Mitral Valve Repair Versus Replacement in Simultaneous Mitral and Aortic Valve Surgery for Rheumatic Disease

Mitral Valve Repair Versus Replacement in Simultaneous Mitral and Aortic Valve Surgery for Rheumatic Disease Kenji Kuwaki, MD, PhD, Nobuyoshi Kawaharada, MD, PhD, Kiyofumi Morishita, MD, PhD, Tetsuya Koyanagi, MD, Hisayoshi Osawa, MD, PhD, Toshiyuki Maeda, MD, and Tetsuya Higami, MD, PhD Department of Thoracic and Cardiovascular Surgery, Sapporo Medical University School of Medicine, Sapporo, Japan Background. The purpose of this study was to compare the late results of combined mitral valve repair and aortic valve replacement with double valve replacement for patients with rheumatic heart disease. Methods. From 1981 to 2003, 128 patients underwent aortic valve replacement with either mitral valve repair (n 47) or mitral valve replacement (n 81) for rheumatic disease. Mean follow-up was 9.1 4.5 years. Results. Rates of actuarial freedom from cardiac-related death (81.4% versus 75.9% at 12 years; p 0.60), thromboembolism (79.8% versus 85.1% at 12 years; p 0.78), and bleeding (97.3% versus 95.7% at 12 years; p 0.77) were similar in both combined mitral valve repair and aortic valve replacement and double valve replacement. However, freedom from mitral valve reoperation was significantly lower in combined mitral valve repair and aortic valve replacement compared with double valve replacement (52.6% versus 76.8% at 12 years; p 0.002). Mitral valve repair (p 0.002) and mitral bioprosthesis (p 0.0001) were independent risk factors for mitral valve reoperation. Conclusions. Potential advantages of preserving, rather than replacing, the native mitral valve, such as better cardiac survival or fewer thromboembolic complications, were not identified in combined mitral valve repair and aortic valve replacement compared with double valve replacement for patients with rheumatic disease. Indeed, combined mitral valve repair and aortic valve replacement was associated with a significantly higher incidence of mitral valve reoperation. Therefore, in double valve surgery for rheumatic disease, mitral valve repair should be limited to the correction of mitral valve lesions only when excellent durability can be expected. (Ann Thorac Surg 2007;83:558 63) 2007 by The Society of Thoracic Surgeons Whether mitral valve repair (MVP) or replacement (MVR) in simultaneous mitral and aortic valve surgery will be associated with a better result remains uncertain. No specific recommendations for treatment, including surgical strategy, of multiple valve disease has been developed in the American College of Cardiology/ American Heart Association practice guidelines for the management of patients with valvular heart disease because there are remarkably few data relating to mixed valve disease [1]. Some groups [2, 3] have recommended MVP plus aortic valve replacement (AVR) as the preferred strategy in mixed patient populations with double valve disease, whenever MVP is possible. However, other groups have advocated double valve replacement (DVR) as the standard procedure [4, 5]. Among several causes of double valve disease, rheumatic disease remains the predominant cause even in reports from Western countries [2, 6, 7]. The aims of this study were to compare the late results of Accepted for publication Aug 9, 2006. Address correspondence to Dr Kuwaki, Department of Thoracic and Cardiovascular Surgery, Sapporo Medical University School of Medicine, South 1, West 16, Chuo-Ku, Sapporo, 060-8543, Japan; e-mail: kuwaki@ sapmed.ac.jp. MVP AVR with those of DVR in patients with rheumatic disease, and to identify the independent clinical predictors of late adverse events in all patients. Patients and Methods Patients Between 1981 and 2003, 128 patients underwent simultaneous mitral and aortic valve surgery for rheumatic disease at Sapporo Medical University Hospital. This cohort excluded 26 patients who underwent DVR involving re-mvr after failure of a mitral valve prosthesis implanted previously. This study was approved