-
PDF
- Split View
-
Views
-
Cite
Cite
Michele De Bonis, Elisabetta Lapenna, Benedetto Del Forno, Stefania Di Sanzo, Andrea Giacomini, Davide Schiavi, Luca Vicentini, Azeem Latib, Alberto Pozzoli, Federico Pappalardo, Giovanni La Canna, Ottavio Alfieri, Minimally invasive or conventional edge-to-edge repair for severe mitral regurgitation due to bileaflet prolapse in Barlow’s disease: does the surgical approach have an impact on the long-term results?, European Journal of Cardio-Thoracic Surgery, Volume 52, Issue 1, July 2017, Pages 131–136, https://doi.org/10.1093/ejcts/ezx032
- Share Icon Share
Abstract
OBJECTIVES: To evaluate whether the adoption of a right minithoracotomy operative approach had an impact on the long-term results of edge-to-edge (EE) repair compared to conventional sternotomy in patients with Barlow’s disease and bileaflet prolapse.
METHODS: We assessed the long-term results of 104 patients with Barlow’s disease treated with a minimally invasive EE technique. An equal number of patients had a conventional median sternotomy EE repair for the same disease and were used as a control group. The inverse probability of treatment weighting was used to create comparable distributions of the covariates that were significantly different at baseline in the two groups. We performed a comparative analysis of the groups.
RESULTS: No hospital deaths were observed. Follow-up was 99.5% complete (median 11.3 years). The cumulative incidence function (CIF) of cardiac death at 12 years, with noncardiac death as a competing risk, showed no difference between the two groups (P = 0.87). At 12 years, the CIF of recurrent MR ≥ 3+, with death as the competing risk, was 7% in the sternotomy group and 5% in the minimally invasive group (P = 0.30), and the CIF of recurrence of MR ≥ 2+ was 15 and 14%, respectively (P = 0.63). The type of surgical approach was not a predictor of cardiac death, reoperation, recurrent MR ≥ 3+ or recurrent MR ≥ 2+.
CONCLUSIONS: A minimally invasive approach does not have a negative impact on the effectiveness and long-term durability of the EE repair for bileaflet prolapse in Barlow’s disease. Long-term outcomes are excellent, and valvular performance remains stable over time with no evidence of mitral stenosis.
INTRODUCTION
Mitral repair is the gold standard for treatment of severe degenerative mitral regurgitation (MR) [1]. However, bileaflet prolapse in Barlow's disease [2] remains a surgical challenge. Several different techniques have been described to treat these patients, including the edge-to-edge (EE) repair [3–7]. We previously reported excellent late durability (at 14 years) of this approach when performed through a median sternotomy [3]. There are no published data on the long-term results of the EE technique in patients with Barlow’s disease who were treated with a minimally invasive approach. Indeed, the only studies addressing this issue were limited by a mean follow-up not exceeding 2 years [8–10]. The aim of this study was therefore to evaluate whether the adoption of a right minithoracotomy approach had an impact on the long-term results of this type of repair compared to conventional sternotomy in this difficult setting.
METHODS
Study population
From 1999 to 2010, 104 patients in our institution with severe MR due to the prolapse of both leaflets in the context of Barlow’s disease were submitted to EE repair via a video-assisted right minithoracotomy (minithoracotomy group). Those patients were compared with the first consecutive 104 patients with Barlow’s disease who underwent surgical mitral repair with the EE technique through a conventional median sternotomy (sternotomy group) between 1995 and 2000. Therefore, the final study population included 208 patients.
The diagnosis of Barlow’s disease was made during a preoperative transoesophageal echocardiographic (TEE) examination. It was then confirmed by the surgeon’s direct assessment of the valve. All patients had excessive leaflet tissue, annular dilatation and bileaflet prolapse. Cases with a forme fruste of Barlow’s disease were excluded. Patients affected by Barlow’s disease and bileaflet prolapse treated with techniques other than EE repair and patients with commissural EE or rescue EE repair that was performed as a bailout procedure following failure of standard repair techniques were not included. In addition, patients with endocarditis, previous right thoracotomy, chest wall deformities and redo cases and patients who required concomitant procedures that could not be performed through a minimally invasive approach (e.g. coronary bypass, ascending aorta or aortic valve surgery) were also excluded. Transthoracic echocardiography (TTE) and transoesophageal echocardiography (TEE) were routinely performed, and an integrative approach was used to define MR severity. A non-linear 4-grade scale was adopted to define MR as mild (1+/4+), moderate (2+/4+), moderate-to-severe (3+/4+) and severe (4+/4+). All data were entered into a dedicated database and reviewed retrospectively. The institutional ethics committee approved this study and waived individual consent for this retrospective analysis.
