostic impact of a new left ventricular ejection index in primary mitral regurgitation because of mitral valve prolapse . Circ Cardiovasc Imaging 2015 ; 8 : e003036 . Google Scholar CrossRef Search ADS PubMed 21 Le Tourneau T , Messika-Zeitoun D , Russo A , Detaint D , Topilsky Y , Mahoney DW et al. Impact of left atrial volume on clinical outcome in organic mitral regurgitation . J Am Coll Cardiol 2010 ; 56 : 570 – 8 . Google Scholar CrossRef Search ADS PubMed 22 Barbieri A , Bursi F , Grigioni F , Tribouilloy C , Avierinos JF , Michelena HI et al. ; Mitral Regurgitation International DAtabase (MIDA) Investigators . Prognostic and therapeutic implications of pulmonary hypertension complicating degenerative mitral regurgitation due to flail leaflet: a multicenter long-term international study . Eur Heart J 2011 ; 32 : 751 – 9 . Google Scholar CrossRef Search ADS PubMed 23 Agricola E , Galderisi M , Oppizzi M , Melisurgo G , Airoldi F , Margonato A. Doppler tissue imaging: a reliable method for estimation of left ventricular filling pressure in patients with mitral regurgitation . Am Heart J 2005 ; 150 : 610 – 5 . Google Scholar CrossRef Search ADS PubMed 24 Hilde JM , Skjørten I , Grøtta OJ , Hansteen V , Melsom MN , Hisdal J et al. Right ventricular dysfunction and remodeling in chronic obstructive pulmonary disease without pulmonary hypertension . J Am Coll Cardiol 2013 ; 62 : 1103 – 11 . Google Scholar CrossRef Search ADS PubMed 25 Grigioni F , Avierinos J-F , Ling LH , Scott CG , Bailey KR , Tajik AJ et al. Atrial fibrillation complicating the course of degenerative mitral regurgitation: determinants and long-term outcome . J Am Coll Cardiol 2002 ; 40 : 84 – 92 . Google Scholar CrossRef Search ADS PubMed 26 Kuppahally SS , Akoum N , Burgon NS , Badger TJ , Kholmovski EG , Vijayakumar S et al. Left atrial strain and strain rate in patients with paroxysmal and persistent atrial fibrillation: relationship to left atrial structural remodeling detected by delayed-enhancement MRI . Circ Cardiovasc Imaging 2010 ; 3 : 231 – 9 . Google Scholar CrossRef Search ADS PubMed 27 Raiten JM , Ghadimi K , Augoustides JGT , Ramakrishna H , Patel PA , Weiss SJ et al. Atrial fibrillation after cardiac surgery: clinical update on mechanisms and prophylactic strategies . J Cardiothorac Vasc Anesth 2015 ; 29 : 806 – 16 . Google Scholar CrossRef Search ADS PubMed 28 Naji P , Griffin BP , Asfahan F , Barr T , Rodriguez LL , Grimm R et al. Predictors of long-term outcomes in patients with significant myxomatous mitral regurgitation undergoing exercise echocardiography . Circulation 2014 ; 129 : 1310 – 9 . Google Scholar CrossRef Search ADS PubMed Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Heart Journal – Cardiovascular Imaging Oxford University Press

Predictors of post-operative cardiovascular events, focused on atrial fibrillation, after valve surgery for primary mitral regurgitation

Loading next page...
 
/lp/ou_press/predictors-of-post-operative-cardiovascular-events-focused-on-atrial-SW97E0JlCm
Publisher
Oxford University Press
Copyright
Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com.
ISSN
2047-2404
D.O.I.
10.1093/ehjci/jey049
Publisher site
See Article on Publisher Site

Abstract

Abstract Aims Primary mitral regurgitation (PMR) can be considered as a heterogeneous clinical disease. The optimal timing of valve surgery for severe PMR remains unknown. To determine whether unbiased clustering analysis using dense phenotypic data (phenomapping) could identify phenotypically distinct PMR categories of patients. Methods and results One hundred and twenty-two patients who underwent surgery were analysed, excluding patients with pre-operative permanent atrial fibrillation (AF), were prospectively included before surgery. They were given an extensive echocardiographic evaluation before surgery, and clinical data were collected. These phenotypic variables were grouped in clusters using hierarchical clustering analysis. Then, different groups were created using a dedicated phenomapping algorithm. Post-operative outcomes were compared between the groups. The primary endpoint was post-operative cardiovascular events (PCE), defined as a composite of: deaths, AF, stroke, and rehospitalization. The secondary endpoint was post-operative AF. Data from three phenogroups with different characteristics and prognoses were identified. Phenogroup-1 (67 patients) was the reference group. Phenogroup-2 (33 patients) included intermediate-risk male and smoker patients with heart remodelling. Phenogroup-3 (22 patients) included older female patients with comorbidities (chronic renal failure, paroxysmal AF) and diastolic dysfunction. They had a higher risk of developing both PCE [(hazard ratio) HR = 3.57(1.72–7.44), P < 0.001] and post-operative AF [HR = 4.75(2.03–11.10), P < 0.001]. Pre-operative paroxysmal AF was identified as an independent risk factor for PCE. Conclusion Classification of PMR can be improved using statistical learning algorithms to define therapeutically homogeneous patient subclasses. High-risk patients can be identified, and these patients should be carefully monitored and may even be treated earlier. mitral regurgitation , machine learning , echocardiography , atrial fibrillation Introduction The optimal timing of mitral valve surgery for severe primary mitral regurgitation (PMR) is still being debated even with the most recent guidelines on the management of valvular heart disease.1,2 Although some risk factors of developing events, including post-operative atrial fibrillation (AF), have been identified, it remains difficult to predict which patients will develop post-operative AF and events before sending the patient to surgery. Therefore, we hypothesized that PMR is a heterogeneous clinical disease that affects variable patient populations and, thus, different phenotypes of patients; available predictive parameters have low positive predictive values.3,4 Machine learning, which can be defined as a process of using data to learn relationships between objects,5 was used in this study to determine the different phenotypes of severe PMR and to assign patients into these groups. Machine learning uses statistical learning algorithms to realize an unbiased hierarchical clustering analysis of phenotypic data. This statistical method is usually performed to analyse genetic data, but the approach was recently used to improve characterization of another heterogeneous cardiovascular syndrome.6 In this study, unbiased clustering analysis using dense phenotypic data, called ‘phenomapping’, resulted in the characterization of three groups of patients with different characteristics, defining a novel HFpEF-phenogroup classification. The aims of this study were (i) to identify different phenogroups of patients with severe PMR, analysing clinical, demographic, electrocardiographic, and echocardiographic data, (ii) to search predictors of post-operative cardiovascular events (PCE), particularly AF, and (iii) to analyse outcomes in each phenogroup to improve management of this heterogeneous population. Methods Data collection and patients From June 2007 to December 2015, data from 201 patients suffering from PMR due to leaflet prolapse, who underwent a pre-operative check-up in our institution, were prospectively collected. Demographic data, patients’ medical backgrounds, haemodynamic status, treatments, symptoms, operative data, and in-hospital outcomes were collected from patients’ medical records. All patients underwent pre-operative transthoracic 2D echocardiography (Vivid 7 or 9, GE Healthcare, Horten, Norway) with Doppler and tissue Doppler imaging. Echo recordings were retrospectively reanalysed, and measurements were performed by experienced blinded cardiologists (EchoPAC BT12, General Electric, Horten, Norway). Post-operative events were collected in hospitalization reports and by phone-calls with referent physicians. Among these 201 patients, 151 underwent surgery (mitral valve repair or replacement) at the end of the follow-up (December 2015). Patients with permanent pre-operative AF were excluded to account for confounders, as were patients presenting a non-severe PMR, a missing follow-up or extreme echocardiographic values (i.e. values above or below three deviations around the mean) (see Figure 1 and Supplementary Data online). Finally, 122 patients who underwent mitral valve surgery for severe PMR were included in this study. This study was conducted in accordance with institutional policies, national legal requirements, and the revised Declaration of Helsinki. The study was approved by an ethics committee (Person Protection Committee West V) (PIME 08/16-675). Figure 1 View largeDownload slide Flowchart. Figure 1 View largeDownload slide Flowchart. Definitions In this study, the PMR aetiology was mitral valve prolapse (Barlow disease and degenerative mitral regurgitation), Type 2 of Carpentier’s classification, defined according to the recommended criteria.7 Rheumatic heart diseases, restrictive lesions (due to drugs or radiation therapy), and infective endocarditis were not included. Quantification of mitral regurgitation was assessed according to recommendations.8 Ventricular functions and volume measurements were also based on the recommendations.9 Global longitudinal strain (GLS) was measured from the three apical views using EchoPAC. Paroxysmal AF was defined according to European Society of Cardiology (ESC) guidelines.10 Renal failure was defined as an estimated glomerular filtration rate <60 mL/min. The standard logistic European System for Cardiac Operative Risk Evaluation (EuroSCOREs 1 and 2) were calculated (www.euroscore.org). Endpoints After identification of three phenogroups (details below), post-operative outcomes were compared between groups: post-operative immediate AF (occurrence before or at 30 days), post-operative long-term AF (after 30 days), all-cause mortality, cardiovascular mortality, stroke, and cardiovascular cause of hospitalization. The primary endpoint was survival free of cardiovascular events. The secondary endpoint was survival free of post-operative long-term AF. Statistical analysis Phenotypic domains and clustering of variables The phenotypic domains consisted of 64 variables, including clinical and demographic data, patients’ medical background (AF history, renal failure, and cardiovascular risk