by the Medical Ethics Committee in Sapporo Medical University. An MVP AVR was performed in 47 patients (37%), and a DVR was performed in 81 patients (63%). Baseline preoperative and intraoperative characteristics of the study cohort are presented in Table 1. Mean follow-up time for MVP AVR and DVR was 9.1 4.2 (range, 2 months to 17 years) and 9.1 4.6 years (range, 3 months to 19 years), respectively (no significant difference). The cumulative follow-up for MVP AVR and DVR was 372 and 653 patient-years, respectively. Closing interval of follow-up was between January 2004 and November 2007 by The Society of Thoracic Surgeons 0003-4975/07/$32.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2006.08.015

Ann Thorac Surg KUWAKI ET AL 2007;83:558 63 MITRAL VALVE REPAIR VERSUS REPLACEMENT IN DOUBLE VALVE SURGERY 559 Table 1. Baseline Preoperative and Intraoperative Characteristics MVP AVR DVR N 47 81 Mean age (y) 48 10 53 8 0.0006 a Male gender 14 34 0.16 Preop NYHA IV 6 12 0.74 Previous MV repair 1 16 0.004 a Sinus rhythm 5 7 0.52 CI 2, L min 1 m 2 7 20 0.29 Mitral stenosis 39 44 0.001 a Mitral regurgitation 4 5 0.72 MSR 4 32 0.0002 a SevereTR (requir TAP) 8 33 0.005 a Renal failure (Cr 2) 1 1 0.51 DM 1 6 0.43 History of stroke or TIA 4 9 0.99 AXT 180 min 0 11 0.007 a CPB 240 min 0 9 0.02 a AVR aortic valve replacement; AXT aortic cross-clamp time; CI cardiac index; CPB cardiopulmonary bypass time; Cr creatinine; DM diabetes mellitus; DVR double valve replacement; MSR mixed mitral valve stenosis and regurgitation; MV mitral valve; MVP mitral valve repair; NYHA Yew York Heart Association functional class; Preop preoperative; TAP tricuspid valve annuloplasty; TIA transient ischemic attack; TR tricuspid valve regurgitation. 2004. Seven patients (3 in MVP AVR and 4 in DVR) were lost to follow-up. The completeness of follow-up was 95%. Complications were defined in accordance with the published guidelines for reporting morbidity and mortality after cardiac valvular operation [8]. Operative Procedures All procedures were carried out through a median sternotomy using cardiopulmonary bypass with moderate systemic hypothermia (27 30 C). Myocardial protection was achieved with antegrade or combined antegrade and retrograde cold blood cardioplegia with topical ice-slush cooling to protect the myocardium during cardiac arrest. The mitral valve prostheses included 66 mechanical valves (DuroMedics, n 23; CarboMedics, n 19; St. Jude Medical, n 16; Bicarbon, n 4; MIRA, n 4) and 15 biological valves (Liotta, n 13; Mitroflow, n 1; Carpentier-Edwards, n 1). Although chordal-sparing MVR was carried out in the majority of patients, we were unable to identify with certainty which patients underwent chordal-sparing MVR from the operative records. Mitral valve repair procedures consisted of open mitral commissurotomy in 44 patients and commissural annuloplasty in 5. Concomitant procedures included DeVega tricuspid annuloplasty in 41 patients, Maze in 7, and coronary artery bypass grafting in 3. Aortic valve prostheses included 128 mechanical valves (Omniscience, n 44; Omnicarbon, n 32; CarboMedics, n 28; St. Jude Medical, n 17; Bicarbon, n 4; ATS, n 2; Lillehei Kaster, n 1). Postoperative Anticoagulation Anticoagulation with warfarin potassium was commenced 2 to 3 days after the operation. Between 1981 and 1997, a thrombo-test (TT) was mainly used for monitoring warfarin anticoagulation therapy; the target level of TT was between 10% and 25%, which corresponded to an international normalized ratio for prothrombin time (INR) of 1.5 to 3.0. Since 1998, the INR has mainly been used to control warfarin anticoagulation; the target level of INR was between 2.0 and 3.0 for MVP AVR patients and between 2.5 and 3.0 for DVR patients. Statistical Analysis Statistical analysis was performed using StatView 5.0 software (Abacus Concept Inc, Berkeley, CA). Timedependent events were evaluated using a univariate analysis of log-rank test and a stepwise Cox proportional hazards multivariable model. s with a probability value