Surgical techniques
Patients in the sternotomy group were operated through a conventional median sternotomy with standard cardiopulmonary bypass (CPB) on moderate hypothermia and intermittent cold blood cardioplegia. The mitral valve was often exposed through a conventional left atrial incision, parallel to the interatrial sulcus.
The minimally invasive approach and the technical aspects of the procedure via an anterolateral minithoracotomy have been described in detail [8]. Briefly, moderately hypothermic (28–30 °C) CPB was instituted via femoral arterial and venous cannulation, performed with the Seldinger technique (through a 2- to 3-cm long incision in the groin) under TEE guidance. At the beginning of our experience, we inserted an additional venous cannula percutaneously in the right internal jugular vein and afterwards in patients who required concomitant tricuspid valve repair or ASD closure. In 100 patients, aortic cross-clamping was achieved using the Chitwood transthoracic clamp (Scanlan International, Inc, Minneapolis, MN, USA) or the Cygnet flexible clamp (Vitalitec, Plymouth, MA, USA), and intermittent cold blood cardioplegia or crystalloid Custodiol cardioplegia was administered through an aortic root catheter. In the remaining four cases, we used an endoaortic balloon (Heartport, Inc, Redwood City, CA), placed under echo guidance in the ascending aorta through the femoral cannula. After balloon inflation, the cardioplegia was delivered through the tip of the catheter. Exposure of the mitral valve was achieved in all cases through a left atriotomy in Sondergaard’s groove, and the left atrial retractor was placed through a parasternal incision. The valve was analysed and repaired using both direct vision and a 30° camera inserted through a 5- or 10-mm port to improve valve assessment and reconstruction.
The mitral valve was repaired in all patients using the EE technique by stitching together the middle portion of the free edges of the mitral leaflets to produce a double-orifice mitral valve. Technical details of this type of repair have been described previously [7, 11].
Follow-up
All patients underwent TTE echocardiography immediately before discharge from the hospital. Clinical and echocardiographic follow-up examinations were performed in our institutional outpatient clinic or via a telephone interview with the patients and the referring cardiologists. Follow-up was 99.5% complete and was significantly longer in the sternotomy group [median 11.5 years, interquartile range (IQR) 10.4; 12.1 vs 10.4 years, IQR 7.0; 12.7].
Statistical analysis
Statistical analyses were performed using SPSS version 22.0 (SPSS Inc., Chicago, IL, USA) for Windows (Microsoft Corp, Redmond, WA) and Stata software version 12. Categorical data were reported as both a number and percentage. Continuous data were expressed as mean ± SD or as the median and IQR. The distribution of variables was evaluated using the Shapiro–Wilk test. The Mann–Whitney U-test was used for independent samples when continuous data were not normally distributed. The χ2 test was used for categorical data, and the Fisher’s exact test was used when the minimum cell size requirements for the chi-square test was not satisfied. P-values < 0.05 (2-tailed) were considered statistically significant. Standardized differences were calculated to compare baseline characteristics in the two surgical groups.
The propensity score and inverse probability of treatment weighting.To adjust for the imbalance of features between minithoracotomy and sternotomy patients, we performed an inverse probability of treatment weighted (IPW) analysis. The IPW was based on the propensity score to create a synthetic sample in which the distribution of measured baseline covariates is independent of the treatment assignment. Weights were estimated through a logistic model that included the following variables, which are considered possible confounders: age, sex, New York Heart Association class III–IV and ejection fraction. To reduce the influence of outlying weights, we performed analyses with stabilized weights [12].
