Although heart valve replacement is a safe and commonly

Mitral Valve Relacement: Randomized Trial of St. Jude and Medtronic Hall Prostheses Andrew C. Fiore, MD, Hendrick B. Barner, MD, Marc T. Swartz, BA, Lawrence R. McBride, MD, Arthur J. Labovitz, MD, Kathy J. Vaca, RN, Jan St. Vrain, RDCS, Gary L. Grunkemeier, PhD, and George C. Kaiser, MD Division of Cardiothoracic Surgery and Cardiology, Saint Louis University Health Sciences Center, St. Louis, Missouri Background. This study was designed to better define the merits of the bileaflet and tilting-disc valves. Methods. We rosectively randomized 156 atients (mean age, 59 years) to receive either the St. Jude (n 80) or the Medtronic Hall (n 76) mitral valve rosthesis between Setember 1986 and December 1997. The two grous were not significantly different with resect to reoerative New York Heart Association class, left ventricular ejection fraction, incidence of mitral stenosis or insufficiency, extent of coronary artery disease, comleteness of revascularization, or cross-clam or byass time. Results. The oerative mortality (11.2% versus 13.1%, St. Jude versus Medtronic Hall, resectively) and late mortality (27% versus 22%, St. Jude versus Medtronic Hall, resectively) were not significantly different. Follow-u was comlete in all hosital survivors with a mean of 60.7 months (range, 1 to 133 months). The analysis of 10-year actuarial survival and freedom from valve-related events demonstrated no significant differences between the cohorts. Freedom from reoeration was higher in the St. Jude grou ( < 0.01). Comarisons of atient functional status and echocardiograhic hemodynamic arameters obtained at the time of follow-u demonstrated no significant differences between the two rostheses. Conclusions. This study suggests that there is no difference between the St. Jude and Medtronic Hall rostheses with resect to late clinical erformance or hemodynamic results and therefore does not suort the referential selection of either rosthesis. (Ann Thorac Surg 1998;66:707 13) 1998 by The Society of Thoracic Surgeons Although heart valve relacement is a safe and commonly erformed rocedure, all available rostheses are associated with valve-related comlications that influence their clinical use. Two commonly used rosthetic valves in the United States are the bileaflet St. Jude and the tilting-disc Medtronic Hall. Both rostheses ossess secific mechanical features that aear advantageous. The wide oening angle (70 degrees), uward gliding disc, and monostrut construction are attractive roerties of the Medtronic Hall valve, whereas the central flow design of the low-rofile bileaflet St. Jude rosthesis is urorted to be hemodynamically advantageous. To determine whether one rosthesis has clinical or hemodynamic sueriority, we designed a rosective, randomized trial to comare these two valves in atients requiring mitral valve relacement. Material and Methods From Setember 1986 until December 1997, 156 atients undergoing elective valve relacement for mitral valve dysfunction were rosectively randomized to receive Presented at the Thirty-fourth Annual Meeting of The Society of Thoracic Surgeons, New Orleans, LA, Jan 26 28, 1998. Address rerint requests to Dr Fiore, Deartment of Surgery, Saint Louis University Health Sciences Center, 3635 Vista at Grand Blvd, St. Louis, MO 63110-0250. either the Medtronic Hall (MH, 76 atients) or the St. Jude (SJ, 80 atients) mitral rosthesis. Patients requiring emergency oerations and those requiring simultaneous aortic and mitral valve relacement were excluded from this study. Randomization was accomlished by a blind drawing from randomly distributed enveloes. The clinical data recorded included age, sex, admitting diagnosis, cardiac risk factors, rocedure erformed, and ostoerative comlications. A New York Heart Association functional classification was determined for each atient. Angina was graded using the Canadian Cardiovascular Society Classification. Cardiac catheterization data noted include left ventricular end-diastolic ressure, ulmonary artery ressure, cardiac index, left ventricular score (regional wall motion scoring as defined by Coronary Artery Surgery Study) [1], number of major coronary vessels diseased (defined as at least 70% luminal obstruction in any angiograhic lane), resence of left main coronary artery stenosis (defined as 50% luminal obstruction), mitral valve area (calculated using the Gorlin formula), and the resence and severity (grade 0 to 4) of angiograhically determined mitral insufficiency. M-mode, two-dimensional, and Doler echocardiograhy were erformed reoeratively and 1 to 7 years ostoeratively (mean, 2.4 years). Ventricular endsystolic and end-diastolic dimensions were measured, 1998 by The Society of Thoracic Surgeons 0003-4975/98/$19.00 Published by Elsevier Science Inc PII S0003-4975(98)00670-5