less than 0.2 on univariate analysis were then entered into a multivariable analysis. A probability value less than 0.05 was considered statistically significant. s examined by multivariable analysis to identify the independent risk factors for late adverse events in all patients included the following: age older than 55 years, sex, New York Heart Association functional class IV, previous MVP, preoperative stroke or transient ischemic attack, mitral valve regurgitation (MR), mitral valve stenosis (MS), mixed MS and MR, concomitant tricuspid valve procedure, atrial fibrillation, preoperative cardiac index less than 2.0 L min 1 m 2, MVP AVR, DVR, bioprosthetic MVR, years of operation, renal dysfunction (serum creatinine 2.0 mg/dl), diabetes mellitus, aortic cross-clamp time greater than 180 minutes, and cardiopulmonary bypass time greater than 240 minutes. Survival and event-free curves were described using the Kaplan Meier method. Probability estimates were expressed as mean standard error of the mean, and continuous data were expressed as mean standard deviation. We also performed a propensity score analysis to examine the potential selection bias associated with the choice of mitral valve procedure (MVP versus MVR), and to evaluate the effect of the potential bias on late outcome. s examined by logistic regression analysis in terms of selection of the surgical procedure on the mitral valve (MVP versus MVR) included the following: age older than 55 years, preoperative New York Heart Association class IV, previous MVP, preoperative stroke or transient ischemic attack, MR, MS, mixed MS and MR, concomitant tricuspid valve procedure, atrial fibrillation, preoperative cardiac index less than 2.0 L min 1 m 2, years of operation (1981 1990), renal dysfunction (serum creatinine 2.0 mg/dl), and diabetes mellitus. The significant independent predictors associated with the surgical procedure on the mitral valve (MVP versus MVR) identified by multivariable logistic regression were previous history of MVP (which was associated with MVR), age older than 55 years at operation (which was associated with MVR), and mixed MS and MR (which

560 KUWAKI ET AL Ann Thorac Surg MITRAL VALVE REPAIR VERSUS REPLACEMENT IN DOUBLE VALVE SURGERY 2007;83:558 63 Table 2. Independent Predictors for the Surgical Procedure (Repair Versus Replacement) on the Mitral Valve Logistic Coefficient SE Previous history of mitral 2.48 1.0 0.003 a valve repair Age more than 55 years 1.26 0.45 0.004 a Mixed mitral valve stenosis 1.98 0.93 0.03 a and regurgitation C-statistics 0.79 SE standard error. was associated with MVR; Table 2). C-statistics in this model was 0.79 (Table 2). Propensity score was used for the multivariable analysis for adjustment of a potential selection bias of a surgical procedure on the mitral valve. Results Early Mortality Early death was defined as death occurring within 30 days of surgery or as an in-hospital death at any time after surgery. Early mortality was 6.3% in MVP AVR and 6.1% in DVR. The causes of early deaths in MVP AVR included low cardiac output in 3 patients, and those in DVR included gastrointestinal bleeding, left ventricular rupture, cerebral hemorrhage, cerebral infarction, and arrhythmia in 1 patient each. Left ventricular rupture occurred in a 68-year-old woman who underwent DVR for mixed rheumatic mitral (mixed MS and MR) and aortic valve disease (aortic stenosis). The patient was weaned from cardiopulmonary bypass uneventfully, but severe bleeding occurred from the back of the heart and was identified from the atrioventricular groove in the posterior wall of the left ventricle. A major attempt was Table 3. Independent Predictors for Cardiac-Related Deaths Univariate Analysis Multivariable Analysis Hazard Ratio 95% CI Male sex 0.09 0.07 2.21 0.84 5.81 NYHA class IV 0.004 a 0.009 a 4.04 1.41 11.56 Propensity score 0.82 CI confidence interval; functional class. NYHA New York Heart Association made to repair the rupture, but the bleeding was not controlled and the patient eventually died. Late