End point analysis.Crude and weighted rates of end-points were calculated and compared between the two treatments using the Kaplan–Meier method. Fine–Gray models [13] were used in competing risk analysis for time to cardiac death with non-cardiac causes of death as competing risk and for time to reoperation and to recurrent MR ≥ 3+ and MR ≥ 2+ with death as the competing risk.
Estimates of cumulative incidence, adjusted for inverse probability weights, were compared to determine the impact of the type of operation on the end-points. Equality across groups was tested with the Pepe–Mori test. We also compared the two treatments by estimating the adjusted hazard ratios through the weighted Fine–Gray competing risks regression model on the type of surgery. A robust variance estimator was used to account for the estimation of weights. The proportionality of all covariates on the hazard for the outcome variables of interest was checked by assessing the lack of significance of the interaction of each covariate with the survival time.
RESULTS
Patient characteristics
At admission, most of the patients were young (age 43 ± 13.6), were in New York Heart Association (NYHA) functional class I or II (86%) and in sinus rhythm (93%). No patients had renal failure, congestive heart failure or peripheral vascular disease. All patients were operated on electively, had a grade 3+ or 4+ MR and normal left ventricular ejection fraction (61 ± 6.4). A bileaflet prolapse was in all cases the mechanism of mitral insufficiency and associated with excessive myxomatous tissue, annular dilatation, elongated chordae and leaflet thickening (Barlow’s disease).
As shown in Table 1, although some baseline differences were noted between the minithoracotomy and sternotomy groups, the two groups were well balanced after IPW adjustment, and absolute standardized differences were below 0.20 [14], indicating an adequate match (Table 2).
. | Minithoracotomy n = 104 . | Sternotomy n = 104 . | P-value . |
---|---|---|---|
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0 .001 |
Female, n (%) | 68 (65) | 38 (36) | <0 .001 |
NYHA III–IV, n (%) | 10 (10) | 19 (18) | 0 .074 |
AF, n (%) | 2 (2) | 11 (11) | 0 .017 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0 .001 |
Associated procedures, n (%) | 12 (11.5) | 6 (5.8) | 0 .1 |
ASD/PFO closure | 3 (2.8) | 3 (2.8) | |
Tricuspid annuloplasty | 9 (8.6) | 3 (2.8) | |
CPB time, min, mean ± SD | 90 ± 19.5 | 71 ± 17.4 | <0 .0001 |
Cross-clamp time, min, mean ± SD | 64 ± 13.6 | 51 ± 10 .8 | <0 .0001 |
. | Minithoracotomy n = 104 . | Sternotomy n = 104 . | P-value . |
---|---|---|---|
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0 .001 |
Female, n (%) | 68 (65) | 38 (36) | <0 .001 |
NYHA III–IV, n (%) | 10 (10) | 19 (18) | 0 .074 |
AF, n (%) | 2 (2) | 11 (11) | 0 .017 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0 .001 |
Associated procedures, n (%) | 12 (11.5) | 6 (5.8) | 0 .1 |
ASD/PFO closure | 3 (2.8) | 3 (2.8) | |
Tricuspid annuloplasty | 9 (8.6) | 3 (2.8) | |
CPB time, min, mean ± SD | 90 ± 19.5 | 71 ± 17.4 | <0 .0001 |
Cross-clamp time, min, mean ± SD | 64 ± 13.6 | 51 ± 10 .8 | <0 .0001 |
NYHA: New York Heart Association; AF: atrial fibrillation; LVEF: left ventricular ejection fraction; ASD: atrial septal defect; PFO: patent foramen ovale; CPB: cardiopulmonary bypass; MR: mitral regurgitation.