708 FIORE ET AL Ann Thorac Surg PROSTHETIC MITRAL VALVE 1998;66:707 13 and ejection fraction and rosthetic valve gradient were calculated. In atients with atrial fibrillation, the mean of ten consecutive cardiac cycles was calculated. The ejection fraction was calculated by Simson s rule as reviously described [2]. Mitral valve areas were calculated by the ressure half-time method with either continuous or ulsed wave recordings [3, 4]. The resence and degree of rosthetic valve regurgitation was qualitatively assessed by ulsed wave, continuous wave, and color Doler echocardiograhy from multile windows. Exercise studies were erformed on a treadmill using the Bruce or modified Bruce rotocol [5]. The heart rate increased from a mean of 74 to a mean of 123 beats er minute (increase of 46%). The metabolic equivalents increased from 2 to 13 with a mean eak value of 6.76. Cardioulmonary byass was established using ascending aortic and bicaval cannulation with systemic hyothermia and hemodilution. Myocardial rotection included the use of cold hyerkalemic blood cardiolegia (6 C). The infusate consisted of a 4:1 dilution of blood with 0.9% normal saline solution that contained 20 meq/l sodium bicarbonate and 0.1 mg/ml lidocaine hydrochloride (Xylocaine). The initial infusion contained 14 meq/l of otassium. In atients in whom the aortic valve was cometent, an initial dose (15 ml/kg) was given through the aortic root and subsequent doses (7.5 ml/kg) were infused through the coronary sinus. In atients with aortic insufficiency, all doses of cardiolegic solution were given through the coronary sinus. Toical myocardial cooling with Ringer s lactate slush and a cardiac insulation ad for hrenic nerve rotection were used in all atients. Proximal vein graft anastomoses were erformed using a artially occlusive aortic clam while distal anastomoses were constructed with the aorta cross-clamed. Myocardial revascularization was erformed before mitral valve relacement to avoid raising the heart after the mitral rosthesis had been inserted, thus decreasing the risk of myocardial ruture. The oerative rocedure was erformed by the same grou of surgeons (A.C.F., H.B.B., L.R.M., G.C.K.) using similar techniques. The mitral rosthesis was inserted with interruted horizontal mattress sutures of 2-0 Ethibond (Ethicon, Somerville, NJ) laced through the left atrial side so as to evert the mitral annulus. When feasible, the osterior mitral leaflet was left intact. The St. Jude rosthesis was oriented either in the antianatomic axis erendicular to the mitral commissures or in the anatomic lane arallel to the mitral commissures. The Medtronic Hall valve was generally oriented with the larger orifice oening osteriorly. This rosthesis could be rotated according to the surgeon s reference to revent entrament of the disc by the retained osterior mitral leaflet. Recorded oerative variables included cardioulmonary byass and cross-clam times. Prohylactic anticoagulation with sodium warfarin was carried out in all atients. Anticoagulation was instituted within 48 hours after the oeration unless clinical contraindications existed. The control of theray was made by interval analysis of the international normalized ratio, which was ket between 2.5 and 3.5. Antibiotic rohylaxis consisted of cefazolin sodium (Ancef; Smith Kline & French Laboratories, Philadelhia, PA) administered reoeratively and for 48 hours ostoeratively. Follow-u data were collected by a registered nurse using written questionnaires and telehone interviews with the atient, the atient s hysician, or both. All atients enrolled were contacted yearly for the duration of their articiation in this rosective randomized trial. Patients surviving the oerative eriod were followed u for a