RESULTS Baseline characteristics and mitral valve repair details Baseline characteristics of the whole study group are presented in Table 1. Median patient age was 66.5 (IQR 57.8–73.8) years and the majority of patients were male. The repair and procedure details are presented in Table 2. A full ring was used in all patients; the majority of patients (n = 456; 94.2%) underwent either a Physio I (n = 223; 46.1%) or Physio II (n = 233; 48.1%) ring (Edwards Lifesciences, Irvine, CA, USA) implantation. In 42 (8.7%) patients, a ring size 30 or smaller was implanted. Table 1: Baseline characteristics n = 484 Age (years) 66.5 (57.8–73.8) Gender (female) 171 (35.3) Body surface area (m2) 1.93 ± 0.22 NYHA  I 146 (30.2)  II 243 (50.2)  III–IV 95 (19.6) Hypertension 221 (45.7) Atrial fibrillation 172 (35.5) Renal impairment (ml/min/1.73 m2)  Moderate (CC 50–85) 226 (46.7)  Severe (CC <50) 41 (8.5) Diabetes mellitus 22 (4.5) Chronic lung disease 42 (8.7) Mitral annular calcification 49 (10.1) Echocardiographic characteristics  LVEF ≤60% 134 (27.7)  LVEDD (mm)a 54 ± 7  LVESD (mm)a 33 ± 7  LAD (mm)a 45.0 (40.0–49.0) EuroSCORE II 1.76 (0.99–3.24) Degenerative disease subtype  Barlow’s disease 140 (28.9)  Forme fruste Barlow’s disease 51 (10.5) Site of leaflet prolapse  Posterior leaflet 299 (61.8)   Single scallop prolapse 234 (48.3)  Anterior leaflet 41 (8.5)  Bileaflet 144 (29.7) Data are presented as n (%) and means ± SD or medians (IQR). a Available for ≥95% of patients. CC: creatinine clearance; IQR: interquartile range; LAD: left atrial diameter; LVEDD: left ventricular end-diastolic diameter; LVEF: left ventricular ejection fraction; LVESD: left ventricular end-systolic diameter; NYHA: New York Heart Association; SD: standard deviation. Table 2: Mitral valve repair details n = 484 Mitral valve annulus  Plication 102 (21.1) Anterior mitral valve leaflet repair (neochords) 125 (25.8) Posterior mitral valve leaflet  Resection 321 (66.3)  Leaflet sliding 222 (45.9)  Neochords 206 (42.6)  Chordal transfer 9 (1.9)  Indentation closure 85 (17.6) Commissural repair 111 (22.9) Annuloplasty ring type  Physio I ring 223 (46.1)  Physio II ring 233 (48.1)  Other 28 (5.8) Annuloplasty ring size  24 1 (0.2)  26 10 (2.1)  28 31 (6.4)  30 62 (12.8)  32 112 (23.1)  34 98 (20.3)  36 77 (15.9)  38 52 (10.7)  40 41 (8.5) Intraoperative mitral valve gradient (mmHg) 1.66 (1.16–2.40) Concomitant procedures  Tricuspid valve repair 321 (51.4)  Radiofrequency ablation for atrial fibrillation 148 (30.6)  Coronary artery bypass surgery 96 (19.8)  Aortic valve intervention 28 (5.8) Data are presented as n (%) and medians (IQR). IQR: interquartile range. Postrepair mitral valve gradient A moderate positive correlation (r = 0.356, n = 391, P < 0.001) between the intraoperative and predischarge MV gradient was seen (Supplementary material, Figure S1). The haemodynamic results of valve repair are demonstrated in Table 3. As expected, postrepair MV gradients decreased with the size of annuloplasty ring implanted (Fig. 1). On linear regression analysis, increasing patient age (coefficient = −0.11, standard error = 0.005, P = 0.034) and annuloplasty ring size (coefficient = −0.181, standard error = 0.018, P < 0.001) were correlated with lower postrepair MV gradients (Table 4). On the other hand, increasing body surface area (coefficient = 0.905, standard error = 0.340, P = 0.008) and, interestingly, use of the Physio I ring (coefficient = 0.414, standard error = 0.122, P < 0.001) were correlated with higher postrepair MV gradients. The linear regression model demonstrated good fit with no evidence of heteroscedasticity or non-linearity (Supplementary material, Figure S2). Figure 1: Postrepair resting mitral valve gradients in relation to the implanted annuloplasty ring size. Smallest size rings (≤size 28) had the most pronounced effect on postrepair gradients. The middle horizontal line presents the median and the upper and lower borders of the box present the upper and lower quartiles. The upper and lower whiskers present the maximum and minimum values of non-outliers. Extra dots present the outliers. Table 3: Postoperative echocardiographic results n = 484 Predischarge