. | Minithoracotomy n = 104 . | Sternotomy n = 104 . | P-value . |
---|---|---|---|
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0 .001 |
Female, n (%) | 68 (65) | 38 (36) | <0 .001 |
NYHA III–IV, n (%) | 10 (10) | 19 (18) | 0 .074 |
AF, n (%) | 2 (2) | 11 (11) | 0 .017 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0 .001 |
Associated procedures, n (%) | 12 (11.5) | 6 (5.8) | 0 .1 |
ASD/PFO closure | 3 (2.8) | 3 (2.8) | |
Tricuspid annuloplasty | 9 (8.6) | 3 (2.8) | |
CPB time, min, mean ± SD | 90 ± 19.5 | 71 ± 17.4 | <0 .0001 |
Cross-clamp time, min, mean ± SD | 64 ± 13.6 | 51 ± 10 .8 | <0 .0001 |
. | Minithoracotomy n = 104 . | Sternotomy n = 104 . | P-value . |
---|---|---|---|
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0 .001 |
Female, n (%) | 68 (65) | 38 (36) | <0 .001 |
NYHA III–IV, n (%) | 10 (10) | 19 (18) | 0 .074 |
AF, n (%) | 2 (2) | 11 (11) | 0 .017 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0 .001 |
Associated procedures, n (%) | 12 (11.5) | 6 (5.8) | 0 .1 |
ASD/PFO closure | 3 (2.8) | 3 (2.8) | |
Tricuspid annuloplasty | 9 (8.6) | 3 (2.8) | |
CPB time, min, mean ± SD | 90 ± 19.5 | 71 ± 17.4 | <0 .0001 |
Cross-clamp time, min, mean ± SD | 64 ± 13.6 | 51 ± 10 .8 | <0 .0001 |
NYHA: New York Heart Association; AF: atrial fibrillation; LVEF: left ventricular ejection fraction; ASD: atrial septal defect; PFO: patent foramen ovale; CPB: cardiopulmonary bypass; MR: mitral regurgitation.
Patient baseline characteristics by surgical cohort, pre- and post-IPW weighting
. | Unweighted sample . | Inverse probability—weighted sample . | ||||||
---|---|---|---|---|---|---|---|---|
. | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . |
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0.001 | 1.34 | 40 ± 10 | 40 ± 15 | 0.66 | 0.03 |
Female, % | 65 | 36 | <0.001 | 0.60 | 52 | 54 | 0.84 | 0.04 |
NYHA III–IV, % | 10 | 18 | 0.074 | 0.25 | 17 | 12 | 0.52 | 0.14 |
AF, % | 2 | 11 | 0.017 | 0.33 | 4 | 6 | 0.53 | 0.09 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0.001 | 0.62 | 63 ± 5 | 61 ± 6 | 0.04 | 0.39 |
. | Unweighted sample . | Inverse probability—weighted sample . | ||||||
---|---|---|---|---|---|---|---|---|
. | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . |
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0.001 | 1.34 | 40 ± 10 | 40 ± 15 | 0.66 | 0.03 |
Female, % | 65 | 36 | <0.001 | 0.60 | 52 | 54 | 0.84 | 0.04 |
NYHA III–IV, % | 10 | 18 | 0.074 | 0.25 | 17 | 12 | 0.52 | 0.14 |
AF, % | 2 | 11 | 0.017 | 0.33 | 4 | 6 | 0.53 | 0.09 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0.001 | 0.62 | 63 ± 5 | 61 ± 6 | 0.04 | 0.39 |
Mini: minithoracotomy; Sterno: sternotomy; NYHA: New York Heart Association; AF: atrial fibrillation; LVEF: left ventricular ejection fraction.
Standardized difference: difference in means or proportions divided by standard error; imbalance defined as absolute value greater than 0.20 (small effect size) [1].
P-value based on the Somers' D statistic.