minimum of 1 month and a maximum of 133 months (mean, 60.7 months). The mean follow-u was 4.9 years for SJ and 5.1 years for MH (not significant [NS]). The cumulative follow-u was 394 atient-years for the SJ and 395 atient-years for the MH grou. No atient was lost to follow-u. The definitions of events, their classification, and the analysis of the data follow the recently suggested guidelines for reorting morbidity and mortality after cardiac valve rosthesis insertion [6]. Oerative mortality is defined as death in the oerating room, in the hosital, or within 30 days of the oerative rocedure regardless of the atient s geograhic location. Late mortality is defined as death beyond 30 days of the oerative rocedure regardless of the atient s geograhic location. Oerative deaths were all classified as valve-related. Data were analyzed using the Statview for Windows 4.53 statistical software ackage (Abacus Concets, Inc, Berkeley, CA). Univariate analysis of discrete variables was erformed using the 2 analysis or Fisher s exact test where aroriate. Unaired Student s t tests were used for continuous variables. Results are exressed as means the standard deviation excet as noted. The actuarial curves were constructed using the methods described by Kalan and Meier and are reorted with 95% confidence limits. Actuarial curves were comared using a log-rank analysis as described by Mantel-Cox [7]. Hosital deaths were included in the actuarial survival analysis. A value of less than 0.05 was considered significant. Results Clinical Characteristics The cohorts were similar with resect to age, sex, hemodynamic and oerative arameters, and the degree of mitral stenosis or mitral insufficiency (Table 1). The majority of atients had mild left ventricular dysfunction with moderate ulmonary hyertension. Fifteen ercent of atients in both grous had rior valve relacement, and 23% had concomitant myocardial revascularization. Thirty-three ercent of the atients in both cohorts had at least one rior cardiac surgical rocedure. Before oeration, most atients were in New York Heart Association functional class III or IV. After oeration, however, the mean New York Heart Association functional class imroved significantly in both cohorts (SJ, 1.4 0.63; MH, 1.36 0.75; NS). Oerative Mortality A total of 19 atients (12.1%) died in the hosital or within 30 days of the oerative rocedure. The overall

Ann Thorac Surg FIORE ET AL 1998;66:707 13 PROSTHETIC MITRAL VALVE 709 oerative mortality was 11.2% for the SJ grou and 13.1% for the MH grou (NS), whereas that for mitral valve relacement with myocardial revascularization was 12.5% and 31%, resectively (NS). The oerative mortality of those atients undergoing redo valve relacement was nearly twofold higher in the SJ cohort (SJ, 15.4%; MH, 8%; NS). The most common cause of oerative death was low cardiac outut syndrome. Eleven atients died as a result of low cardiac outut associated with biventricular failure. In two instances (SJ, 1; MH, 1) oerative death was caused by atrioventricular sulcus ruture. The cause of death in 3 additional atients receiving the Medtronic Hall valve included fatal dysrhythmia, massive cerebral hemorrhage on ostoerative day 1, and an aortic dissection at the aortic cannulation site in a atient with Ehlers-Danlos syndrome. Three additional atients (SJ, 2; MH, 1) died of multiorgan system failure. One atient with a history of chronic heatitis died 13 days ostoeratively of liver and renal failure. A second atient remained comatose ostoeratively. Renal and resiratory failure develoed and the atient died on the fourth ostoerative day. The remaining atient had undergone urgent third-time mitral valve relacement. He died of multiorgan system failure after several hyotensive eisodes on ostoerative day 5. The atients who died in the hosital had a mean New York Heart Association functional class of 3.1 0.6, and 47% of hosital deaths were in atients with mitral Table 1. Preoerative and Oerative Clinical Profile a Variable SJ MH Age (y) 60.8 11 58.1 10.6 0.11 Male sex (%) 41 34 0.37 NYHA class 2.9 0.8 2.9 0.6 0.69 LVEDP (mm Hg) 14.4 6.5 13.5 6.4 0.46 LVEF 0.55 0.13 0.57 0.12 0.60 Pulmonary artery 48.8 18.2 51.2 18 0.46 systolic ressure (mm Hg) Cardiac index 2.2 0.50 2.34 0.56 0.17 (L min 1 m 2 ) Mitral area (cm 2 ) 1.03 