Mortality and Cardiac-Related Death Late death occurred in 6 patients (1.6%/patient-year) in MVP-AVR and 19 (2.9%/patient-year) in DVR. In operative survivors, the actuarial overall survival rates at 12 years in MVP AVR and in DVR were 78.6% 8.6% and 68.4% 7.1%, respectively (no significant difference, p 0.23). The causes of late deaths in MVP AVR were cardiac-related in 5 patients and malignancy in 1, and those in DVR were cardiac-related in 14 patients, malignancy in 3, aortic aneurysm in 1, and pneumonia in 1. Cardiac-related deaths in MVP AVR included early death after repeat heart valve operation in 2 patients, cerebral infarction in 1, and cerebral bleeding and unknown in 1 patient each, and those in DVR included cardiac failure in 6 patients, sudden death in 3, early death after repeat heart valve operation in 3, and cerebral infarction in 2. Actuarial freedom from cardiac-related death at 12 years in MVP AVR and in DVR was 81.4% 8.2% and 75.9% 7.0%, respectively (no significant difference, p 0.60; Fig 1). Preoperative New York Heart Association functional class IV was identified by multi- Fig 1. Freedom from cardiac-related death in mitral valve repair plus aortic valve replacement (MVP AVR; solid line) and double valve replacement (DVR; dashed line) patients. No statistically significant difference was seen between the two groups. Fig 2. Freedom from mitral valve reoperation in mitral valve repair plus aortic valve replacement (MVP AVR; solid line) and double valve replacement (DVR; dashed line) patients. A statistically significant difference (52% versus 76% at 12 years; p 0.002) was seen between the two groups.

Ann Thorac Surg KUWAKI ET AL 2007;83:558 63 MITRAL VALVE REPAIR VERSUS REPLACEMENT IN DOUBLE VALVE SURGERY 561 Table 4. Independent Predictors for Mitral Valve Reoperation Univariate Analysis Multivariable Analysis Hazard Ratio 95% CI Male gender 0.02 a 0.08 0.34 0.14 0.85 Previous MVP 0.17 0.55 0.38 0.09 1.61 Concomitant 0.04 a 0.08 2.53 0.97 6.58 tricuspid procedure Cardiac index 2.0 0.11 0.19 2.27 0.79 6.53 L/min/m 2 MVP AVR 0.002 a 0.004 a 2.93 1.42 5.97 Bioprosthetic MVR 0.0001 a 0.0001 a 10.3 4.48 24.06 Propensity score 0.15 AVR aortic valve replacement; CI confidence interval; MVP mitral valve repair; MVR mitral valve replacement. variable analysis as an independent predictor of cardiacrelated deaths (p 0.009; 95% confidence interval, 1.4 to 11.6; hazard ratio, 4.0; Table 3). Mitral Valve Reoperation Twenty-one patients (5.65%/patient-year) in MVP AVR required mitral valve reoperation. At reoperation, all 21 patients underwent MVR. Progression of rheumatic disease in the repaired mitral valve was the reason for reoperation in all patients except for 1 patient who required MVR because of MR 2 months after open mitral commissurotomy. Repair failure would be the reason for MR in this patient. Mitral valve lesions in 20 patients at reoperation were mitral valve restenosis in 19 patients and combined mitral valve restenosis and regurgitation in 1. Twelve patients (1.83%/patient-year) in the DVR group required mitral valve reoperation. Primary tissue Fig 3. Freedom from thromboembolism in mitral valve repair plus aortic valve replacement (MVP AVR; solid line) and double valve replacement (DVR; dashed line) patients. No statistically significant difference was seen between the two groups. Table 5. Independent Predictors for Combined Thromboembolic and Bleeding Events Univariate Analysis Multivariable Analysis Hazard Ratio 95% CI Age 55 years old 0.19 0.26 0.62 0.22 1.73 Male sex 0.09 0.11 2.21 0.78 6.29 Cardiac index 2.0 0.13 0.14 2.58 0.57 11.66 L min 1 m 2 Propensity score 0.71 CI confidence interval. failure in the mitral bioprosthesis was the reason for reoperation in 10 patients, and thrombosis and pannus formation on a mechanical mitral prosthesis were reported in 1 patient each. All patients underwent repeat MVR with a mechanical prosthesis. Actuarial freedom from mitral valve reoperation in MVP AVR was significantly lower than that in DVR (52.6% 9.2% versus 