factors), EuroSCORE, treatments, symptoms, and echocardiographic parameters (indexed variables) (Supplementary Data online). As described above, we excluded patients presenting extreme values, defined as values below or above three standard deviations around the mean. In case of asymmetric distribution, we performed either log, square-root, inverse, or power transformation to improve normality of variables, which was graphically checked. We used the missForest algorithm11 to impute missing data before the clustering analysis. To remove redundancies, we performed ascending hierarchical variable clustering using the ClustVarLV R package.12 This algorithm is based on the aggregation of variables around latent components, with the capability to take into account the direction of correlations (i.e. position or negative association) between variables of mixed types. Clustering of patients The next step consisted of identifying clusters of similar patients from the previous latent components, which provided summarized information of the original variable. We used hierarchical clustering with the dissimilarity matrix given by Euclidean distance and the Ward’s minimum variance method for aggregation. The final clusters were determined by consensus across a set of criteria used to select the optimal number of clusters.13 Comparison of clinical characteristics and survival among phenogroups We compared the clinical, demographic, electrocardiographic, and echocardiographic characteristics between clusters using the Kruskal–Wallis test for continuous variables and the χ2 test (or Fisher’s exact test when appropriate) for categorical variables. Statistical significance was considered as a two-sided P-value <0.05. In this case, we also computed pairwise comparisons using the Conover test for continuous variables and the Fisher’s exact test for qualitative variables. P-values were adjusted with the Bonferroni correction. We used Cox regression models to calculate between-group differences in PCEs and post-operative AF. Group 1 was considered as the reference group for survival analysis. We analysed Schoenfeld residuals to test the assumption of proportional hazards and used the Kaplan–Meier method to calculate survival curves. We used R statistical software, version 3.3.3, for all analyses.14 Results Patients and description of phenogroups Performing ascending hierarchical variable clustering, the 64 phenotypic variables were grouped in six clusters of variables (Supplementary Data online). Then, we identified three groups of patients with the biclustering procedure. The baseline characteristics of patients are depicted in Table 1. Data from 122 patients (among 151 operated, Figure 1) were analysed (median age 63 years old; 68% male). Mitral valve repair was performed in 105 (86%) patients, whereas 17 patients received mitral valve replacements (14%). Table 1 Baseline patient characteristics Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Quantitative data are expressed as means and standard deviations. Categorical variables are expressed as numbers (%). AF, atrial fibrillation; BMI, body mass index; BSA, body surface area; CABG, coronary artery bypass graft; GFR, glomerular filtration rate; NYHA, New York Heart Association. Table 1 Baseline patient characteristics Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Overall Group 1 Group 2 Group 3 (n = 122) (n = 67) (n = 33) (n = 22) P-value Age (years) 63 ± 11 61 ± 11 61 ± 10 73 ± 5 <0.001 Men 83 (68) 47 (70) 31 (94) 5 (23) <0.001 Heart rate (bpm) 71 ± 12 72 ± 13 72 ± 12 67 ± 8.8 0.212 Systolic blood pressure (mmHg) 140 ± 22 140 ± 20 130 ± 18 160 ± 25 0.025 Diastolic blood pressure (mmHg) 81 ± 11 80 ± 9 80 ± 12 86 ± 11 0.171 Hypertension 49 (40) 19 (29) 16 (48) 14 (64) 0.009 Height (cm) 170 ± 10 170 ± 9.3 170 ± 9.8 160 ± 8.4 <0.001 BSA (m2) 1.8 ± 0.21 1.8 ± 0.19 1.9 ± 0.21 1.6 ± 0.18 <0.001 BMI (kg/m2) 25 ± 3.5 25 ± 3.5 26 ± 3.5 24 ± 3.2 0.091 Diabetes mellitus 7 (5.7) 1 (1.5) 3 (9.1) 3 (14) 0.044 Dyslipidaemia 42 (34) 23 (34) 12 (36) 7 (32) 0.941 Smoking 25 (20) 6 (9) 15 (45) 4 (18) <0.001 Chronic obstructive pulmonary disease 21 (17) 4 (6) 12 (36) 5 (23) <0.001 Coronary artery disease 11 (9) 5 (7.5) 2 (6.1) 4 (18) 0.294 Paroxysmal AF 25 (20) 9 (13) 6 (18) 10 (45) 0.009 Renal failure (GFR <60 mL/min) 40 (33) 19 (29) 4 (12) 17 (77) <0.001 GFR (mL/min) 78 ± 25 80 ± 24 88 ± 23 57 ± 17 <0.001 EuroSCORE 1 4.5 ± 3.9 3.3 ± 2 4.2 ± 3.8 8.8 ± 5.5 <0.001 Pre-operative NYHA class 0.434  I 27 (22) 16 (24) 8 (24) 3 (14)  II 83 (68) 47 (70) 20 (61) 16 (73)  ≥III 12 (9.8) 4 (6) 5 (15) 3 (14) Prolapse site 0.622  Anterior 10 (9.1) 5 (8.1) 3 (10) 2 (11)  Posterior 85 (77) 49 (79) 20 (69) 16 (84)  Both leaflets 15 (14) 8 (13) 6 (21) 1 (5.3) Flail leaflet 64 (55) 34 (53) 18 (60) 12 (55) 0.821 Medical therapies  Beta blockers 38 (32) 15 (23) 8 (24) 15 (71) <0.001  Diuretic 45 (37) 16 (24) 12 (36) 17 (77) <0.001  Angiotensin receptor blockers 17 (14) 7 (11) 4 (12) 6 (29) 0.141  Angiotensin conversion enzyme inhibitors 30 (25) 13 (20) 10 (30) 7 (33) 0.323  Aspirin 24 (20) 11 (17) 9 (27) 4 (19) 0.454 Mitral surgery (type of procedure) 0.728  Repair 105 (86) 59 (88) 28 (85) 18 (82)  Replacement 17 (14) 8 (12) 5 (15) 