0.45 0.97 0.53 0.59 Mitral insufficiency 2.8 1.6 2.6 1.7 0.42 Coronary artery 30 21 0.20 disease (%) Prior valve 15 13 0.74 relacement (%) Duration of CPB (min) 146 52.6 148 61.3 0.81 Cross-clam time (min) 93 36 97 33.6 0.51 Simultaneous CABG (%) 21 17 0.52 Prosthesis size (mm) 28.8 4.2 28.7 2.5 0.82 a s exressed as ercent or mean the standard deviation. CABG coronary artery byass grafting; CPB cardioulmonary byass; LVEDP left ventricular end-diastolic ressure; LVEF left ventricular ejection fraction; MH Medtronic Hall; NYHA New York Heart Association; SJ St. Jude. Fig 1. Actuarial survival rates after mitral valve relacement with the St. Jude (SJ) and Medtronic Hall (MH) valves. Oerative and late deaths are included. (NS not significant.) regurgitation as the rimary athologic rocess. Univariate analysis demonstrated that the resence of left ventricular ejection fraction less than 0.40 correlated with oerative mortality ( 0.05). Coronary artery disease, a history of hyertension, and cardioulmonary byass time were marginally significant redictors of oerative mortality ( 0.05 to 0.1). Sex, ulmonary artery systolic ressure, and global ischemic time were not correlated with hosital death. Multivariate analysis failed to demonstrate any redictors of oerative mortality. Late Mortality There were a total of 39 late deaths recorded after mitral valve relacement: 22 in atients receiving the St. Jude valve and 17 in atients receiving the Medtronic Hall valve (SJ, 27%; MH, 22%; NS). Valve-related comlications were the cause of late death in 12 atients (SJ 5, 12.8%; MH 7, 17.9%; NS). The most common cause of valve-related death was cerebrovascular accident, which accounted for five valve-related late deaths (SJ, 4; MH, 1) at 31, 47, 54, 72, and 91 months ostoeratively. Sudden death occurred at home in 3 atients (all in the MH grou) at 23, 24, and 34 months ostoeratively. Two of these atients had known coronary artery disease, an ejection fraction less than 40%, and ulmonary hyertension. Three atients (SJ, 2; MH, 1) died of congestive heart failure at 3.3, 53, and 55 months ostoeratively secondary to aravalvular leak. The remaining valve-related late death (MH) was secondary to generalized sesis from bacterial endocarditis 2.3 months after valve imlantation. Patient Survival The actuarial survival curve for the oulation is shown in Figure 1. At 5 and 10 years the survival rates were nearly identical for atients receiving either the Medtronic Hall or the St. Jude mitral valve. Patients who underwent mitral valve relacement were found to have a significantly higher 10-year actuarial survival in the absence of coexisting coronary artery disease (63% versus 30%; 0.05).

710 FIORE ET AL Ann Thorac Surg PROSTHETIC MITRAL VALVE 1998;66:707 13 Table 2. Actuarial 10-Year Freedom From Events a Event SJ MH Survival 53 7 58 6 0.60 Thromboembolism 72 8.0 72 7.0 0.56 Hemorrhage 83 6.0 89 5 0.24 Endocarditis 100 0 94 3 0.07 Paravalvular leak 89 5 77 10 0.79 Reoeration 100 0 79 10 0.01 All valve-related events 43 8 35 9 0.45 a s exressed as ercent standard error of mean. MH Medtronic Hall; NS not significant; SJ St. Jude. Fig 3. Actuarial freedom from thromboembolism after mitral valve relacement with the St. Jude (SJ) and Medtronic Hall (MH) valves. (NS not significant.) Valve-Related Comlications During the eriod of oerative hositalization, there were no valve-related events in any atient other than death. The actuarial freedom from these morbid events is shown in Table 2 and Figure 2. STRUCTURAL DETERIORATION. Structural deterioration did not occur in any atient in this study. THROMBOEMBOLISM. There were a total of 26 eisodes of thromboembolism occurring in 26 atients and 5 were fatal. Valve thrombosis occurred in 4 atients, 2 in each grou. One atient died of metastatic disease with clot on the St. Jude valve at 1.8 months ostoeratively. Two atients (SJ, 1; MH, 1) underwent successful reoeration with insertion of a orcine biorosthesis at 18 and 43 months ostoeratively. The remaining atient with valve thrombosis had successful thrombolytic theray 72 months after imlant. These latter 3 atients are longterm survivors. Nine atients who exerienced transient ischemic attack (SJ, 4; MH, 5) had resumed embolic events at a mean of 42 months ostoeratively in both cohorts. The