Patient baseline characteristics by surgical cohort, pre- and post-IPW weighting
. | Unweighted sample . | Inverse probability—weighted sample . | ||||||
---|---|---|---|---|---|---|---|---|
. | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . |
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0.001 | 1.34 | 40 ± 10 | 40 ± 15 | 0.66 | 0.03 |
Female, % | 65 | 36 | <0.001 | 0.60 | 52 | 54 | 0.84 | 0.04 |
NYHA III–IV, % | 10 | 18 | 0.074 | 0.25 | 17 | 12 | 0.52 | 0.14 |
AF, % | 2 | 11 | 0.017 | 0.33 | 4 | 6 | 0.53 | 0.09 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0.001 | 0.62 | 63 ± 5 | 61 ± 6 | 0.04 | 0.39 |
. | Unweighted sample . | Inverse probability—weighted sample . | ||||||
---|---|---|---|---|---|---|---|---|
. | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . | Mini n = 104 . | Sterno n = 104 . | P-value* . | Absolute Stand. Diff.a . |
Age, years, mean ± SD | 36 ± 8 | 51 ± 13 | <0.001 | 1.34 | 40 ± 10 | 40 ± 15 | 0.66 | 0.03 |
Female, % | 65 | 36 | <0.001 | 0.60 | 52 | 54 | 0.84 | 0.04 |
NYHA III–IV, % | 10 | 18 | 0.074 | 0.25 | 17 | 12 | 0.52 | 0.14 |
AF, % | 2 | 11 | 0.017 | 0.33 | 4 | 6 | 0.53 | 0.09 |
LVEF, %, mean ± SD | 63 ± 5 | 59 ± 8 | <0.001 | 0.62 | 63 ± 5 | 61 ± 6 | 0.04 | 0.39 |
Mini: minithoracotomy; Sterno: sternotomy; NYHA: New York Heart Association; AF: atrial fibrillation; LVEF: left ventricular ejection fraction.
Standardized difference: difference in means or proportions divided by standard error; imbalance defined as absolute value greater than 0.20 (small effect size) [1].
P-value based on the Somers' D statistic.
Procedural data
All patients underwent an associated annuloplasty with a prosthetic ring (205 pts, 98.5%) or autologous pericardium (3 pts, 1.4%) to reshape and reduce the mitral annulus and stabilize the repair. A Carpentier–Edwards classic complete ring was implanted in 11 patients, a St. Jude Medical Seguin complete semi-rigid ring, in 165 and a St. Jude Medical Tailor flexible posterior ring, in 29 patients. The mean ring size was 36.3 ± 2.19 (median 36, IQR 35–38). Autologous pericardium was used exclusively in 3 patients at the beginning of our experience and then abandoned. Concomitant procedures were performed in 8.7% of cases, including tricuspid valve annuloplasty in 12 patients and suture of an atrial septal defect or patent foramen ovalis in 6 patients. No significant difference was detected between the two groups (P = 0.1). Mean CPB and cross-clamp times were significantly longer in the minithoracotomy group (90 ± 19.5 vs 71 ± 17.4 min, P < 0.0001 and 64 ± 13.6 vs 51 ± 10.8 min, P < 0.0001, respectively). There were no conversions to sternotomy.
Clinical hospital outcomes
There were no in-hospital deaths. No significant differences were observed postoperatively in terms of low cardiac output syndrome (4.8% vs 4.8%, P = 1.0), re-exploration for bleeding (2.8% vs 2.8%, P = 1.0) or pacemaker implantation (1.9% vs 1.9%, P = 1.0). There were no cases of perioperative myocardial infarction, neurological complication, acute renal failure, respiratory failure requiring tracheostomy and sepsis in either group. One patient in the minithoracotomy group required postoperative drainage of a right pleural effusion, and 1 patient in the sternotomy group needed subxiphoid drainage of a pericardial effusion. In the minimally invasive group, 1 patient (1/104, 0.9%) had a retroperitoneal haematoma immediately after the operation, and 1 patient (1/104, 0.9%) developed lymphoedema in a leg on postoperative day 5. Finally, 1 patient (1/104, 0.9%) in the sternotomy group required sternal rewiring.
No significant differences were observed in terms of intensive care unit stay (22.8 ± 11.2 vs 23.1 ± 11.8 h, P = 0.7) and hospital length of stay (5.2 ± 2.2 vs 5.8 ± 2.9 days, P = 0.8) in either group.
Echocardiography at hospital discharge
TTE was performed at discharge for all patients. No or trivial MR was detected in 1 patient in the sternotomy group who had residual 2+ MR (1/104, 0.9%, P = 1.0).
Long-term survival
Follow-up was 99.5% complete (median 11.5 years, IQR 10.4–12.1 vs 10.4 years, IQR 7.0–12.7). At 12 years, overall survival of the total population was 93 ± 2%, which was not significantly different in the minithoracotomy or the sternotomy groups in the unadjusted and in the IPW-adjusted estimates (95% vs 92%, P = 0.19; 96% vs 97%, P = 0.60).