4 (18) Tricuspid annuloplasty 20 (16) 5 (7.5) 7 (21) 8 (36) 0.004 CABG 0.192 Quantitative data are expressed as means and standard deviations. Categorical variables are expressed as numbers (%). AF, atrial fibrillation; BMI, body mass index; BSA, body surface area; CABG, coronary artery bypass graft; GFR, glomerular filtration rate; NYHA, New York Heart Association. Phenogroup-1 included 67 (median age 61 years old; 70% male) patients. Phenogroup-2 included 33 (median age 61 years old; 94% male) patients, whereas phenogroup-3 included 22 (median age 73 years old; 23% male) patients. The baseline characteristics were imbalanced between groups. Phenogroups-1 and 2 were predominantly male (70% and 94% of patients, respectively), whereas phenogroup-3 was mostly composed of women (77% of patients, P < 0.001) (Table 1 and Supplementary data online, Table S1). Patients in Group 3 were older than in the others (73 ± 5 years old vs. 61 ± 10 years old in Group 2 and 61 ± 11 in Group 1; P < 0.0001 after pairwise comparison). Regarding cardiovascular risk factors (besides age and overweight, mentioned above), patients in phenogroup-3 were more likely to have high blood pressure than those of the first phenogroup: 64% of patients in Group 3 had hypertension vs. 29% in Group 1 (P = 0.015) (Supplementary data online, Table S1). Patients in Group 3 were also more likely to have diabetes mellitus (Table 1). Regarding comorbidities, chronic obstructive pulmonary disease (COPD) was more frequent in Group 2 (see Supplementary data online, Table S1). Before surgery, the predicted operative mortality (estimated using EuroSCORE) in phenogroup-3 was higher than in other groups: the EuroSCORE 1 was 8.8% in Group 3 vs. 4.2% in Group 2 (P < 0.0001) and 3.3% in the Group 1 (P < 0.0001). The EuroSCORE 2 was 2.3% in Group 3 vs. 1.1% in Group 2 (P < 0.0001) and 1.0% in the Group 1 (P < 0.0001). Patients in Group 3 were more likely to have pre-operative paroxysmal AF: 45% of patients in Group 3 had a history of AF vs. 18% in Group 2 and only 13% in the Group 1 (P = 0.009) (Table 1). Pre-operative echocardiographic characteristics Regarding left ventricular (LV) systolic function: LV ejection fraction (LVEF) was comparable between groups, with a mean of 67 ± 7% (P = 0.127) (Table 2). There was a trend towards a lower GLS in phenogroup-2, but this difference was weakly significant: GLS was −19 ± 3.1% in Group 2 vs. −20 ± 3.0% in Group 3 and −21 ± 3.1% in the Group 1, P = 0.049 (see Supplementary data online, Table S2). In phenogroup-2, the left ventricle was more dilated than in the Group 1. Table 2 Pre-operative echocardiographic data Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Quantitative data are expressed as means and standard deviations. EROA, effective regurgitant orifice area; GLS, global longitudinal strain; LA, left atrial; LV, left ventricular; LVEDD, left ventricular end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; LVESV, left ventricular end-systolic volume; PASP, pulmonary artery systolic pressure; RA, right atrial; RV, right ventricular; TAPSE, tricuspid annular plane systolic excursion; VTI, velocity time integral. Table 2 Pre-operative echocardiographic data Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Left ventricle  LVEF (%) 67 ± 7 66 ± 6.8 66 ± 7.6 69 ± 6.3 0.127  GLS LV (%) −20 ± 3.1 −21 ± 3.1 −19 ± 3.1 −20 ± 3 0.049  Indexed LVEDD (mm/m2) 31 ± 4.2 30 ± 4 33 ± 4.2 32 ± 3.6 0.002  Indexed LVESD (mm/m2) 20 ± 3.4 19 ± 2.7 21 ± 4.3 20 ± 2.6 0.002  Indexed LVEDV (mL/m2) 81 ± 21 76 ± 17 97 ± 22 73 ± 17 <0.001  Indexed LVESV (mL/m2) 28 ± 10 26 ± 8.1 34 ± 12 24 ± 7.1 <0.001  Indexed stroke volume (mL/m2) 37 ± 8.9 37 ± 7.7 39 ± 9.9 36 ± 11 0.562  Indexed cardiac output (L/min/m2) 2.7 ± 0.65 2.7 ± 0.58 2.9 ± 0.71 2.5 ± 0.69 0.100 Mitral characteristics  Regurgitation volume (mL) 92 ± 32 85 ± 30 110 ± 33 86 ± 27 0.014  EROA (mm2) 60 ± 20 56 ± 19 69 ± 20 55 ± 17 0.010  E velocity (cm/s) 130 ± 36 120 ± 31 130 ± 33 150 ± 44 0.002  E-wave deceleration time (ms) 170 ± 49 170 ± 48 180 ± 56 170 ± 39 0.595  A velocity (cm/s) 68 ± 27 69 ± 29 64 ± 23 75 ± 29 0.347  E/A ratio 1.8 ± 0.57 1.7 ± 0.53 1.9 ± 0.49 2 ± 0.78 0.114  e’ velocity (cm/s) 11 ± 3.7 11 ± 3.3 11 ± 4.1 8.8 ± 4 0.013  E/e’ ratio 13 ± 5.9 12 ± 4.9 13 ± 5.4 18 ± 7.4 0.002  S velocity (cm/s) 9.4 ± 2.4 9.5 ± 2.2 10 ± 2.7 7.6 ± 1.4 <0.001 Left atrial  Indexed LA volume (mL/m2) 53 ± 23 44 ± 19 66 ± 25 58 ± 19 <0.001  LA diameter (mm) 44 ± 7.4 42 ± 6.7 48 ± 7.9 44 ± 6.7 0.004  LA peak systolic strain 25 ± 9 27 ± 8.8 23 ± 7.8 19 ± 9.4 0.006 Right atrial  RA volume (mL) 50 ± 20 44 ± 15 65 ± 24 50 ± 14 <0.001 Right ventricle  PASP (mmHg) 40 ± 14 35 ± 9.7 42 ± 17 48 ± 13 <0.001  TAPSE (mm) 24 ± 4.3 24 ± 4.2 23 ± 4.9 24 ± 4 0.709  RV fractional area change (%) 44 ± 10 46 ± 8 41 ± 12 43 ± 11 0.300  Pulsed Doppler S’ wave (cm/s) 15 ± 3.3 15 ± 3.1 15 ± 3.7 14 ± 3.2 0.853  GLS RV (%) −21 ± 4.5 −23 ± 4.2 −19 ± 4.1 −20 ± 4.8 0.009  Free wall RV longitudinal strain −24 ± 6.8 −25 ± 6.4 −22 ± 7 −23 ± 7.1 0.178 Quantitative data are expressed as means and standard deviations. EROA, effective regurgitant orifice area; GLS, global longitudinal strain; LA, left atrial; LV, left ventricular; LVEDD, left