earliest transient ischemic attack occurred at 0.5 months ostoeratively. Seventeen atients exerienced a cerebrovascular accident (SJ, 8; MH, 9) of which 4 (SJ, 3; MH, 1) were fatal. The actuarial freedom from thromboembolism was equal in both cohorts (Fig 3). HEMORRHAGE. There were 14 instances of hemorrhage related to anticoagulant treatment (SJ, 9; MH, 5). Bleeding occurred intracerebrally in 2 atients and was fatal in 1 at 72 weeks ostoeratively. Surgical bleeding develoed in 3 atients (SJ, 2; MH, 1) after elective rocedures whereas 7 atients (SJ, 4; MH, 3) had nonfatal gastrointestinal hemorrhage that did not require oeration. The 2 remaining atients (MH, 2) exerienced nonfatal hemorrhage. One atient had a retroeritoneal hematoma after a motor vehicle accident and the final atient had disseminated intravascular coagulation associated with a ulmonary embolism. Both atients were treated nonoeratively and are long-term survivors. The actuarial freedom from hemorrhage at 10 years was slightly higher in the Medtronic Hall grou, but the difference was not significant (Fig 4). ENDOCARDITIS. Bacterial endocarditis with a aravalvular leak develoed in 4 atients, all in the Medtronic Hall grou. One atient underwent reoeration at 1.5 months ostoeratively with insertion of a orcine valve and survived 34 months, dying of comlications secondary to chronic dialysis. Two atients who did not undergo reoeration died at 2 and 34 months ostoeratively of multiorgan system failure. The remaining atient had stretococcal endocarditis, which was treated medically, and he continues to be a long-term survivor. The 10-year actuarial freedom from endocarditis was not significantly different between cohorts. PARAVALVULAR LEAK. Paravalvular leak develoed in eleven atients (SJ, 5; MH, 6). Three atients underwent reo- Fig 2. Actuarial freedom from all valve-related events after mitral valve relacement with the St. Jude (SJ) and Medtronic Hall (MH) valves. (NS not significant.) Fig 4. Actuarial freedom from hemorrhage after mitral valve relacement with the St. Jude (SJ) and Medtronic Hall (MH) valves. (NS not significant.)

Ann Thorac Surg FIORE ET AL 1998;66:707 13 PROSTHETIC MITRAL VALVE 711 Table 4. Postoerative Exercise Transvalvular Echocardiograhic Mean Gradient Valve Size (mm) Gradient (mm Hg) SJ (n 22) MH (n 26) Fig 5. Actuarial freedom from reoeration after mitral valve relacement with the St. Jude (SJ) and Medtronic Hall (MH) valve. eration with one oerative death and 2 long-term survivors. Of the remaining 8 atients, 2 died at 3 and 53 months of congestive heart failure and sudden unexlained death, resectively. The remaining 6 atients are long-term survivors. The 10-year actuarial freedom from aravalvular leak was similar in both cohorts. REOPERATIONS. Seven atients, all in the Medtronic Hall grou, underwent reoeration. The indications for reoeration were aravalvular leak (3), endocarditis (2), valve thrombosis (1), and endocarditis with eriheral emboli in the remaining atient. The oerative mortality for reoeration was 28%. The 10-year actuarial freedom from reoeration was significantly higher in the SJ cohort (Fig 5). POSTOPERATIVE ECHOCARDIOGRAPHY. The ostoerative resting and exercise echocardiograhic analysis are summarized on Tables 3 and 4. The mean transvalvular gradient at rest and with exercise was not significantly different between grous at any valve size. Comment The bileaflet St. Jude and the ivoting-disc Medtronic Hall valve have retained widesread oularity for more than 20 years. These devices have undergone virtually no alteration in design since their develoment in 1977 and 1978 and have remained unmodified throughout this study, which commenced in 1986. Although a number of investigators have reorted their clinical and hemodynamic results with these valves, this reort reresents the only rosective randomized trial comaring these rostheses in the mitral osition Table 3. Postoerative Resting Transvalvular Echocardiograhic Mean Gradient Valve Size (mm) Gradient (mm Hg) SJ (n 42) MH (n 41) 25 3.04 0.84 4.02 2.2 0.48 27 3.34 1.4 4.26 1.26 0.61 29 3.82 0.48 3.10 1.7 0.39 31 1.54 0.67 2.92 1.1 0.06 33 2.48 1.48 2.68 1.1 0.64 MH Medtronic Hall; n number of atients; NS not significant; SJ St. Jude. 25 6.7 5.2 8.2 3.3 0.52 27 6.2 3.2 7.2 2.5 0.59 29 