76.8% 6.1% at 12 years; p 0.002; Fig 2). The surgical procedures of MVP AVR (p 0.002; 95% confidence interval, 1.4 to 5.9; hazard ratio, 2.9) and mitral bioprosthesis (p 0.0001; 95% confidence interval, 4.5 to 24.1; hazard ratio, 10.3) were identified by multivariable analysis as independent predictors of mitral valve reoperation (Table 4). The mortality of mitral valve reoperation in MVP AVR and DVR was 4.7% (1 of 21) and 0% (0 of 12), respectively. Thromboembolism and Bleeding There were 5 thromboembolic events (1.34%/patientyear) in MVP AVR and 10 (1.53%/patient-year) in DVR. Thromboembolic events in MVP AVR included cerebral infarction in 2 patients, transient ischemic attack in 2, and acute occlusion of a peripheral artery in 1, and those in DVR included cerebral infarction in 6, transient ischemic attack in 2, acute occlusion of a peripheral artery in 1, and thrombosis of a mechanical mitral prosthesis in 1. Actuarial freedom from thromboembolic events at 12 years in MVP AVR and in DVR was 79.8% 8.8%/ patient-year and 85.1% 5.8%/patient-year, respectively (no significant difference, p 0.78; Fig 3). Bleeding events requiring hospitalization occurred in 1 patient (0.27%/patient-year) in MVP AVR and 3 (0.46%/patient-year) in DVR. Bleeding events in MVP AVR included intracranial hemorrhage in 1 patient and those in DVR included intracranial hemorrhage in 2 patients and hemoptysis associated with pneumonia in 1. Actuarial freedom from bleeding events at 12 years in MVP AVR and in DVR was 97.3% 2.7%/patient-year and 95.7% 4.3%/patient-year, respectively (no significant difference, p 0.77). These excellent results in terms of bleeding complication in both MVP AVR and DVR might be explained by the relatively lower target level of INR in our series. Combined thromboembolic and bleeding events occurred in 6 patients in MVP AVR (1.61%/patient-year)

562 KUWAKI ET AL Ann Thorac Surg MITRAL VALVE REPAIR VERSUS REPLACEMENT IN DOUBLE VALVE SURGERY 2007;83:558 63 and 13 in DVR (1.99%/patient-year). No independent predictor for combined thromboembolic and bleeding events was identified in this study (Table 5). Comment The present study demonstrated that MVP AVR was associated with similar long-term survival, bleeding, and thromboembolic complications when compared with DVR in patients with rheumatic double valve disease. However, MVP AVR resulted in a significantly higher percentage of patients who required late mitral valve reoperation because of progression of underlying rheumatic mitral valve disease. Survival In the present study of double valve surgery for rheumatic valve disease, no survival advantage of MVP AVR over DVR was seen; the survival rate at 12 years from cardiac-related death was 81.4% and 75.9% for MVP AVR and DVR, respectively (p 0.60). Previous reports comparing MVP AVR and DVR in mixed patient populations also demonstrated no statistically significant difference in survival between these procedures [4 6]. These previous authors documented 10-year survival rates of 64% to 92% after MVP AVR and 87% to 90% after DVR. Yau and colleagues [9] found that in patients with isolated rheumatic mitral disease, better late cardiac survival was independently predicted by MVP than by MVR. They demonstrated that MVR with a mechanical prosthesis minimized reoperation, but limited survival and increased thromboembolic complications. The poorest cardiac survival in patients who had undergone MVR with a mechanical valve was attributed largely to thromboembolic events and the bleeding complications of long-term anticoagulation. However, in double valve surgery in relatively young patients with rheumatic disease, a mechanical valve is likely to be chosen for aortic valve replacement, and therefore long-term anticoagulation is necessary even in patients who undergo reparative mitral valve surgery. In the present series, all patients in both the DVR and MVP AVR groups received a mechanical valve in the aortic position and thus required long-term anticoagulation. This might be a reason for the