ventricular end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESD, left ventricular end-systolic diameter; LVESV, left ventricular end-systolic volume; PASP, pulmonary artery systolic pressure; RA, right atrial; RV, right ventricular; TAPSE, tricuspid annular plane systolic excursion; VTI, velocity time integral. The left atrium (LA) was larger in Group 2: the indexed LA volume was 66 ± 25 mL/m2 (which represent a severe LA dilatation) vs. 58 ± 19 mL/m2 in Group 3 and 44 ± 19 mL/m2 in the Group 1 (P < 0.001). The pairwise comparison highlighted LA enlargements in both Groups 2 and 3 compared with the Group 1, but there was no longer any statistically significant difference between Groups 2 and 3 (see Supplementary data online, Table S2). The right atrium was also enlarged in Group 2 (Table 2). Differences between groups regarding quantitative evaluation of mitral regurgitation were observed. Indeed, a larger effective regurgitant orifice area (EROA) and a larger regurgitation volume were assessed in Group 2: the average EROA was 69 mm2 vs. 55 mm2 in Group 3 and 56 mm2 in Group 1 (P = 0.010) (Table 2). Regarding the right heart, pulmonary artery systolic pressure was estimated at ∼48 ± 13 mmHg in Group 3 vs. ∼42 ± 17 mmHg in Group 2 and ∼35 ± 9.7 mmHg in the Group 1 (P < 0.001) (Table 2). The right ventricle (RV) was more dilated in Group 2 (RV end-diastolic area = 22 ± 3.6 cm2 vs. 17 ± 3.1 cm2 in Group 3 and 15 ± 2.7 cm2 in the Group 1; P < 0.001). Post-operative outcomes and endpoints Regarding the primary endpoint (Table 3 and Figure 2) compared with patients in the Group 1, patients in Group 3 were more likely to develop PCE. Table 3 Association of phenogroups with post-operative outcomes on a univariate Cox model: hazard ratio with (95% confidence interval) and P-value Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Table 3 Association of phenogroups with post-operative outcomes on a univariate Cox model: hazard ratio with (95% confidence interval) and P-value Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) All cardiovascular events 1.0 1.84 (0.89–3.79), P = 0.099 3.57 (1.72–7.44), P < 0.001 Post-operative long-term atrial fibrillation 1.0 1.55 (0.60–4.00), P = 0.368 4.75 (2.03–11.10), P < 0.001 Figure 2 View largeDownload slide Survival free of cardiovascular events, stratified by phenogroup. Figure 2 View largeDownload slide Survival free of cardiovascular events, stratified by phenogroup. Post-operative outcomes are detailed in Table 4. Regarding the secondary endpoint, 29 (24%) patients have developed post-operative long-term AF in the overall population. Post-operative long-term AF occurred significantly more often in phenogroup-3 than in phenogroup-1. There was no statistical difference between the Groups 1 and 2 or between the Groups 2 and 3 (Table 4 and Supplementary data online, Table S3). The risk of developing post-operative long-term AF was significantly increased in Group 3 (Table 3 and Figure 3). Table 4 Post-operative outcomes Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Categorical variables are expressed as numbers (%). Table 4 Post-operative outcomes Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Overall (n = 122) Group 1 (n = 67) Group 2 (n = 33) Group 3 (n = 22) P-value Post-operative immediate AF (≤30 days) 26 (21) 12 (18) 7 (21) 7 (32) 0.409 Post-operative long-term AF (>30 days) 29 (24) 11 (16) 7 (21) 11 (50) 0.005 All-cause mortality 4 (3.3) 2 (3) 0 (0) 2 (9.1) 0.153 Cardiovascular mortality 0 (0) 0 (0) 0 (0) 0 (0) 1 Stroke 10 (8.3) 5 (7.6) 3 (9.1) 2 (9.1) 1 Cardiovascular-cause of hospitalization 22 (18) 6 (9.1) 11 (33) 5 (23) 0.009 Categorical variables are expressed as numbers (%). Figure 3 View largeDownload slide Survival free of post-operative long-term AF, stratified by phenogroup. Figure 3 View largeDownload slide Survival free of post-operative long-term AF, stratified by phenogroup. Regarding mortality, only four patients died during the follow-up, including two patients in Groups 1 and 2 patients in Group 3. There was no difference between groups (P = 0.153). The incidence of cardiovascular causes of hospitalization was significantly higher in phenogroup-2 compared with the first group. Other between-groups differences were not statistically significant. Paroxysmal pre-operative AF was confirmed as a risk factor of developing both PCE and post-operative AF (Supplementary data online, Figures S1 and S2). The risk of developing post-operative AF (including both immediate and long-term AF) was significantly increased in patients with pre-operative AF: hazard ratio = 3.19(1.49–6.82) (P = 0.00284). Discussion In this study, using machine learning and hierarchical clustering analysis applied on dense phenotypic data, the main findings were as follows (i) PMR is a heterogeneous disease, (ii) using a phenomapping dedicated algorithm, three phenogroups of patients were identified, despite PMR heterogeneity, (iii) these phenogroups had different prognoses, suggesting different risk profiles, and (iv) pre-operative paroxysmal AF is an important risk factor of PCE. A previous study, ‘Phenomapping for Novel Classification of Heart Failure With Preserved