3.9 3.0 6.0 2.7 0.49 31 4.2 2.9 6.7 4.0 0.34 33 4.6 2.7 6.5 2.3 0.26 MH Medtronic Hall; n number of atients; NS not significant; SJ St. Jude. throughout a 10-year eriod of observation at a single institution [8, 9]. This is imortant because referral atterns may vary from one institution to another as may oerative techniques and ostoerative management, factors that often make comarisons roblematic. The overall oerative mortality in this series is similar to that reorted by other investigators. Oerative mortality rates of 4.7% to 12% have been reorted for mitral valve relacement with St. Jude and 7.5% to 13.1% with the Medtronic Hall valve [10, 11]. In our series nearly 50% of the oerative deaths were in atients with mitral insufficiency, and the most accurate redictor of oerative death was left ventricular dysfunction. It was not unexected that the leading cause of hosital mortality was erioerative heart failure. Although indeendent determinants of oerative survival for mitral valve relacement from revious studies have included coexisting coronary artery disease and female gender, they were not demonstrated to be redictors in this series. Patient survival based on 5- and 10-year actuarial estimates was not significantly different between grous. At 5 years our survival estimates (SJ, 65%; MH, 70.5%) comared favorably with those reorted in other series, which range from 59% to 78% for the St. Jude and 67% to 74% for the Medtronic Hall [12, 13]. Late survival is deendent on many factors related to the individual atient s disease and to the comlications of the rosthesis. In our study, most late deaths were a result of cardiac failure and were not related to rosthetic valve dysfunction or its attendant comlications. A more sensitive measure of valve erformance is thromboembolism, which is a major risk factor for atients with rosthetic valves. Thromboembolism was the most common valve-related comlication in this series. The incidence of thromboembolic comlications in the third-generation valvular devices has decreased dramatically. This has been because of imroved hemodynamic design resulting in better flow velocity rofile, less turbulence, and reduction of stagnant areas. In the Medtronic Hall valve, a more central osition of the disc in the oen osition enlarged the minor orifice, thus reducing stagnation of blood and turbulence in this area. The central flow design of the bileaflet St. Jude rosthesis may contribute to its low incidence of thromboembolism. In this reort, the 5-year actuarial freedom from thromboembolism was slightly better for the Medtronic Hall

712 FIORE ET AL Ann Thorac Surg PROSTHETIC MITRAL VALVE 1998;66:707 13 grou (SJ, 87%; MH, 94%) but by 10 years the actuarial freedom was identical in both cohorts. These figures comare favorably with those reorted by other investigators whose five-year actuarial estimates have ranged between 85% and 92% for the St. Jude valve and 85% and 94% with the Medtronic Hall valve [14, 15]. There were four cases of valve thrombosis. Noncomliance with sodium warfarin anticoagulation in the resence of atrial fibrillation was believed to have contributed to thrombus formation in 3 atients. The fourth case was a atient with metastatic lung cancer in whom valve thrombosis develoed desite an international normalized ratio of 3.0. He was not thought to be a candidate for thrombolytic theray or reoeration. All atients with thromboembolic comlications had diyridamole (300 mg daily) or asirin (5 grains) added to their anti-coagulant regimen to reduce the recurrent risk of thromboembolism. Hemorrhage was the second most common valverelated comlication. This comlication was fatal in 1 hyertensive atient in whom an intracerebral hemorrhage develoed. The actuarial freedom from hemorrhage was similarly low in both cohorts and comared favorably with that reorted in other series for the Medtronic Hall and St. Jude valves. As in most revious reorts, we noted a low rate of endocarditis with little difference between the two valves [16]. In this reort, we measured the mean transvalvular gradient at rest and after treadmill stress testing between 12 and 62 months ostoeratively using Doler echocardiograhy. No significant difference in the resting or stress-induced mean gradient could be demonstrated when both rostheses