Reoperation for mitral paravalvular leak: a single-centre experience with 200 patients†

Abstract

OBJECTIVES

Paravalvular leak (PVL) is a major cause of morbidity and mortality after mitral valve replacement. Risk factors and data on long-term surgical outcomes are lacking.

METHODS

Between January 1995 and December 2012, 206 [118 males (57%)] patients underwent reoperation due to mitral PVL. Mean age was 64 ± 11 years. Haemolytic anaemia was present in 85 (41%) patients, while 137 (67%) patients were in New York Heart Association Class III or IV. Baseline creatinine was above 1.5 in 91 (44%) patients, while chronic steroids were used in 14 (7%) patients. Active endocarditis was present in 8 (4%) patients. Device occlusion was attempted in 21 (10%) patients.

RESULTS

PVL was most common at the aortomitral curtain (82 patients, 40%). Repair was possible in 105 (51%) patients. Early mortality occurred in 11 (5%) patients. Mean follow-up was 5 (maximum 19) years. Overall survival at 1, 5 and 15 years was 83%, 62% and 16%, respectively. Death due to heart failure or cardiogenic shock occurred in 39 patients. Recurrence occurred in 43 (21%) patients and reoperation was required in 19 patients. Multivariate analysis revealed advanced New York Heart Association class (P < 0.0001), active endocarditis (P = 0.013), chronic steroids (P = 0.022), previous coronary artery bypass grafting (P = 0.026), baseline creatinine above 1.5 (P = 0.001), postoperative need for dialysis (P = 0.036) and residual PVL (P < 0.0001) to be predictors of late mortality. Active endocarditis (P = 0.0004) and chronic steroids (P = 0.002) were significant predictors for PVL recurrence. Freedom from reoperation due to PVL recurrence was 89% and 84%, while freedom from late intervention was 94% and 61% at 5 and 15 years, respectively.

CONCLUSIONS

PVL after mitral valve replacement is associated with increased morbidity and mortality. Re-repair is possible, but recurrent PVL is a risk factor for late mortality, and reoperation should be performed prior to the onset of advanced heart failure.

INTRODUCTION

Paravalvular leak (PVL) after mitral valve (MV) replacement (MVR) is not uncommon with an estimated incidence of 0.2–1.4% per year [1]. Patients’ presentations range from being asymptomatic with mild PVL to haemolysis and heart failure in the presence of severe leak [2]. Percutaneous device closure has emerged as an alternative to surgery in select patient population; however, surgery is still considered the gold standard therapy. Repeat operations for PVL have been associated with high mortality and significant morbidities [3]. There are limited data on long-term outcomes after reoperation for mitral PVL. Our goal in this study is to report our early and long-term outcomes of repeat operation for mitral PVL.

MATERIALS AND METHODS

We conducted a retrospective review of a prospectively collected database to identify all patients who underwent surgery for mitral PVL at our institution. Mayo Foundation Institutional Review Board approved this study, and informed consent was obtained from patients for their participation in the research. From January 1995 through December 2012, 206 patients underwent repeat operation for mitral PVL at our clinic. We included patients who had concomitant coronary artery bypass grafting (CABG) or surgery for concomitant aortic and or tricuspid valve disease. Medical records were reviewed to obtain demographic characteristics, associated medical conditions, details of operations and results of the echocardiographic examinations. All patients underwent preoperative angiography and transoesophageal echocardiography to determine the degree of severity and location of the mitral PVL. PVL was classified as mild, moderate or severe according to echocardiography. Only those who had moderate or more degree of PVL underwent reoperation.

Indications for surgery were as follows: (i) symptoms of heart failure and (ii) haemolytic anaemia requiring repeat blood transfusions.

Late survival was determined for all patients from their medical records, Accurint (www.accurint.com) and surveys. Late clinical status was determined by responses to questionnaires that are sent to patients 1, 3, 5, 7, 10 and 15 years after operation. This was supplemented by information from medical records when new information was available as well as written correspondence or telephone conversations with the patients or their physicians, or both.

Statistical analysis

Preoperative patient characteristics included age, gender, body mass index, preoperative ejection fraction, location and degree of PVL, preoperative dialysis, preoperative CABG, previous attempt at percutaneous device closure and preoperative stroke. Intraoperative findings included the location of the PVL, the type of surgery performed and the type of prosthesis used. Postoperative complications included the need for new dialysis and a new stroke. The main outcomes in this study were 30-day mortality, 30-day unplanned reoperation, overall long-term survival and freedom from reoperation or recurrence of PVL. We used multiple logistic regression models to assess associations with the short-term outcomes. The models were built a priori based on clinical knowledge and included the independent variable mentioned above. For long-term survival, we used Kaplan–Meier survival analyses to describe 20-year survival by key variables and comparisons made by the log-rank test for equality and then used multivariable Cox proportional hazards models to identify risk factors associated with long-term survival. All statistical tests were 2-sided, and P–values <0.05 were considered statistically significant. Statistical analyses were conducted using STATA special edition software (version 13.1; StataCorp, College Station, TX, USA).