Ejection Fraction’,6 provided a novel classification of a cardiovascular disorder using phenomapping and was ‘the first study that applies machine learning techniques to resolve heterogeneity in a cardiovascular syndrome using dense phenotypic data’. Using a similar algorithm, our study demonstrates that it is also possible to identify different phenogroups in PMR disease. Indeed, all patients had criteria for severe PMR, but phenomapping showed that PMR is a heterogeneous disease. A novel classification of PMR patients was defined using correlations between phenotypic variables to find patterns in dense data. Once the three groups were identified, the differences between them were striking regarding clinical and echocardiographic characteristics. The three phenogroups represented three archetypes of PMR according to their demographic, clinical, and echocardiographic characteristics: Phenogroup-1 represented low-risk patients (less risk factors, less dilatation of the ventricle…) but their main characteristics did not differ from those of the overall population. Phenogroup-2, which represented intermediate-risk patients, included predominantly male patients, aged approximately 61 years old, with large body surface areas, who were smokers and with histories of COPD. Their heart remodelling appeared to be more pronounced. Phenogroup-3, which represented high-risk patients, included predominantly women who were older, thinner, and smaller, with hypertension and comorbidities such as renal failure or a pre-operative history of AF. Their EuroSCORE was higher than those of the other groups. Few studies have explored gender influence on valvular disease. However, women seemed to have worse outcomes than men after mitral surgery,15 although this difference can be partially driven by the rate of mitral valve replacement in women. The heart remodelling in phenogroup-3 appeared to be surprisingly less marked than that in Group 2. Regarding echocardiographic characteristics: LA peak systolic strain is confirmed as a predictor of cardiovascular events in mitral regurgitation16 and is associated with elevated filling pressures17. E-wave velocity was also the highest in Group 3 (150 ± 44 cm/s): it may be explained by both an important mitral regurgitation (increasing stroke volume) and a more severe diastolic dysfunction in these patients, as described previously.18 LV diastolic dysfunction in PMR is known to result in poor outcomes19 and seemed to predict better cardiovascular outcomes than LVEF (mean pre-operative LVEF was 67%, with no difference between groups). LVEF may remain in the normal range for a long period of time, whereas subclinical myocardial LV dysfunction may be present.20 Therefore, it is necessary to look for other echocardiographic predictors of outcomes, such as increased LA volume,21 abnormal LV GLS,4 pulmonary hypertension,22 increased E/e’ ratio,23 or abnormal LA peak systolic strain.16 Most of these parameters were altered in phenogroup-3 (except LV GLS). Of note, regurgitant volume and EROA were not good predictors of post-operative outcomes. Phenogroup-2 had global cardiac chamber enlargement (LA and right atrial volumes, LV volume, and RV size) without obvious diastolic dysfunction or even elevated filling pressures. Although cardiac chamber enlargement suggests myocardial remodelling, it did not seem to result in an increased risk of PCE. The enlargement in the right chambers and the lowest RV global strain in phenogroup-2 can also be explained by the important number of patients with COPD (36% of patients had a history of COPD in phenogroup-2).24 The risk of developing PCE was not significantly increased in phenogroup-2. We hypothesized that phenogroup-2 likely included intermediate-risk patients. Pre-operative AF is known as an important predictor of cardiovascular outcomes,25 which was confirmed in this study, where 45% of phenogroup-3 patients, who presented the highest risk of PCE occurrence, had pre-operative paroxysmal AF. The high incidences of pre-operative AF in this group can be explained by several factors. LA dilatation is considered as an important risk factor for the occurrence of AF,21 and indexed LA volumes in phenogroup-3 (mean 58 ± 19 mL/m2) were close to the cut-off of 60 mL/m2.21 Nevertheless, left atrium was also dilated in Group 2, although these patients were less likely to have post-operative long-term AF than those in Group 3. Therefore, an enlarged LA is not sufficient to explain the highest prevalence of AF in Group 3. Phenogroup-3 patients were older10 and were also more likely to have hypertension and chronic renal failure.10 LA peak systolic strain was reduced in phenogroup-3, reflecting an LA dysfunction (reduced reservoir function).26 Finally, pre-operative AF was a predictor of post-operative AF (Supplementary data online, Figure S2).27 Perspectives Determining optimal timing for surgery in severe PMR is still problematic, and is currently based on symptoms, severity of the regurgitation and its impact on LV volume and systolic function. Predictors of outcome currently considered in regurgitant valvular diseases are LV increased volumes and reduced LVEF, indicating an alteration