P found suitable valve substitutes for clinical use.

Transcription

1 Porcine Versus Pericardial Bioprostheses: A Comparison of Late Results in 1,593 Patients L. Conrad Pelletier, MD, Michel Carrier, MD, Yves Leclerc, MD, Gilles Lepage, MD, Pierre deguise, MD, and Ihor Dyrda, MD Departments of Surgery and Medicine, Institut de Cardiologie de Montreal, Montreal, Quebec, Canada From 1976 to 1988, 1,593 patients underwent valve replacement with a porcine (878 patients) or a pericardial bioprosthesis (715 patients). There were 701 aortic, 678 mitral, and 214 multiple-valve replacements. Follow-up was obtained for 1,559 patients (98%). Early mortality was 9% (79 patients) in the porcine valve group and 5% (37 patients) among patients with a pericardial valve (p < 0.01). Late survival after replacement with porcine valves was 80% k 1% and 62% k 3% at 5 and 10 years, respectively. With pericardial valves, 5-year survival was 79% k 2%. Among valve-related complications, rates of freedom from thromboembolism, endocarditis, and hemorrhage after 6 years were similar for both valve groups. Freedom from reoperation at 6 years was also similar after aortic (96% versus 91%) or multiple-valve replacement (95% versus 88%). However, for mitral valve replacement, freedom from reoperation was significantly better with porcine valves than with pericardial valves at 6 years (92% versus 68%; p < 0.001). This difference was mainly due to the Ionescu-Shiley valve, which accounted for 83% of primary tissue failures among pericardial bioprostheses implanted in the mitral position (10/12 patients). After 6 years, freedom from primary tissue failure of mitral valves was 92% k 2% with porcine and 70% k 11% with pericardial bioprostheses (p < ). The degree of clinical improvement among survivors was similar with both valve types. Thus, in the aortic position, pericardial valves compare with porcine valves up to 6 years, whereas in the mitral position, the durability of the former is significantly less, mainly because of the suboptimal performance of the Ionescu-Shiley pericardial bioprosthesis. ( 1989;47:352-61) orcine and pericardial bioprostheses have both been P found suitable valve substitutes for clinical use. However, comparisons between different series are often difficult because group disparities can preclude meaningful conclusions. Porcine bioprostheses have been in continual use since 1976, and pericardial valves of various types have been implanted starting in 1979 at the Montreal Heart Institute. It was therefore of interest to compare two concurrent groups of patients operated on by the same group of surgeons during an overlapping period and followed prospectively with the same clinical criteria. Porcine bioprostheses have been used more extensively and over a longer period than pericardial bioprostheses. Consequently, they can serve as the gold standard for comparative analysis of other bioprostheses, particularly in regard to valve durability, because the main concern with their use remains primary tissue failure, the critical period for which is somewhere between the sixth and 12th year [l-31. The aim of the present study was to compare pericardial and porcine bioprostheses in terms of patient survival, valve survival, and valve-related complications. Presented at the Twenty-fourth Annual Meeting of The Society of Thoracic Surgeons, New Orleans, LA, Sep 2628, Address reprint requests to Dr Pelletier, Department of Surgery, Institut de Cardiologie de Montrbal, 5000 Belanger St E, Montreal, Que, Canada HIT 1C8. Material and Methods From 1976 to 1988, 1,593 patients underwent heart valve replacement with a bioprosthesis. One thousand twentyfour Carpentier-Edwards porcine bioprostheses were used in 878 patients (634 standard, 239 improvedannulus, and 151 supraannular models), and 789 pericardial bioprostheses were implanted in 715 patients (316 Ionescu-Shiley, 295 Carpentier-Edwards, and 178 Mitroflow valves). There were 808 male and 785 female patients aged 13 to 80 years. The male to female ratio was 0.93 in the porcine group compared with 1.16 in the pericardial group (p < 0.05). The latter group included a greater proportion of patients older than 60 years (Fig l), and the mean age ( years [mean f standard deviation]) was significantly higher than that of patients in the porcine group (54 f 12 years) (p < 0.01). Preoperatively, 89 patients (5.6%) were in New York Heart Association functional class I, 551 (34.6%) were in class 11, 829 (52%) were in class 111, and 124 (7.8%) were in class IV. In the porcine valve group, 61.8% (543 patients) were in functional class 111 or IV compared with 57.3% (410 patients) of the pericardial valve patients, a difference that is not significant ( p > 0.05). Aortic stenosis was more frequent in the pericardial valve group (p < O.OOOl), whereas aortic regurgitation (p < O.OOOl), mitral stenosis, and mitral regurgitation were seen more often in the porcine valve patients (Table 1). A cardiac operation had been done previously in 359 patients (22.5%), 189 in the porcine valve group (21.5%) and 170 in the pericardial by The Society of Thoracic Surgeons /89/$3.50