RESULTS

Patients’ characteristics

Between January 1995 and December 2012, 206 patients [118 males (57%)] underwent reoperation due to mitral PVL. Mean age was 64 ± 11 years. Baseline characteristics of the study group are presented in Table 1. The indications for surgery were based on symptoms of heart failure, presence of haemolytic anaemia and the need for repeat blood transfusions. Haemolytic anaemia was present in 85 (41%) patients, while 137 (67%) patients were in New York Heart Association (NYHA) Class III or IV. Preoperative cardiogenic shock was present in 7 (3%) patients and 6 patients needed an intra-aortic balloon pump prior to surgery. Baseline creatinine was above 1.5 in 91 (44%) patients, with 2 patients being dialysis dependent, while chronic steroids were used in 14 (7%) patients. Active endocarditis was present in 8 (4%) patients. Transcatheter device closure was attempted in 21 (10%) patients (Fig. 1A and B). Concomitant aortic PVL was present in 16 patients who underwent previous aortic and MVR. PVL occurred after previous mechanical mitral prostheses in 164 (80%) patients, while 42 (20%) patients had an initial bioprosthesis. The majority of patients (127, 62%) had one episode of PVL, while recurrent PVL occurred in the remaining 79 (38%) patients (48 patients had 2 PVL, 23 patients had 3 PVL, 6 patients had 4 PVL and 2 patients had 5 PVL). The leak was severe in 179 (87%) patients and moderate in the remaining 27 (13%) patients.

Surgical technique

Right thoracotomy was performed in 14 (7%) patients, while repeat sternotomy was the main approach in the majority of patients with 33 (16%) patients having had more than 3 sternotomies. Techniques of exposure of the mitral prosthesis varied with the standard left atriotomy being the most common in the majority of patients (100, 49%), followed by vertical trans-septal in 85 (41%) patients, extended vertical trans-septal in 19 (9%) patients and Dubost approach in the remaining 2 (1%) patients. Severe mitral annular calcifications were present in 48 (23%) patients. Standard cardioplegic arrest was the standard, while fibrillatory arrest was utilized only in 10 (5%) patients. Five (2%) patients required concomitant circulatory arrest. The mean aortic cross-clamp and cardiopulmonary bypass times were 70 ± 44.6 and 118 ± 60.4 min, respectively.

The most common location of the PVL was at the aortomitral curtain (82 patients, 40%), followed by the medial annulus (53, 26%) and along the posterior annulus in 28 (14%) patients. The location was unclear in 17 (8%) patients and more than 1 regurgitant jet was present in 26 (13%) patients. Repair of the PVL was possible in 105 (51%) patients. Several techniques and exposure strategies have been utilized for repair, such as synthetic patch repair (4 patients), 2-layer suture repair (16 patients) and single-layer pledgeted sutures (85 patients) (Fig. 2A–G). MV re-replacement was necessary in the remaining 101 (49%) patients, and the most common prosthesis used for re-replacement was mechanical (82 patients, 40%).

Early outcome

Early mortality occurred in 11 (5%) patients. Predictors of early mortality are presented in Table 2. Mechanical circulatory support with an intra-aortic balloon pump was needed in 20 (10%) patients. Postoperative renal failure needing dialysis was required in 18 (9%) patients. Prolonged ventilation needing tracheostomy occurred in 12 (6%) patients. Postoperative complete heart block occurred in 23 (11%) patients, and they underwent permanent pacemaker placement. The mean length of hospital stay was 14 ± 10.5 (maximum 60) days.

Follow-up and late outcome

The mean follow-up was 5 (maximum 19) years. Overall survival at 1, 5 and 15 years was 83%, 62% and 16%, respectively (Fig. 3A–D). Death due to heart failure or cardiogenic shock occurred in 39 (19%) patients. Recurrence of PVL occurred in 43 (21%) patients. Transcatheter device closure was performed in 14 (6%) patients, and reoperation was required in 19 patients. We do not have information on residual PVL at the time of repair.

Univariate analysis (Table 3) showed recurrent PVL (P < 0.001), late intervention with transcatheter devices (P < 0.001), late reoperation for recurrent PVL (P = 0.021) and even residual PVL (P < 0.001) to be significant risk factors for late death. On the multivariate level (Table 4), residual PVL (P < 0.0001) and late reoperation due to recurrent PVL (P = 0.019) were significant. Other predictors were advanced NYHA class (P < 0.0001), active endocarditis (P = 0.013), chronic steroids (P = 0.022), previous CABG (P = 0.026), baseline creatinine above 1.5 (P = 0.001), concomitant tricuspid valve surgery (P = 0.03) and postoperative need for dialysis (P = 0.036). There was no difference between repair of PVL and MV re-replacement in terms of recurrence of PVL (P = 0.29).