During the follow-up period, 3 (2.9%) deaths were observed in the minithoracotomy group, 2 of which were cardiac related (sudden deaths). In the sternotomy group, 8 patients (7.6%) died. Four of the deaths were of cardiac origin, including 1 case of congestive heart failure with pulmonary oedema; 1 patient died suddenly, whereas 2 suffered acute myocardial infarctions.
The overall unweighted cumulative incidence function (CIF) of cardiac death at 12 years, with non-cardiac death as the competing risk, was 3.9% (95% CI: 0.6–13.3) in minithoracotomy patients vs 3.8% (95% CI: 1.3–8.8) in sternotomy patients. IPW-adjusted CIFs of cardiac death at 12 years were 2.9% (95% CI: 0.001–19.8) in the minithoracotomy group and 1.8% (95% CI: 0.5–4.6) in the sternotomy group (Fig. 1). Using the Pepe–Mori test, no statistically significant differences were detected (P = 0.87). Analysis with the Fine–Gray Model showed that the surgical approach was not a significant predictor of cardiac death (HR 0.88, 95% CI: 0.14–5.37, P = 0.87).

IPW-adjusted cumulative incidence function of cardiac death at 12 years, with non-cardiac death as competing risk. IPW: inverse probability of treatment weighted.
Reoperation
At 12 years, the unadjusted CIF of reoperation, with overall death as competing risk, was 8.5% (95% CI: 3.9–15.4) in the sternotomy and 1.1% (95% CI 0.1–5.3) in the minimally invasive group (P = 0.055). After IPW adjustment, CIF of reoperation, with overall deaths as competing risk, the results with the minithoracotomy group were similar to those with the sternotomy groups (0.9%, 95% CI 0.1–9.6 vs 4.9%, 95% CI 2.3–8.9, P = 0.29) (Fig. 2). Analysis with the Fine–Gray Model indicated that the surgical approach was not a significant predictor of reoperation (HR 0.21, 95% CI 0.03–1.80, P = 0.16). Eleven patients (5.3%) required a new cardiac operation between the first month and 14.2 years after the initial repair (median 7.6 years, IQR 2.8–11.6): 1 (0.9%) in the minithoracotomy group and 10 in the sternotomy group (9.6%). Mitral stenosis requiring a redo operation was detected in 1 patient (0.4%), whereas the remaining 10 patients (4.8%) were all re-operated on because of severe MR. One patient (in the sternotomy group) already had residual moderate (2+/4+) MR at hospital discharge. Recurrent MR in 1 patient was the result of tearing of the leaflet by the EE suture; 2 patients had recurrent leaflet prolapse; and 1 patient had infective endocarditis. In the remaining 7 patients, the mechanism responsible for recurrent MR remained unclear. Ten patients underwent conventional MV replacement, and 1 patient received a re-repair with a new EE.

IPW-adjusted cumulative incidence function of reoperation at 12 years, with overall deaths as competing risk. IPW: inverse probability of treatment weighted.
Recurrence of mitral regurgitation
At 12 years, the unadjusted CIF of recurrent MR ≥ 3+, with death as the competing risk, was 11.4% (95% CI 5.9–18.8) in the sternotomy and 5.8% (95% CI 1.8–13.4) in the minimally invasive group, respectively (P = 0.077). After IPW adjustment, the CIF of recurrent MR ≥ 3 was 7.3% (95% CI 4.1–11.8) in the sternotomy and 5.4% (95% CI 0.6–18.4) in the minimally invasive group (P = 0.30) (Fig. 3). The unadjusted CIF of recurrent MR ≥ 2+, with death as the competing risk, was 22.6% (95% CI 14.0–32.0) in the sternotomy and 14.8% (95% CI 7.3-24.7) in the minimally invasive group (P = 0.20), whereas the IPW-adjusted CIF of recurrent MR ≥ 2+ was 14.8% (95% CI 9.7–21.0) in the sternotomy and 13.5% (95% CI 3.8–29.5) in the minimally invasive group (P = 0.63) (Fig. 4). Analysis with the Fine–Gray model showed that the surgical approach was not significantly associated with recurrence of MR ≥ 3+ (HR 0.52, 95% CI 0.15–1.86, P = 0.32) and of MR ≥ 2+ (HR 1.04, 95% CI 0.45–2.39, P = 0.93).