of myocardial contractility. However, reduced LVEF is often a late consequence of valve dysfunction and may even imply irreversible myocardial injury. Machine learning identified three phenogroups with different prognoses. Therefore, the management of patients with severe PMR could be improved, as high-risk patients can now be identified earlier. We suggest that high-risk patients should be carefully monitored (i.e. by more frequent visits to their usual cardiologist). Sinus rhythm should be maintained as long as possible, and paroxysmal AF can be detected by Holter monitoring. Pulmonary vein isolation could be performed during valve surgery in patients with pre-operative paroxysmal AF to prevent post-operative AF.10 Mitral surgery may be performed early in (asymptomatic) phenogroup-3 individuals to prevent the AF-occurrence risk and can be encouraged by early detection of the cardiac consequences of PMR by performing exercise echocardiography.28 Study limitations We acknowledge several limitations. Hierarchical clustering led to an imbalanced number of patients between groups. This imbalance impeded definite conclusions about the value of phenomapping in PMR and resulted in a lack of statistical power, particularly because of the small sample size. The retrospective design of our study also implies that we had to perform an imputation procedure to deal with missing values and to exclude several patients having extreme values due to mistyping in electronic health records. Therefore, our results should be considered as hypothesis generating and interpreted with caution. To confirm the reproducibility of our results (i.e. the identification of phenogroups, outcomes, and risk-levels), we will have to further control our results with a validation cohort. Conclusion In this study, PMR was confirmed as a heterogeneous clinical disease, but this heterogeneity may be resolved using a dedicated algorithm to perform phenomapping analysis. Phenotypic data can be grouped in clusters using machine learning to identify three phenogroups of patients with different prognoses. Supplementary data Supplementary data are available at European Heart Journal - Cardiovascular Imaging online. Acknowledgements The authors thank the study nurses from the CIC-IT1414 but also the CORECT programmes and the scientific committee of the CHU-Rennes for their support. Conflict of interest: None declared. References 1 Baumgartner H , Falk V , Bax JJ , De Bonis M , Hamm C , Holm PJ et al. 2017 ESC/EACTS Guidelines for the management of valvular heart disease: the Task Force for the Management of Valvular Heart Disease of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS) . Eur Heart J 2017 ; 38 : 2739 – 91 . Google Scholar CrossRef Search ADS PubMed 2 Nishimura RA , Otto CM , Bonow RO , Carabello BA , Erwin JP , Fleisher LA et al. 2017 AHA/ACC Focused Update of the 2014 AHA/ACC guideline for the management of patients with valvular heart disease: a Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines . J Am Coll Cardiol 2017 ; 70 : 252 – 89 . Google Scholar CrossRef Search ADS PubMed 3 Tribouilloy C , Rusinaru D , Szymanski C , Mezghani S , Fournier A , Lévy F et al. 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 . Eur J Echocardiogr 2011 ; 12 : 702 – 10 . Google Scholar CrossRef Search ADS PubMed 4 Mascle S , Schnell F , Thebault C , Corbineau H , Laurent M , Hamonic S et al. Predictive value of global longitudinal strain in a surgical population of organic mitral regurgitation . J Am Soc Echocardiogr 2012 ; 25 : 766 – 72 . Google Scholar CrossRef Search ADS PubMed 5 Hastie T , Tibshirani R , Friedman J. Elements of Statistical Learning: Data Mining, Inference, and Prediction , 2nd ed . Springer Ed ; 2009 . https://statweb.stanford.edu/∼tibs/ElemStatLearn/ (12 September 2017, date last accessed). 6 Shah SJ , Katz DH , Selvaraj S , Burke MA , Yancy CW , Gheorghiade M et al. Phenomapping for novel classification of heart failure with preserved ejection fraction . Circulation 2015 ; 131 : 269 – 79 . Google Scholar CrossRef Search ADS PubMed 7 Zoghbi WA , Adams D , Bonow RO , Enriquez-Sarano M , Foster E , Grayburn PA et al. Recommendations for noninvasive evaluation of native valvular regurgitation: a report from the American Society of Echocardiography Developed in Collaboration with the Society for Cardiovascular Magnetic Resonance . J Am Soc Echocardiogr 2017 ; 30 : 303 – 71 . Google Scholar CrossRef Search ADS PubMed 8 Lancellotti P , Tribouilloy C , Hagendorff A , Popescu BA , Edvardsen T , Pierard LA et al. ; Scientific Document Committee of the European Association of Cardiovascular Imaging . Recommendations for the echocardiographic assessment of native valvular regurgitation: an executive summary from the European Association of Cardiovascular Imaging . Eur Heart J Cardiovasc Imaging 2013 ; 14 : 611 – 44 . Google Scholar CrossRef Search ADS PubMed 9 Lang RM , Badano LP , Mor-Avi V , Afilalo J , Armstrong A , Ernande L et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging . Eur Heart J Cardiovasc Imaging 2015 ; 16 : 233 – 71 . Google Scholar CrossRef Search ADS PubMed 10 Kirchhof P , Benussi S , Kotecha D , Ahlsson A , Atar D , Casadei B et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS . Eur Heart J 2016 ; 37 : 2893 – 962 . Google Scholar CrossRef Search ADS PubMed 11 Stekhoven DJ , Bühlmann P. MissForest—non-parametric missing value imputation for mixed-type data . Bioinformatics 2012 ; 28 : 112 – 8 . Google Scholar CrossRef Search ADS PubMed 12 ClustVarLV: Clustering of Variables Around Latent Variables Version 1.5.1 from CRAN. https://rdrr.io/cran/ClustVarLV (12 September 2017, date last accessed). 