Active endocarditis (P = 0.0004) and chronic steroids (P = 0.002) were significant predictors for PVL recurrence.

Freedom from reoperation due to PVL recurrence was 93%, 89%, 84% and 84%, while freedom from late intervention was 97%, 94%, 86% and 61% at 1, 5, 10 and 15 years, respectively (Fig. 4A–D).

DISCUSSION

Despite the advances in valve replacement techniques, periprosthetic regurgitation or PVL continues to be a serious problem that leads to increased morbidity and mortality. The reported incidences are about 7–17% in MVR [4]. Several known risk factors have been associated with PVL, such as severe annular calcifications, endocarditis and tissue friability. Patients may be seen with heart failure, symptomatic haemolysis or both [5]. Options for treatment include both reoperation and transcatheter techniques. The literature is rich with case reports and retrospective studies about the use of device closure for PVL, which is currently being performed off-label; however, very little has been published on surgical outcomes [6].

Repeat operations have been associated with high mortality and significant morbidities [7] and part of this may be related to late referral and delayed surgery. The majority of these patients at the time of surgery are already in NYHA Class III or IV and have been hospitalized preoperatively due to severe heart failure and the need for transfusion secondary to significant clinical haemolysis. In this study, haemolytic anaemia was present in 85 (41%) patients, while 137 (67%) patients were in NYHA Class III or IV. In fact, it is not uncommon for these patients to require some sort of mechanical circulatory support prior to surgery, most commonly an intra-aortic balloon pump due to cardiogenic shock (occurred in 3% of the study population).

Taramasso et al. [8] addressed the issue of PVL in relation to both mitral and aortic prostheses. Their retrospective study included 122 consecutive patients (82 with mitral PVL and 40 with aortic PVL) who underwent reoperation for symptomatic PVL. The mean age was 62 ±11 years. Symptomatic haemolysis was present in 31% of the patients, and most of these patients (60%) were in NYHA Class III or IV. Repair of the PVL was feasible in most cases (79 patients or 65%), whereas reoperative replacement was needed in the remaining cases (43 patients or 35%). The early mortality was high (10.7%), and it was even higher for the mitral prostheses than for the aortic prostheses (13% for mitral PVL vs 5% for aortic PVL). Multivariate analysis identified preoperative chronic renal failure and more than 1 previous cardiac reoperation as risk factors for late mortality. In this study, we had a total of 206 patients who underwent repeat operation for mitral PVL only. Early mortality was 5%, which is significantly lower than what is reported in the literature. There were no specific factors that were identified for early mortality on the multivariate analysis; however, the long-term survival is disappointing but similar to the published literature. In the study by Taramasso et al., the overall actuarial survival at 12 years was less than 40% of their patients. However, this can be explained by the multiple comorbidities and the need for repeat sternotomies that exist in these patients even with their relatively younger age (current study mean age was 64 ± 11 years).

Our overall survival at 1, 5 and 15 years was 83%, 62% and 16%, respectively. Death due to heart failure or cardiogenic shock occurred in 39 (19%) patients. On the multivariate level, residual PVL (P < 0.0001) and late reoperation due to recurrent PVL (P = 0.019) were significant predictors for late mortality as were advanced NYHA class (P < 0.0001), active endocarditis (P = 0.013), chronic steroids (P = 0.022), previous CABG (P = 0.026), baseline creatinine above 1.5 (P = 0.001), concomitant tricuspid valve surgery (P = 0.03) and postoperative need for dialysis (P = 0.036). Bouhout et al. [9] reported their long-term results after surgical treatment of PVL in the aortic and mitral position. Their study included 190 patients, in whom 120 (63%) patients had mitral PVL. The mean age at surgery was 63 ± 12 years. The operative mortality occurred in 13 (7%) patients. Survival at 1, 5 and 10 years was 85%, 73% and 56%, respectively. They identified the number of previous surgeries as predictor of survival and PVL recurrence.

In this study, repair of the PVL was possible in 105 (51%) patients, and we have utilized several techniques for repair such as synthetic patch repair (4 patients), 2-layer suture repair (16 patients) and single-layer pledgeted sutures (85 patients). MV re-replacement was necessary in the remaining 101 (49%) patients, and the most common prostheses used for re-replacement were mechanical (82 patients, 40%). We did not identify a significant difference between repair of the PVL and MV re-replacement or difference in recurrence of PVL based on the type of prosthesis used. In this study, mechanical mitral prostheses were used in 164 (80%) patients, while 42 (20%) patients had an initial bioprosthesis. We do believe it may be related to the era of surgery where mechanical valves were commonly used specially in younger patients as in the current cohort. Currently, there are more bioprostheses being used. In a similar study by Choi et al. [10], the authors compared repair versus re-replacement for surgical correction of mitral PVL. There were a total of 52 patients included, and there was no difference between the 2 groups.