IPW-adjusted cumulative incidence function of recurrent MR ≥ 3+, with death as competing risk. IPW: inverse probability of treatment weighted.

IPW-adjusted cumulative incidence function of recurrent MR ≥ 2+, with death as competing risk. IPW: inverse probability of treatment weighted.
No or mild recurrent MR was detected in the last echocardiograms in 166 (166/206, 80.5%) patients and moderate MR was detected in 24 (24/206, 11.6%). Recurrence of MR ≥ 3+ occurred in 16 patients (16/206, 7.7%) at 7 ± 3.9 years after the initial repair: 12 in the sternotomy group (12/104, 11.5%) and 4 in the minithoracotomy group (4/102, 3.9%). All patients with recurrent severe (4+/4+) MR were re-operated on.
At the last follow-up examination, mean mitral valve area and gradient were 2.9 ± 0.7 cm2 and 3.3 ± 1.0 mmHg, respectively. Mitral stenosis requiring reoperation was detected in 1 patient (0.4%) about 1 month after surgery due to an atrioventricular obstruction at the subvalvular level. The obstruction was the result of the impingement of the hypertrophic papillary muscles in the orifices of the valve.
DISCUSSION
In patients with Barlow’s disease, mitral repair can be technically demanding and not easily reproducible through a minimally invasive approach. Nevertheless, several surgical techniques have been used through a right minithoracotomy to treat these patients; mid- [8, 9, 15, 16] and long-term results have been described [5]. We previously reported excellent late durability (at 14 years) of the double orifice EE repair performed through a median sternotomy in bileaflet prolapse due to Barlow’s disease [3]. However, no data have been published to date on the long-term results of this technique using a minimally invasive approach. Indeed, the only studies addressing this issue were limited by a mean follow-up period not exceeding 1 year [8–10]. Therefore, we chose to assess whether the use of a right minithoracotomy has an impact on the long-term results of EE repair compared to conventional sternotomy in this setting. For this purpose, we enrolled 104 selected minithoracotomy EE patients from 1999 to 2010. These patients were compared to the same number of sternotomy patients from 1995 to 2000, when the double orifice EE repair was the routine approach in Barlow’s disease in our institution. To reduce as much as possible the number of confounding factors, we excluded patients undergoing minithoracotomy MV repair with a forme fruste of Barlow’s disease, those treated with other concomitant techniques and those with predominant involvement of the commissural region in whom a commissural and not a central EE repair was used.
To provide a meaningful comparison between the two surgical approaches, first we had to adjust for the imbalance of features between minithoracotomy and sternotomy patients using the IPW method. After IPW adjustment, the two groups were well balanced and an adequate match was achieved. Although the LVEF remained slightly higher in the minithoracotomy group, this difference was not considered clinically relevant because it was preserved (>60%) in both groups. No conversion to sternotomy was necessary in the minithoracotomy group. Although the EE technique is a relatively straightforward approach that can be performed with short myocardial ischaemic time, both CPB and cross-clamp times were approximately 20 min longer in the minimally invasive group. The slightly longer procedural times had no impact on the immediate results or on the postoperative course because the overall mean aortic cross-clamp time in the minithoracotomy group was only 64 min. This ischaemic time is significantly shorter than that reported with any other minimally invasive surgical technique (chordal loop, resection, artificial chordae) [9, 15, 16]. These procedures can be particularly time-consuming in patients with multisegment disease. When the EE was compared to the loop technique in patients with Barlow’s disease who had a minimally invasive MV repair, a statistically significant 33% reduction in aortic cross-clamp time was observed [9].
Postoperative morbidity and late complications were low and not significantly different in the two groups. In our opinion, cosmesis remains the only and most important advantage of the minithoracotomy approach. Therefore, it represents our preferred technique in relatively young patients (<55 years old).