lack of difference between MVP AVR and DVR in terms of late cardiac survival in this study. Mitral Valve Reoperation Our study demonstrated that the durability of MVP was significantly inferior to that of MVR in double valve surgery. Actuarial freedom from mitral valve reoperation at 12 years was 52.6% and 76.8% in MVP AVR and DVR, respectively (p 0.002). Rheumatic mitral disease is well known to be associated with inferior durability of MVP [2, 4, 5, 9 14]. After repair of a rheumatic mitral valve, actuarial freedom from mitral valve reoperation at 10 years has been reported to be 46% to 89% [2, 4, 5, 9 11]. Even with the inferior durability of MVP, many authors have recommended MVP for isolated rheumatic mitral disease when technically feasible because of better cardiac survival and fewer thromboembolic complications when compared with MVR [9, 11, 12]. Moreover, Yau and associates [9] concluded that mitral valve reoperation did not carry a demonstrably increased risk of death as they performed reoperation without mortality in 23 patients after initial MVP for isolated rheumatic mitral disease. In the present study, 21 of 47 patients in the MVP AVR group required mitral valve reoperation, and the operative mortality for these 21 patients was 4.7% (1 of 21). This mortality is acceptable considering the high risk of these 19 patients who had double valve disease, but it is a relatively high mortality. In addition, no significant advantages of MVP AVR over DVR with regard to cardiac survival and thromboembolic complications, both of which were reported to be advantages of MVP over MVR in the previous studies for isolated rheumatic mitral disease, were seen in our series. Several risk factors for late mitral valve failure after MVP in rheumatic disease, such as mixed rheumatic MS and MR, leaflet calcification, or severe subvalvular disease, have been identified [9, 15, 16]. Mitral valves with these lesions should probably be replaced instead of repaired in double valve surgery, and MVP should be limited to lesions in which excellent durability of the repair can be expected. Thromboembolism and Bleeding With regard to both thromboembolic and bleeding complications, we did not note any statistical advantage of MVP AVR over DVR. Similar results have been demonstrated in most previous studies [4 6] in double valve surgery. The lack of the need for warfarin anticoagulation is a major advantage of MVP, unless the patient has atrial fibrillation. A lower incidence of hemorrhage and thromboembolic complications after MVP, compared with MVR, for isolated mitral valve disease has been reported previously [9]. However, in the present study, all MVP AVR patients required lifelong warfarin anticoagulation because of the presence of a mechanical prosthesis in the aortic position, even when the mitral valve had been preserved by a repair technique. All DVR patients also required warfarin anticoagulation because of the insertion of at least one mechanical valve. Therefore, there was no difference between MVP AVR and DVR in terms of the need for lifelong anticoagulation. This could be a major reason for the lack of any significant advantage of MVP AVR over DVR with regard to embolic and bleeding complications. Study Limitations The present study was not randomized, and therefore a propensity score was incorporated into the multivariable analysis to reduce bias in comparing outcomes after MVP AVR and DVR. However, it is possible that significant bias may still exist by unmeasured variables. For example, in most patients we could not precisely quantify the severity of the rheumatic mitral valve lesion from either