Transcatheter techniques have been utilized for PVL closure, but it is hard to know what the best management option is. Several unanswered questions remain, such as which patient should be referred for surgery and which one should be treated with percutaneous device closure remains unknown? From our experience, however, we do believe that percutaneous device closure has several limitations and they do not work that well in the presence of previous mechanical prostheses due to the risks associated with interference with leaflet mobility. We do believe if the patient has previous mechanical prosthesis, multiple PVL and/or are good candidates for surgery, then they should be referred to surgery. And should patients be referred for surgery sooner? There are limited data that compare outcomes between surgery and transcatheter closure of PVL and the published series usually include a heterogenous group of patients.

In another study by Taramasso et al. [11], the authors compared conventional surgery and transcatheter closure for PVL. The study included 139 patients, 122 (87%) patients underwent surgery and 17 (12%) patients underwent transcatheter closure. Mitral PVL was present in 68% of these patients. The in-hospital mortality was 9.3% and occurred in those who underwent surgery. The authors identified surgery as a risk factor for in-hospital death. They reported 98% acute procedural success and 9.3% in-hospital mortality in the surgical group versus no deaths in the transcatheter group. But were the groups really equal? They also reported less than moderate residual valve regurgitation in all patients, but is this how we define success? There is no doubt that this is a high-risk group of patients often referred late for surgery because of the fear of reoperation. This may be unfair comparison.

In looking at 1 of the 2 largest series for device closure of PVL, which comes from the Mayo Clinic [12] and includes 30-day outcomes of 115 patients, the technique was successful only in 77% of the patients, and 15% of the patients had multiple defects. The study had 3-year follow-up data, and the surprising findings were that there were no differences in survival according to the degree of residual regurgitation, even if it was severe, and that haemolytic anaemia was difficult to resolve among survivors and was an independent predictor for poor survival and need for cardiac surgery.

In this study, we identified recurrent PVL, late interventions or reoperations for recurrent PVL and even residual PVL to be associated with poor long-term outcome. We believe this should be taken seriously and recurrent PVL or residual PVL should not be ignored. Of our cohort, recurrence of PVL occurred in 43 (21%) patients. Transcatheter device closure was performed in 14 (6%) patients, and reoperation was required in 19 patients.

Several other important points are worth discussing: should the indications for reoperation be based on the presence of symptoms, or should these patients be referred earlier for surgery even if they are free of symptoms? The majority of published data share the high-risk profile of the PVL patients, which of course raises the risk of any type of intervention dramatically. If patients are referred to surgery earlier prior to the development of heart failure and advanced NYHA class, that risk may change, but it is unknown. The concept of considering most PVL as being benign may not be entirely accurate and only results in delaying interventions. This starts with proper evaluation, which may require combination of imaging modalities such as echocardiography, magnetic resonance imaging and angiography, especially in cases where there is discrepancy in the assessment. But does the degree of PVL really matter? Or does the mere presence of a leak mandate intervention? In recent data from the Placement of Aortic Transcatheter Valves (PARTNER) trial, even mild PVL was associated with a significant risk of mortality [13].

Another dilemma is what to do with patients who had multiple recurrent PVL leaks? In our study, we had patients who had more than 3 recurrences of PVL (31 patients, 15%). This is a particularly difficult group of patients, and it is hard to know the right answer for them. Finally, what are the outcomes of patients who had failure of device closure and had to undergo reoperation? They may be worse; we do not know. We could not answer this question from this study due to the small number of patients who underwent preoperative attempt at device closure (21 patients, 10%).

Limitations

This is a retrospective review of a single centre. The denominator of the number of prostheses is not well known, not all patients underwent their initial valve replacement at our institution. So the true incidence of PVL remains uncertain. Also, those patients with mild PVL were not included in this study. Follow-up was complete only in 90% of our patients. We were not able to have a control group with patients who underwent percutaneous device closure at the time of the study. It remains important to identify which subset of patients should be referred for surgery and which one is better served with percutaneous device closure.

CONCLUSION

PVL after MVR is associated with increased morbidity and mortality. PVL should be aggressively addressed, and to ensure good outcomes, referral of the patient to the surgeon should not be delayed. Re-repair is possible but recurrent PVL is a risk factor for late mortality, and reoperation should be performed prior to the onset of advanced heart failure.

Conflict of interest: none declared.

REFERENCES

Author notes