were comared in each valve size. Imroved hemodynamic erformance has been associated with symtomatic imrovement. In this series, the St. Jude and Medtronic Hall valves achieved equal imrovement in atient symtoms as evidenced by the low ostoerative NYHA functional score observed in both cohorts. The results of this study indicate that in the first 10 years after imlantation, the erformance of the St. Jude and Medtronic Hall rostheses are comarable. These two valves are among the best of the modern generation of rostheses, and either may be confidently recommended for valve relacement. The choice of valve should be based on the surgeon s exerience and reference, which may be linked to other factors such as ease of insertion, availability, and cost. Clearly, this study suggests that neither valve can claim suerior erformance or safety. We thank Terri Wriley for rearation of the manuscrit and gratefully acknowledge the articiation of Karen Powers, RN, and Debbie Moroney, RN, in obtaining accurate follow-u data. References 1. Princial investigators of CASS and their associates. The National Heart, Lung and Blood Institute Coronary Artery Surgery Study. Circulation 1981;63(Sul 1):1 81. 2. Mehdirad AA, Williams GA, Bryg RJ, et al. Evaluation of left ventricular function with uright exercise: correlation of exercise Doler with ost-exercise wall motion and single lane echocardiograhic ejection fraction. Circulation 1987; 75:413 9. 3. Hatle L, Angelsen B, Tromsdal A. Non-invasive assessment of atrioventricular ressure half-time by Doler ultrasound. Circulation 1969;60:1096 104. 4. Bryg RB, Williams GA, Labovitz AJ, Aker U, Kennedy HL. Effect of atrial fibrillation and mitral regurgitation on calculated mitral valve area: comarison of Doler, twodimensional echocardiograhy and cardiac catheterization. Am J Cardiol 1986;57:634 8. 5. Edmunds LH Jr, Clark RE, Cohn LH, Grunkemeier GL, Miller DC, Weisel RD. Guidelines for reorting morbidity and mortality after cardiac valvular oerations. Ann Thorac Surg 1996;62:932 5. 6. Mantel N, Haenszel W. Statistical asects of the analysis of data from retrosective studies of disease. J Natl Cancer Inst 1959;22:719 48. 7. Anthunes MJ. Clinical erformance of St. Jude and Medtronic Hall rostheses: a randomized comarative study. Ann Thorac Surg 1990;50:743 7. 8. Fiore AC, Naunheim KS, D Orazio S, et al. Mitral valve relacement: randomized trial of St. Jude and Medtronic Hall rostheses. Ann Thorac Surg 1992;54:68 73. 9. Keenan RJ, Armitage JM, Trento A, et al. Clinical exerience with the Medtronic Hall valve rosthesis. Ann Thorac Surg 1990;50:748 53. 10. Arom KV, Nicoloff DM, Kersten TE, et al. Ten years exerience with the St. Jude Medical valve rosthesis. Ann Thorac Surg 1989;47:831 7. 11. Vallejo JL, Gonzales-Santos JM, Albertos J, et al. Eight years exerience with the Medtronic Hall valve rosthesis. Ann Thorac Surg 1990;50:429 36. 12. Ibrahim M, Cleland J, O Kane H, Gladstone D, Mulholland C, Craig B. The St. Jude Medical rosthesis: a thirteen-year exerience. J Thorac Cardiovasc Surg 1994;108:221 30. 13. Anthunes MJ, Wessels A, Sadowski RG, et al. Medtronic Hall relacement in a third-world oulation grou. J Thorac Cardiovasc Surg 1988;95:980 93. 14. Baudet EM, Oca CC, Roques XF, et al. A 5 1 2 year exerience with the St. Jude Medical cardiac valve rosthesis. J Thorac Cardiovasc Surg 1985;90:137 44. 15. Nitter-Hauge S, Abdelnoor M. Ten-year exerience with the Medtronic Hall valvular rosthesis. Circulation 1989;80 (Sul 1):43 8. DISCUSSION DR ALEJANDRO ARIS (Barcelona, Sain): I congratulate Dr Fiore for roving to us that bileaflet valves are basically no better than disc valves. I think that bileaflet valves were at the right lace at the right time, but by no means have solved the roblems of other mechanical valves. Two years ago at this meeting in Orlando we resented our exerience with the Monostrut valve in a large study with more than 8,000 valves imlanted in Sain. This valve, incidentally, has just been aroved by the Food and Drug Administration for its use in the United States. Our results at 10 years were very similar to those resented here today. There are several misconcetions regarding bileaflet valves. First, it rovides better hemodynamic erformance. This is wrong. We have shown that in the small aortic annulus the