2 1989; PELLETIER ET AL 353 NUMBER OF PATIENTS Rnn HPORCINE V PERICARDIAL V k AGE GROUPS Fig 1. Patient distribution by age groups indicates a preponderance of porcine valves (V) in patients less than 60 years old and a majority of pericardial valves in patients older than 70 years. Table 2. Associated Surgical Procedures Porcine Pericardial Valve Valve Group Group -~ Procedure No. % No. % Coronary artery bypass grafting Valve repair Enlargement aortoplasty Bentall procedure Resection of left ventricular aneurysm Closure of ventricular septa1 defect Postoperative intraaortic balloon pumping valve group (23.8%)(p > 0.05). However, failure of a valve prosthesis was the indication for operation more often in the pericardial valve group (67 patients, 9.4%) than in the porcine valve group (24 patients, 2.7%) (p < ). A Carpentier-Edwards porcine bioprosthesis was used for isolated aortic valve replacement (AVR) in 331 patients (37.7%), for isolated mitral valve replacement (MVR) in 419 patients (47.7%), and for multiple-valve replacement (MR) in 128 patients (14.6%). A pericardial bioprosthesis was implanted in 370 patients (51.7%) for AVR, in 259 (36.2%) for MVR, and in 86 (12%) for MR. The difference between the two groups was significant (p < ). Coronary artery bypass grafting was associated with the valve procedure in 135 (15.4%) of the patients with a porcine valve and 155 (21.7%) of those with a pericardial bioprosthesis (p < 0.01). Associated surgical procedures are listed in Table 2. Patients were followed on a yearly basis either at the outpatient clinic or through their own physician. Overall, 34 patients were lost to follow-up, 19 (2.16%) in the porcine valve group and 15 (2.10%) in the pericardial valve group, for a global follow-up of 97.9%. The duration of follow-up of patients with a porcine bioprosthesis averaged 74 months, totaling 4,899 patient-years, compared with 34 months and 1,936 patient-years for the pericardial valve group. At the last follow-up visit, 28.4% (227 patients) of patients with porcine valves and 26% (176 patients) of those with pericardial valves were on a regimen of anticoagulation therapy, and antiplatelet agents were administered to 11.5% (92 patients) and 19.9% (135 patients) of the patients, respectively. Definitions An embolic episode was defined as any new neurological deficit, either transient or permanent, or any suggestive symptom, as well as all events of clinically detectable noncerebrovascular embolism, excluding perioperative cerebrovascular accidents (first week). Endocarditis was defined as any systemic infection with positive hemoculture or signs of infection of the prosthesis at reoperation or autopsy, including the early postoperative period. Table 1. Preoperative Dia,Qnosis Porcine Valve Group (n = 878) Pericardial Valve Group (n = 715) Diagnosis No. % No. % p Value Aortic valve < Stenosis Regurgitation Mixed Mitral valve NS Stenosis Regurgitation Mixed Prosthetic valve dysfunction < a Porcine valve group versus pericardial valve group. NS = not significant.

3 354 PELLETIER ET AL 1989; Periprosthetic leak was defined as any major valvular regurgitation demonstrated at heart catheterization (>grade 2/4), reoperation, or autopsy. Hemorrhage was considered a complication whenever treatment or hospitalization of a patient on a regimen of anticoagulant therapy was required. Primary tissue failure of a bioprosthesis included dysfunction with regurgitation due to leaflet tear, stretching, or creeping of the stent, or valvular obstruction by leaflet fibrosis or calcification. Valve-related complications were defined as any complications caused by the presence of a heart valve prosthesis. All late deaths from any cardiac-related causes were considered valve-related deaths. Valve survival was computed from freedom from valverelated death (cardiac-related death) and from reoperation for replacement or repair of a prosthesis. Data Analysis Student's t test and the 2 test were used for statistical analysis of the data. Results are expressed either as linearized incidence or using the life-table method with standard error of the mean. The accepted level of significance was 95%. Results Survival There were 79 early deaths among the porcine valve group, a 30-day mortality of 9.0%, and 37 deaths in the pericardial valve group, a 30-day mortality of 5.2% (p < 0.01). The respective early mortality rates were 6.6% (22 patients) and 3.8% (14 patients) for AVR, 8.8% (37 patients) and 5.4% (14 patients) for MVR, and 15.6% (20 patients) and 10.5% (9 patients) for MR. None of these differences were significant (p > 0.05). Among patients with a porcine valve, there were 141 late deaths compared with 68 in the pericardial valve group, giving linearized late mortality rates of 2.9%/ patient-year and 3.5%/patient-year, respectively. The overall survival curves of both groups were similar, as were those for AVR, MVR, and