mitral valve gradient (mmHg) 3.46 ± 1.43 Predischarge heart rate (min−1) 80 ± 15 Predischarge stroke volume (ml)a 74 (58–93) Predischarge cardiac index (l/m2)a 5.7 (4.6–7.4) Predischarge haemoglobin value (mmol/l) 6.4 ± 0.8 Data are presented as means ± standard deviations and medians (IQR). a Available for ≥90% of patients. IQR: interquartile range. Table 4: Univariable and multivariable linear regression analysis on risk factors associated with postrepair mitral valve gradients Univariable analysis Multivariable analysis Coefficient Standard error P-value Coefficient Standard error P-value Gender (female) 0.114 0.137 0.40 0.069 0.153 0.65 Age (years) 0.000 0.006 0.98 −0.110 0.005 0.034 Body surface area 0.294 0.302 0.33 0.905 0.340 0.008 Atrial fibrillation −0.137 0.136 0.32 −0.008 0.121 0.95 Posterior mitral annular plication 0.347 0.160 0.030 0.109 0.155 0.48 Annular decalcification −0.682 0.223 0.002 −0.235 0.195 0.23 PMVL resection 0.028 0.139 0.84 0.188 0.126 0.14 PMVL indentation closure −0.154 0.172 0.37 −0.123 0.149 0.41 Annuloplasty ring size −0.164 0.017 <0.001 −0.181 0.018 <0.001 Annuloplasty ring type  Physio I ring 0.486 0.129 <0.001 0.414 0.122 0.001 Postoperative blood haemoglobin value −0.288 0.082 <0.001 −0.301 0.071 <0.001 Postoperative heart rate 0.029 0.004 <0.001 0.024 0.004 <0.001 PMVL: posterior mitral valve leaflet. Late clinical results From cubic spline-based analyses, no evidence of non-linearity for post repair MV gradient was found for all time-to-event outcomes (Supplementary materials, Figures S3–S5). Hence, postrepair MV gradient was entered into the Cox proportional hazards regression analysis models as a continuous variable for all subsequent analyses. During a median follow-up period of 5.8 (IQR 3.4–9.3) years (99.5% complete, 2 patients were lost to follow-up due to emigration), 84 patients died. When corrected for patient age, gender, chronic pulmonary disease, renal impairment, left ventricular function, symptomatic MR and atrial fibrillation, postrepair MV gradients was not associated with patient survival [hazard ratio (HR) 1.034, 95% CI 0.889–1.203; P = 0.66; Supplementary material, Table S1]. Follow-up on clinical related events was 90% complete with a mean duration of 5.2 (IQR 2.7–8.8) years. When adjusted for patient age, tricuspid valve repair and preoperative left atrial diameter, postrepair MV gradient was not associated with late atrial fibrillation occurrence (HR 0.849, 95% CI 0.682–1.057; P = 0.14; Supplementary material, Table S1). When adjusted for anterior MV leaflet repair, postrepair MV gradient was associated with a higher risk of MV reintervention (HR 1.378, 95% CI 1.033–1.838; P = 0.029; Supplementary material, Table S1). A total of 11 late reinterventions were performed. The indication for reintervention was recurrent MR in 9 patients and elevated MV gradient in the remaining 2 patients. In both of the latter patients, a postrepair gradient of ≥5 mmHg (5.88 and 5.23 mmHg, respectively) was seen on predischarge echocardiography, despite an acceptable gradient seen on intraoperative echocardiography (2.53 and 4.23 mmHg, respectively). In the first of the latter patients, an annuloplasty ring size 26 was initially implanted and a re-repair due to heart failure symptoms (no significant MR or other explanation for this observation were found) was performed 1.3 years after the initial operation. Upon reoperation, the annuloplasty ring was explanted, resolving the problem of elevated gradient. In the other patient, an annuloplasty ring size 32 was implanted during the initial operation. Severe pannus formation occurred and the patient underwent a valve replacement 1.1 years after the initial operation. Late echocardiographic results On mixed-model linear regression analysis on the evolution of mean MV gradient during the follow-up period, a slight decline in MV gradients was seen (estimate = −0.05, 95% CI  −0.07 to −0.02; P < 0.001) during the follow-up period (Fig. 2). Figure 2: Longitudinal changes in mean resting mitral valve gradients. During the follow-up period, a slight decrease, when compared with the predischarge postrepair mean mitral valve gradient, can be observed.