Ann Thorac Surg KUWAKI ET AL 2007;83:558 63 MITRAL VALVE REPAIR VERSUS REPLACEMENT IN DOUBLE VALVE SURGERY 563 preoperative echocardiographic data or operative records (eg, severity of calcification in leaflets or annulus, degree of annular dilatation, number of chordae affected), which might have influenced treatment choice. Another limitation is the small number of patients in the study, which might be reasons why this study was unable to demonstrate advantages of survival and thromboembolic complications of MVP AVR over DVR in double valve surgery. Conclusions Similar results with regard to late morbidity and mortality were seen between MVP AVR and DVR patients, except for the need for mitral valve reoperation. Compared with DVR, MVP AVR was associated with higher incidence of mitral valve reoperation that was related to progression of the rheumatic mitral lesion. Potential advantages of MVP over MVR, such as improved cardiac survival and fewer thromboembolic events, were not identified. Therefore, in double valve surgery for rheumatic disease, MVP should be performed only in a highly selected group of patients in whom excellent durability of the repaired mitral valve can be expected. Patients with pliable mitral leaflets without subvalvular disease should be good candidates for MVP in double valve surgery. However, moderately or severely deteriorated mitral valves, such as those with leaflet calcification, thickened leaflets, or mitral valves associated with severe subvalvular disease, should probably be replaced. References 1. Bonow RO, Carabello B, de Leon AC Jr, et al. Guidelines for the management of patients with valvular heart disease. Executive summary: a report of the American College Of Cardiology/American Heart Association Task Force in Practice Guidelines (Committee on Management of Patients With Valvular Heart Disease). Circulation 1998;98:1949 84. 2. Gillinov AM, Blackstone EH, Cosgrove DM III, et al. Mitral valve repair with aortic valve replacement is superior to double valve replacement. J Thorac Cardiovasc Surg 2003;125:1372 87. 3. Szentpetery S, Rich JB, Azar H, Newton JR, Tenzer MM. Mitral valve repair combined with aortic valve replacement. J Heart Valve Dis 1997;6:32 6. 4. Hamamoto M, Bando K, Kobayashi J, et al. Durability and outcome of aortic valve replacement with mitral valve repair versus double valve replacement. Ann Thorac Surg 2003;75: 28 34. 5. Grossi EA, Galloway AC, Miller JS, et al. Valve repair versus replacement for mitral insufficiency: when is a mechanical valve still indicated? J Thorac Cardiovasc Surg 1998;115:389 96. 6. Mueller XM, Tevaearai HT, Stumpe F, et al. Long-term results of mitral-aortic valve operations. J Thorac Cardiovasc Surg 1998;115:1298 309. 7. Turina J, Stark T, Seifert B, Turina M. Predictors of the long-term outcome after combined aortic and mitral valve surgery. Circulation 1999;100(Supple 2):II-48 53. 8. Edmunds LH Jr, Clark RE, Cohn LH, Grunkemeier GL, Miller DC, Weisel RD. Guidelines for reporting morbidity and mortality after cardiac valvular operations. Ann Thorac Surg 1999;62:932 5. 9. Yau TM, Farag EL, Ghoneimi YA, Armstrong S, Ivanov J, David TE. Mitral valve repair and replacement for rheumatic disease. J Thorac Cardiovasc Surg 2000;119:53 61. 10. Deloche A, Jebara VA, Relland JYM, et al. Valve repair with Carpentier technique. The second decade. J Thorac Cardiovasc Surg 1990;99:990 1002. 11. Chauvaud S, Fuzellier JF, Berrebi A, Deloche A, Fabiani JN, Carpentier A. Long-term (29 years) results of reconstructive surgery in rheumatic mitral valve insufficiency. Circulation 2001;104(Suppl 1):I-12 5. 12. Mavioglu I, Dogan OV, Ozeren M, Dolgun A, Yucel E. Valve repair for rheumatic mitral disease. J Heart Valve Dis 2001; 10:596 602. 13. Fernandez J, Joyce DH, Hirschfeld K, et al. Factors affecting mitral valve reoperation in 317 survivors after mitral valve reconstruction. Ann Thorac Surg 1992;54:440 8. 14. Gillinov AM, Cosgrove DM, Lytle BM, et al. Reoperation for failure of mitral valve failure. J Thorac Cardiovasc Surg 1997;113:467 75. 15. Detter C, Fishlein T, Feldmeier C, Nollert G, Reichenspurner H, Reichart B. Mitral commissurotomy, a technique outdated? Long-term follow-up over a period of 35 years. Ann Thorac SUrg 1999;68:2112 8. 16. Choudhary SK, Dhareshwar J, Bovil A, Airan B, Kumar AS. Open mitral commissurotomy in the current era: indications, technique, and results. Ann Thorac Surg 2003;75:41 6.