MR separately (Figure 2). Global actuarial survival of patients with porcine valves was 80% f 1% at 5 years and 62% f 3% at 10 years, whereas that of patients with pericardial valves was 79% f 2% after 5 years. Causes of death are shown in Table 3. The most frequent cause in both groups was congestive heart failure (32% and 30%, respectively). A noncardiac cause was responsible for 11% of deaths in each of the two groups. However, autopsy was done in slightly less than half of the patients who died in each group. Complications THROMBOEMBOLISM. With the porcine valve, there were 53 thromboembolic episodes (1.08%/patient-year): seven fatal, 14 with residual deficits, and 32 minor accidents. In the group of patients with pericardial bioprostheses, 39 sustained this complication (2.0l%/patient-year); 4 died, 11 had permanent deficits, and 24 were without permanent deficits. Although the linearized incidence was higher with pericardial valves in AVR, MVR, and MR x SURVIVAL 70 70?,,"A PERICARDIAL VALVES AORTIC VALVE MITRAL VALVE 60 MULTIPLE REPL Fig 2. Actuarial survival curves for porcine valve and pericardial valve patients indicate that the dijferences are not significant (N.S.). (REPL = replacements.) (Table 4), the actuarial curves of freedom from thromboembolism were not significantly different (Fig 3). Freedom rates at 10 years for porcine and at 6 years for pericardial valves were 96% * 1% and 94% f 2% after AVR, 89% * 2% and 89% 2 3% after MVR, and 86% * 5% and 88% f 4% after MR, respectively. There were three cases of valve thrombosis, two in the porcine valve group (one aortic and one mitral) and one in the pericardial valve group (aortic). The patient with the porcine mitral valve Table 3. Causes of Death" Porcine Valve Pericardial Valve Group Group Cause (n = 220) (n = 105) Congestive heart failure Sudden death 33 8 Myocardial infarction Arrhythmias 21 9 Reoperation 19 9 Perioperative hemorrhage 11 4 Cerebrovascular accident 6 5 Endocardi tis 3 6 Hemorrhage due to 4 3 anticoagulants Valve failure 2 2 Valve thrombosis 1 0 Unknown 2 4 Noncardiac a Autopsy was performed on 101 patients (45.9%) in the porcine valve group and 52 (49.5%) in the pericardial valve group.

4 1989; PELLETIER ET AL 355 PORCINE VERSUS PERlCARDlAL VALVES Table 4. Valve-Related Complications Porcine Valve Group Pericardial Valve Group Complication AVR MVR MR Total AVR MVR MR Total Thromboembolism Minor Major Fatal Total %/patient-year Endocarditis Fatal Total %/patient-year Hemorrhage Fatal Total %/patient-year Reopera tion Fatal Total %/patient-year Primary tissue failure Total %/patient-year All complications Total %/pa tien t- year AVR = aortic valve replacement; MR = multiple-valve replacement; MVR = mitral valve replacement. died, whereas the 2 with thrombosed aortic valves underwent reoperation successfully. ENDOCARDITIS. Twenty-three episodes of endocarditis occurred in patients with porcine valves (0.47%/patient-. -,,,,, * PERICARDIAL VALVES 46 EVENT FREE AORTIC VALVE 96i1% Fig 3. Freedom from thromboembolic complications. (N.S. = not significant; REPL = replacements.) I year) compared with 27 among those with pericardial bioprostheses (1.39%/patient-year). The linearized rates were higher with the latter valve in all positions (see Table 4). However, freedom-rate curves were not significantly different for AVR (92%? 4% freedom at 6 years for pericardial valves and 95% * 2% freedom at 10 years for porcine valves) and MVR (97% 5 1% freedom for pericardial valves at 6 years and 97% * 1% for porcine valves at 10 years). In the MR group, endocarditis was significantly more frequent with pericardial than with porcine valves; respective freedom rates were 88% 2 5% at 6 years and 96% * 3% at 10 years ( p < 0.05) (Fig 4). HEMORRHAGE. Patients with porcine valves sustained 17 hemorrhagic accidents (0.35%/patient-year), 4 of which were fatal, compared with 8 in the pericardial valve group (0.41%/patient-year), 2 of which were fatal accidents. Only three (12%) of all 25 episodes occurred after isolated AVR. HEMOLYSIS. No clinically significant hemolysis was detected with any of these bioprostheses in the absence of major valve dysfunction. REOPERATION. Reoperation was performed in 105 patients with a porcine bioprosthesis (2.14%/patient-year) and in