13 Charrad. NbClust: An R Package for determining the relevant number of clusters in a data set. J Stat Softwe. https://www.jstatsoft.org/article/view/v061i06 (12 September 2017, date last accessed). 14 R Core Team . R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2013. https://www.r-project.org (12 September 2017, date last accessed). 15 Seeburger J , Eifert S , Pfannmüller B , Garbade J , Vollroth M , Misfeld M et al. Gender differences in mitral valve surgery . Thorac Cardiovasc Surg 2013 ; 61 : 42 – 6 . Google Scholar PubMed 16 Debonnaire P , Leong DP , Witkowski TG , Al Amri I , Joyce E , Katsanos S et al. Left atrial function by two-dimensional speckle-tracking echocardiography in patients with severe organic mitral regurgitation: association with guidelines-based surgical indication and postoperative (long-term) survival . J Am Soc Echocardiogr 2013 ; 26 : 1053 – 62 . Google Scholar CrossRef Search ADS PubMed 17 Wakami K , Ohte N , Asada K , Fukuta H , Goto T , Mukai S et al. Correlation between left ventricular end-diastolic pressure and peak left atrial wall strain during left ventricular systole . J Am Soc Echocardiogr 2009 ; 22 : 847 – 51 . Google Scholar CrossRef Search ADS PubMed 18 Nishimura RA , Tajik AJ. Evaluation of diastolic filling of left ventricle in health and disease: Doppler echocardiography is the clinician’s Rosetta Stone . J Am Coll Cardiol 1997 ; 30 : 8 – 18 . Google Scholar CrossRef Search ADS PubMed 19 Magne J , Mahjoub H , Pierard LA , O'Connor K , Pirlet C , Pibarot P et al. Prognostic importance of brain natriuretic peptide and left ventricular longitudinal function in asymptomatic degenerative mitral regurgitation . Heart 2012 ; 98 : 584 – 91 . Google Scholar CrossRef Search ADS PubMed 20 Magne J , Szymanski C , Fournier A , Malaquin D , Avierinos JF , Tribouilloy C. Clinical and prognostic impact of a new left ventricular ejection index in primary mitral regurgitation because of mitral valve prolapse . Circ Cardiovasc Imaging 2015 ; 8 : e003036 . Google Scholar CrossRef Search ADS PubMed 21 Le Tourneau T , Messika-Zeitoun D , Russo A , Detaint D , Topilsky Y , Mahoney DW et al. Impact of left atrial volume on clinical outcome in organic mitral regurgitation . J Am Coll Cardiol 2010 ; 56 : 570 – 8 . Google Scholar CrossRef Search ADS PubMed 22 Barbieri A , Bursi F , Grigioni F , Tribouilloy C , Avierinos JF , Michelena HI et al. ; Mitral Regurgitation International DAtabase (MIDA) Investigators . Prognostic and therapeutic implications of pulmonary hypertension complicating degenerative mitral regurgitation due to flail leaflet: a multicenter long-term international study . Eur Heart J 2011 ; 32 : 751 – 9 . Google Scholar CrossRef Search ADS PubMed 23 Agricola E , Galderisi M , Oppizzi M , Melisurgo G , Airoldi F , Margonato A. Doppler tissue imaging: a reliable method for estimation of left ventricular filling pressure in patients with mitral regurgitation . Am Heart J 2005 ; 150 : 610 – 5 . Google Scholar CrossRef Search ADS PubMed 24 Hilde JM , Skjørten I , Grøtta OJ , Hansteen V , Melsom MN , Hisdal J et al. Right ventricular dysfunction and remodeling in chronic obstructive pulmonary disease without pulmonary hypertension . J Am Coll Cardiol 2013 ; 62 : 1103 – 11 . Google Scholar CrossRef Search ADS PubMed 25 Grigioni F , Avierinos J-F , Ling LH , Scott CG , Bailey KR , Tajik AJ et al. Atrial fibrillation complicating the course of degenerative mitral regurgitation: determinants and long-term outcome . J Am Coll Cardiol 2002 ; 40 : 84 – 92 . Google Scholar CrossRef Search ADS PubMed 26 Kuppahally SS , Akoum N , Burgon NS , Badger TJ , Kholmovski EG , Vijayakumar S et al. Left atrial strain and strain rate in patients with paroxysmal and persistent atrial fibrillation: relationship to left atrial structural remodeling detected by delayed-enhancement MRI . Circ Cardiovasc Imaging 2010 ; 3 : 231 – 9 . Google Scholar CrossRef Search ADS PubMed 27 Raiten JM , Ghadimi K , Augoustides JGT , Ramakrishna H , Patel PA , Weiss SJ et al. Atrial fibrillation after cardiac surgery: clinical update on mechanisms and prophylactic strategies . J Cardiothorac Vasc Anesth 2015 ; 29 : 806 – 16 . Google Scholar CrossRef Search ADS PubMed 28 Naji P , Griffin BP , Asfahan F , Barr T , Rodriguez LL , Grimm R et al. Predictors of long-term outcomes in patients with significant myxomatous mitral regurgitation undergoing exercise echocardiography . Circulation 2014 ; 129 : 1310 – 9 . Google Scholar CrossRef Search ADS PubMed Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

European Heart Journal – Cardiovascular ImagingOxford University Press

Published: Mar 28, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off