We decided not to compare the rate of blood transfusions in the two groups because of the extensive use of Custodiol cardioplegia. In recent years, minithoracotomy patients would have faced a major bias compared to the control group of sternotomy cases who mainly received cold-blood cardioplegia and, therefore, a lower degree of haemodilution.
The minimally invasive approach did not impair the quality of the repair because the only patient who had residual MR 2+ at hospital discharge belonged to the sternotomy group, and all remaining patients had no or trivial residual MR. One strength of this study is that long-term outcomes were assessed on the basis of clinical and echocardiographic data and therefore on the rate of recurrence of MR, not only on freedom from reoperation. Because the duration of follow-up was slightly longer in the sternotomy group, statistical methods were used to compare long-term results, taking into account this issue by considering the number of patients at risk in both groups at each time point. At 12 years, overall survival was similar; CIFs of cardiac death, reoperation and recurrence of MR ≥ 3+ and of MR ≥ 2+ were comparable. In addition, the minithoracotomy approach was not a predictor of cardiac death, reoperation or recurrence of MR. In particular, our echocardiographic data demonstrate for the first time the excellent durability of the minimally invasive EE technique in the long term with an IPW-adjusted CIF of recurrent MR ≥ 3+ at 12 years of 5.4%. Only 4 patients (3.9%) had recurrent MR 3+ or 4+ in the minithoracotomy group without any significant difference when compared with the sternotomy EE patients.
Another issue that is commonly raised with the EE technique is the risk of inducing mitral stenosis. In a direct comparison between the EE and the loop technique performed through a right minithoracotomy in Barlow’s disease, Borger et al. [9] showed that the EE repair results in mildly increased transvalvular gradients and mildly decreased valve opening areas without any difference in residual MR. In that study, however, the valvular haemodynamics following EE repair remained well within the nonstenotic range (i.e. mean gradient, 3.3 mmHg; mean orifice area, 2.8 cm2) for up to 2 years after the operation.
Our long-term echocardiographic data demonstrate that valvular haemodynamics remain stable over time, thereby confirming in this minimally invasive study population similar findings previously observed in the EE sternotomy group [3]. At the last follow-up examination, the mean mitral valve area and gradient were 2.9 ± 0.7 cm2 and 3.3 ± 1.0 mmHg, respectively, and no mitral stenosis was detected. In conclusion, a minimally invasive approach has no negative impact on the effectiveness and long-term durability of the EE repair for bileaflet prolapse in Barlow’s disease. Survival, freedom from cardiac death, reoperation and recurrence of MR were similar in the minithoracotomy and sternotomy groups. Despite the challenges seen in patients with Barlow’s disease, a minimally invasive EE repair is associated with shorter myocardial ischaemic time and excellent long-term outcomes compared with other techniques. Valvular performance remains stable over time with no evidence of mitral stenosis.
Limitations
Our study is retrospective in nature and therefore subject to the inherent weaknesses of a retrospective analysis. The two groups included selected patients who were enrolled consecutively rather than simultaneously, which might have been a potential source of bias. Baseline systolic pulmonary artery pressure was not available in all patients and therefore was not analysed. To minimize any potential bias, we adjusted the baseline characteristics with the IPW statistic and used statistical methods, taking into account the difference in the time frame by considering the number of patients at risk in both groups at each time point. Moreover, we cannot exclude the fact that the rate of recurrent MR might have been underestimated in the patients whose echocardiographic examinations were performed at other institutions. However, the good late outcomes observed on echocardiograms are consistent with the satisfactory clinical conditions described by the patients. Left ventricular dimensions, right ventricular function and systolic pulmonary artery pressure at follow-up were not always available and therefore they could not be analysed. Finally, another limitation of this study is the fact that the causes of recurrent MR remained unknown in most patients, primarily because they were re-operated at other institutions.
Conflict of interest: none declared.
REFERENCES
Author notes
Presented at the 30th Annual Meeting of the European Association for Cardio-Thoracic Surgery, Barcelona, Spain, 1–5 October 2016.
The first two authors contributed equally to this work.