5 356 PELLETIER ET AL 1989;47: x EVENT FREE *,* PERICARDIAL VALVES AORTIC VALVE, Table 5. lndications for Reoperation Indication Porcine Valve Group Pericardial Valve Group No. % No. % Leaflet tear or stretching Prosthetic valve obstruction Paravalvular leak Active endocarditis Valve thrombosis Total Fig 4. Freedom from endocarditis. (N.S. = not significant; REPL = replacements.) 40 patients with a pericardial valve (2.07%/patient-year). With either type of prosthesis, the linearized rate of reoperation was higher after MVR or MR than after isolated AVR by a ratio of almost 2 to 1 (see Table 4). Freedom from reoperation for any cause was 96% * 1% and 83% * 5% at 6 and 10 years, respectively, with porcine aortic valves and 91% * 4% at 6 years with pericardial valves ( p > 0.05) (Fig 5). In the mitral position, freedom rates for porcine valves were 92% f 2% at 6 years and 53% * 6% at 10 years compared with 68% * 11% at 6 years with pericardial valves (p < 0.001). After MR, freedom from reoperation was 95% * 2% and 41% * 9% at 6 and 10 years, respectively, with porcine valves and "" I, A PERICARDIAL VALVES MITRAL VALVE Fig 5. Freedom from reoperation due to all causes. (N.S. = nificunt; REPL = replacements.) not sig- 88% f 4% at 6 years with pericardial valves (p > 0.05). The mortality rate of this complication was 18.1% (19 patients) with porcine valves and 22.5% (9 patients) with pericardial valves. Indications for reoperation are shown in Table 5. Primary tissue failure was responsible for 88% (92 patients) of reoperations in the porcine group and for 48% (19 patients) of reoperation among patients with a pericardial valve ( p < ). A much higher rate of reoperation was due to active prosthetic valve endocarditis in the latter group (12 patients, 30%) than in the former (4 patients, 3.8%) ( p < O.OOOl), reflecting the higher incidence of endocarditis in the pericardial valve group (1.39%/patientyear versus 0.47%/patient-year). VALVE DYSFUNCTION. In addition to those who underwent reoperation, 6 patients with a porcine valve (0.12%/patient-year) were found to have major valve dysfunction, 2 of whom died. In the pericardial valve group, a diagnosis of valve dysfunction was established at autopsy in 2 additional patients (O.lO%/patient-year). Overall, primary tissue failure occurred in 98 porcine valves (2.0%/patient-year) and in 19 pericardial valves (0.98%/patient-year) ( p < 0.01). With both bioprostheses, the incidence was higher among mitral than aortic valves (see Table 4). After 6 years, the actuarial freedom rate was similar for both valves in the aortic position (98% f 1% and 94%? 4%; p > 0.05), as well as for patients with MR (95% f 2% and 94% f 4%), but there was a significant difference between the two curves for mitral bioprostheses (92% 5 2% versus 70% + 11% at 6 years; p < ) (Fig 6). After 10 years, freedom from primary tissue failure for porcine valves was 82% * 5% for AVR, 54% f 6% for MVR, and 42% +. 9% for MR. Freedom from all valve-related complications after 6 years was significantly better with porcine than pericardial prostheses, with respective rates of 90% f 2% and 79%+ 5% for AVR (p = 0.05), 80% * 2% and 56% + 10% for MVR (p < O.OOOl), and 85% + 4% and 69% f 6% for MR (p < 0.05) (Fig 7). As for valve survival at 6 years, it was slightly better with porcine valves in AVR (89% f 2% versus 82% f 4%) and MR (84% f 4% versus 74% f lo%), although not significantly so. However, in the mitral position, porcine valves survived significantly

6 1989; PELLETIER ET AL 357 % EVENT -* PERCARDIAL VALVES FREE ao _. 90 SURVIVAL r, A PERCARDIAL VALVES AORTIC VALVE Fig 6. Freedom from primary tissue failure. (N.S. = not significant; REPL = replacements.) longer than pericardial bioprostheses (80%? 2% versus 62% * 10% at 6 years; p < 0.01) (Fig 8). Among the three different types of pericardial bioprosthesis implanted in this series, the Ionescu-Shiley valve accounted for 58% of reoperations (23/40 patients) and for 63% of primary tissue failures of the valve (12/19 patients). The higher complication rate in the pericardial group was almost completely due to the use of this prosthesis in the mitral position; the Ionescu-Shiley valve was responsible for 65% of reoperations (11/17 patients) and for 83% of primary tissue failure (10/12 patients) of mitral pericardial bioprostheses (Fig 9). % PERICARDIAL VALVES AORTIC VALVE EVENT FREE '::w Fig 8. Actuarial valve survival (freedom from cardiac-related death and from reoperation). (N.S. = not significant.) CLINICAL IMPROVEMENT. Of the 573 patients with porcine valves who survived without reoperation, 94% (540 patients) remained in functional class I or I1 at their last follow-up visit an average of 79 months postoperatively. With pericardial valves, 98% (568 patients) of the 581 survivors without reoperation were in class I or I1 after an average follow-up of 36 months. Overall, only 46 patients (4%) remained markedly incapacitated (class I11 or IV) when last seen (Fig 10). Comment The choice of a valve substitute remains a matter of controversy. Comparing various series in the literature can be misleading because of differences in patient choice, management, or follow-up. In addition, nonconcurrent studies can introduce a bias favoring the more recent prosthesis. Two studies [4, 51 published in 1987 compared CARPENTIER-EDWARDS MITROFLOW NUMBER OF PATIENTS REOPERATION 40 Q IONESCU-SHILEY PRIMARY TISSUE FAILURE Fig 7. Freedom from all valve-related complications. (REPL = replacements.) 10 3 Rsissss AVR MVR MR ALL i AVR MVR MR ALL VALVE POSITION Fig 9. Reoperation and prinzary tissue failure of pericardial bioprostheses according to valve type and position. (AVR = aortic valve replacement; MR = multiple-valve replacement; MVR = mitral valve replacernen t.)

7 358 PELLETIER ET AL 1989; PORCINE VALVES PERICARDIAL VALVES and 2.4%/patient-year for pericardial valves, which is YO much lower than that reported with most mechanical PREOP I 70 1 POSTOP valves [ ! Endocarditis had a low incidence with both valves, 50 -! although it was significantly more common among pa i tients who underwent MR with pericardial prostheses. A 30 -! more aggressive attitude toward early surgical treatment 20 - I of complicated endocarditis in recent years might partly i 10 - account for this difference (101. The higher rate of reopern" - b ation because of endocarditis among pericardial valve I II 111 IV I II 111 IV patients is probably also the result of this therapeutic FUNCTIONAL CLASS (NYHA) trend. With the exception of this latter group, freedom from endocarditis after 5 years was 95% or higher with Fig 10. Nm York Heart Association (NYHA) functional class preboth types of valves, and it remained so up to 10 years operatively and postoperatively for each group. with porcine bioprostheses. No clinically significant hemolysis occurred when valve function was normal in the present study. In a previous biological and mechanical valves. After 5 years, the prob- analysis of the Carpentier-Edwards pericardial valve [ 111, ability of remaining free from valve-related complications only trivial biochemical signs of hemolysis were found, was significantly higher with bioprostheses than with whereas a slightly higher degree of hemolysis was remechanical valves in both the aortic and mitral positions ported with the Ionescu-Shiley valve, although hemolysis [4]. This difference was mainly due to the increased remained at subclinical levels [12]. incidence of major though nonfatal episodes of bleeding Of all valve-related complications, degeneration and with the latter type. At 10 years, no significant differences failure of bioprostheses remains the major issue at stake. in the overall valve-related mortality and morbidity were Although freedom from reoperation was similar for porfound between the two types of valves [5]. Although cine and pericardial valves in the aortic position up to 6 mechanical valves had a greater incidence of thromboem- years postoperatively, in the mitral position it remained bolism and reoperation for perivalvular leak, biopros- consistently better with porcine valves after the third theses had a higher rate of primary valve failure. These year. With MR, there was also a tendency for less reopstudies suggest that there is still an important place for eration with porcine valves but the difference did not bioprostheses in clinical use, despite present evidence reach significance. Primary tissue failure was responsible indicating that they will all fail with time. for 88% of reoperations in the porcine valve group and In the present series, the two groups of patients overlap 48% in the pericardial valve group. However, a much over most of the study period, although a majority of higher proportion of the latter group underwent reoperporcine valves were implanted before 1982 and most ation because of endocarditis, 30% in contrast to less than pericardial valves were used during the last 6 years. This 4%, particularly among patients who underwent MR. may partly account for the higher early mortality (9%) in With either type, reoperation was less likely after AVR the porcine valve group compared with the pericardial than MVR, and at 10 years, freedom from reoperation was valve group (5.2%). However, overall patient survival 83% and 53%, respectively, after replacement with porafter up to 6 years was similar in the two groups. Our cine bioprostheses. Whereas the actuarial freedom curves 5-year and 10-year survivals of 80% and 62%, respec- for the aortic position showed a constant but progressive tively, compare with those of others [3, 51. There was no fall, mitral valve curves indicated a steeper decline startdifference in late survival between the two groups, what- ing at the 3-year mark for pericardial valves and at 8 years ever the valve position. for porcine valves. Completeness of follow-up was excellent, as postoper- Freedom curves for primary tissue failure of porcine ative evolution status was known for 97.9% of all patients, valves were very similar to those for reoperation because and lost patients represented a similar proportion in each valve failure was responsible for most reoperations. group. Among late survivors, excellent clinical improve- Curves of primary failure of pericardial valves were not ment was obtained after valve replacement with either different from those for porcine valves in the aortic type after an average follow-up of 6.5 years with porcine position and with MR. This is in contrast with the results valves and 3 years with pericardial valves. of a study comparing Ionescu-Shiley pericardial with Although there was a tendency for a higher incidence of Hancock porcine aortic valves, which indicate a greater thromboembolic events with pericardial valves in all po- tendency to tissue failure with the former, starting at the sitions, freedom curves did not reach significance up to 6 sixth year [l]. However, in the present study, Ionescuyears. The proportion of patients receiving anticoagulants Shiley valves accounted for only one third of AVRs in the was similar, but that of patients with antiplatelet agents pericardial group, which may explain this discrepancy. In was greater among pericardial patients. On the other a previous study [ll] of Carpentier-Edwards aortic perihand, the linearized rate of hemorrhagic complications cardial valves, no primary tissue failure occurred up to 5 was strikingly low with both valves (0.35%/patient-year years and freedom from reoperation for any cause was and 0.4l%/patient-year). The overall rate for those two 98%. On the other hand, pericardial valves showed a complications was 1.4%/patient-year for porcine valves much higher degree of degeneration in the mitral posi-

8 1989; PELLETIER ET AL 359 tion, with a freedom rate after 6 years of 70% compared with 92% with porcine prostheses. The Ionescu-Shiley pericardial valve was responsible for 83% of cases of mitral valve failure. Early degeneration of Ionescu-Shiley and Hancock pericardial valves has been found to occur at alarming rates, leading to their withdrawal from clinical use [ In an animal model, pericardial valves (Ionescu-Shiley and Mitroflow) were shown to develop calcification more extensively and more rapidly than Carpentier-Edwards and Hancock porcine valves [17]. An in vitro study (181 stressed the role of mechanical and design factors in the genesis of tissue degeneration, pericardial valves failing mainly because of damage sustained during valve closure, whereas porcine valves appeared to be damaged mainly during the opening phase of the cardiac cycle. Whereas calcification of porcine valves appears to be related to areas of maximum stress on the leaflet [19], leaflet tears of Ionescu-Shiley pericardial valves are mostly related to the holding commissural suture at the summit of the stent [20, 211. Our long-term results with the Carpentier-Edwards porcine bioprosthesis are similar to those of others with either the same valve or the Hancock porcine valve [3, 22, 231, with an overall valve survival rate of 88% * 1% and 51% * 3% after 5 and 10 years, respectively. Reported rates of valve survival with frozen aortic homografts range from 70% to approximately 85% at 5 years and from 57% to 59% after 10 years [24, 251. Although there is a tendency for survival of porcine valves to be better than that of pericardial valves in all positions, it reaches significance only with mitral prostheses. However, if freedom from all valve-related complications is considered, results are significantly better with the former in all positions. In conclusion, the Carpentier-Edwards porcine bioprosthesis appears to perform better than pericardial valves after the third to fourth year. This is particularly true in terms of the durability of mitral valves. However, the differences between the two types of bioprostheses appear to be mostly due to early tissue degeneration of the Ionescu-Shiley pericardial valve. A more detailed analysis of each type of pericardial valve should be undertaken to clarify its specific performance individually. We thank Manon Bellemare for her assistance in patient follow-up and data collection. References Nistal F, Garcia-Satue E, Artinano E, Duran CMG, Gallo I. Comparative study of primary tissue valve failure between Ionescu-Shiley pericardial and Hancock porcine valves in the aortic position. Am J Cardiol 1986;57:1614. Jamieson WRE, Rosado LJ, Munro AI, et al. Carpentier- Edwards standard porcine bioprosthesis: primary tissue failure (structural valve deterioration) by age groups. Ann Thorac Surg 1988;46: Milano AD, Bortolotti U, Mazzucco A, et al. Performance of the Hancock porcine bioprosthesis following aortic valve replacement: considerations based on a 15-year experience. 1988;46: Hammermeister KE, Henderson WG, Burchfield CM, et al. Comparison of outcome after valve replacement with a bioprosthesis versus a mechanical prosthesis: initial 5 year results of a randomized trial. J Am Coll Cardiol 1987;10: Hammond GL, Geha AS, Kopf GS, Hashim SW. Biological versus mechanical valves. Analysis of 1,116 valves inserted in 1,012 adult patients with a 4,818 patient-year and a 5,327 valve-year follow-up. J Thorac Cardiovasc Surg 1987;93: Karp RB, Cyrus RJ, Blackstone EH, Kirklin JW, Kouchoukos NT, Pacific0 AD. The Bjork-Shiley valve. Intermediate follow-up. J Thorac Cardiovasc Surg 1981;81: Miller DC, Oyer PE, Stinson EB, et al. Ten- to fifteen-year reassessment of the performance characteristics of the Starr- Edwards model 6120 mitral valve prosthesis. J Thorac Cardiovasc Surg 1983;85: Nitter-Hauge S, Semb 8, Abdelnoor M, Hall KV. A 5-year experience with the Medtronic Hall disc valve prosthesis. Circulation 1983;68(Suppl 2): Carrier M, Martineau JP, Bonan R, Pelletier LC. Clinical and hemodynamic assessment of the Omniscience prosthetic heart valve. J Thorac Cardiovasc Surg 1987;93:30G Pelletier LC, Baillot R, Auger P, Dyrda I. Early valve replacement in active infective endocarditis. Can J Surg 1984;2738> Pelletier LC, Leclerc Y, Bonan R, Crepeau J, Dyrda I. Aortic valve replacement with the Carpentier-Edwards pericardial bioprosthesis: clinical and hemodynamic results. J Cardiac Surg 1988;3(SuppI): Febres-Roman PR, Bourg WC, Crone RA, Davis RC Jr, Williams TH. Chronic intravascular hemolysis after aortic valve replacement with Ionescu-Shiley xenograft: comparative study with Bjork-Shiley prosthesis. Am J Cardiol1980;46: Reul GJ Jr, Cooley DA, Duncan JM, et al. Valve failure with Ionescu-Shiley bovine pericardial bioprosthesis: analysis of 2,680 patients. J Vasc Surg 1985;2: Gallo 1, Nistal F, Revuelta JM, et al. Incidence of primary tissue valve failure with the Ionescu-Shiley bovine pericardial bioprosthesis. Primary results. J Thorac Cardiovasc Surg 1985;90:27% Wheatley DJ, Fisher J, Reece IJ, et al. Primary tissue failure in pericardial heart valves. J Thorac Cardiovasc Surg 1987;94: Bortolotti U, Milano A, Thiene G, et al. Early mechanical failures of the Hancock pericardial xenograft. J Thorac Cardiovasc Surg 1987;94:20@ Gallo I, Nistal F, Artinano E, et al. The behavior of pericardial versus porcine valve xenografts in the growing sheep model. J Thorac Cardiovasc Surg 1987;93: Gabbay S, Kadam P, Factor S, Cheung TK. Do heart valve bioprostheses degenerate for metabolic or mechanical reasons? J Thorac Cardiovasc Surg 1988;95:20% Sabbah HN, Hamid MS, Stein I'D. Mechanical stresses on closed cusps of porcine bioprosthetic valves: correlation with sites of calcification. 1986;42: Walley VM, Keon WJ. Patterns of failure in lonescu-shiley bovine pericardial bioprosthetic valves. J Thorac Cardiovasc Surg 1987;93: Trowbridge EA, Lawford PV, Crofts CE, Roberts KM. Pericardial heterografts: why do these valves fail? J Thorac Cardiovasc Surg 1988;95: Bolooki H, Kaiser GA, Mallon SM, Palatianos GM. Comparison of long-term results of Carpentier-Edwards and Hancock bioprosthetic valves. 1986;42: Nistal F, Artiiiano E, Gallo I. Primary tissue valve degener-

9 360 PELLETIER ET AL 1989;47:35241 ation in glutaraldehyde-preserved porcine bioprostheses: Hancock 1 versus Carpentier-Edwards at 4- to 7-years follow-up. 1986;42:56% Angell WW, Angell ID, Oury JH, Lamberti JJ, Grehl TM. Long-term follow-up of viable frozen aortic homografts. A 25. viable homograft valve bank. J Thorac Cardiovasc Surg 1987;93: Matsuki 0, Robles A, Gibbs S, Bodnar E, Ross DN. Longterm performance of 555 aortic homografts in the aortic position. 1988;46: DISCUSSION DR GEORGE J. REUL (Houston, TX): I congratulate Dr Pelletier and the group at the Montreal Heart Institute for a very concise analysis completing an important chapter of valve surgery. Our experience with the Ionescu-Shiley pericardial valve at the Texas Heart Institute closely parallels their experience with the pericardial valve. The major difference was in freedom from reoperation in the mitral valve and aortic valve patients. At 6 years, freedom from reoperation was the same in the aortic valve and mitral valve patients, and at 7 years it was higher in our aortic valve group. Dr Pelletier has shown that the reoperation rate was higher in the mitral valve patients at 4 years. Furthermore, at 6 to 7 years there was almost a 10% increase in our reoperation rate. The reason that aortic valve reoperation was required more frequently than mitral valve reoperation in our patients was because of the mode of tissue failure. Dr Pelletier has shown that leaflet disruption was the major mode of tissue failure. Most of the follow-up data of others have shown that the pericardial valve calcifies at a faster rate than the porcine valve. Our increased incidence of aortic valve reoperation was mostly due to Calcification. Under the age of 30 years, the mean time for occurrence of calcification was almost 3.5 years, whereas in the older age group, the mean time of calcification was about 5 years. Because of the older mean age in Dr Pelletier s series, the aortic valve failure rate will probably equal the mitral valve failure with more time for follow-up. Because of bias in patient selection, some of Dr Pelletier s comparisons may not be valid. The pericardial valve was used more frequently in patients with aortic stenosis and with prosthetic valve failure. Why did the surgeon choose the pericardial valve in preference to the porcine valve? Is this because a pericardial valve can be used in a patient with a smaller annulus compared with the porcine valve? Only 8% of the patients were functional class IV and only about 2% of the patients had enlargement procedures on the annulus. Were mechanical valves used in other patients not included during this time frame? Another area of concern was the higher incidence of endocarditis in the pericardial valve compared with the porcine valve. Our incidence of endocarditis with the lonescu-shiley valve was exactly the same as their incidence of endocarditis with the porcine valve, that is, 0.4%/patient-year. Was there a higher incidence of endocarditis in the pericardial valve patients because endocarditis was present preoperatively more frequently in the pericardial valve patients? It would be of interest to stratify the results in the pericardial group according to valve type. Do you have enough data to show that the Ionescu-Shiley valve is different from the Carpentier- Edwards valve and the other pericardial valves utilized? Finally, a question that we d all like you to answer based on these data is whether or not there is any indication at the present time for the use of a pericardial valve. DR ROBERT W. M. FRATER (Bronx, NY): I m a great admirer of Dr Pelletier s work, and I m very glad that in his last sentence he made the point that it is necessary to distinguish between different kinds of pericardial valves. The key point about pericar- dial valves is that they are constructed-they are not harvested, God didn t make them-and as a result of that, every single pericardial valve is different. 1 don t believe you can group pericardial valves for follow-up, and 1 don t believe you can draw conclusions about pericardial valves in general. The conclusions must be type specific. As far as the Ionescu-Shiley valve is concerned, we had cohorts of patients operated on between 1975 and 1980 with Starr ball valves, Hancock porcine valves, valve repairs of an extensive kind, and the Ionescu-Shiley bovine valve, all in the mitral position. There s absolutely no question about the inferior performance in the mitral position of the Ionescu-Shiley valve. The steep portion of the freedom from valve-related deterioration curve represents tears, the portion of the curve that is parallel with the Hancock valve represents calcification. However, when we look at valve survival in the aortic position, the curves of freedom from valve-related death and valve-related deterioration for the bovine Ionescu-Shiley and the porcine Hancock valves are precisely the same. This same device made of the same material behaves completely differently in the aortic and the mitral position as Dr Pelletier pointed out, and the reason in this particular case is probably a very simple one. When you stretch pericardium across a large orifice it cannot stand the stress. When you stretch it across a small orifice the mode of failure becomes calcification as opposed to tearing. 1 would just ask Dr Pelletier to confirm for this audience, please, that when you talk about pericardial valves you must talk about them specifically and not in general. DR SHLOMO GABBAY (Newark, NJ): I congratulate Dr Pelletier for this excellent presentation. I was delighted as well to hear the last sentence about the Carpentier-Edwards pericardial valve, which in up to 6 years of follow-up shows excellent results. Pericardial valve obviously is not a generic name. Each known pericardial valve stands on its own merit due to the difference in design. We cannot blame the Carpentier-Edwards pericardial valve for the problem of the Ionescu-Shiley valve. The Bjork- Shiley valve is a pyrolitic carbon valve as is the St. Jude valve, and surgeons do not think all pyrolitic carbon valves are the same. I predicted the superior clinical behavior of the Carpentier- Edwards pericardial valve over the Ionescu-Shiley valve more than 5 years ago. I based my prediction on in vitro accelerated fatigue testing and closed observations of clinically explanted bioprostheses. Dr Pelletier s conclusions about the superiority of the Carpentier-Edwards pericardial valve can only enforce the claim that pericardial valves are not all alike, as well as stressing the importance of in vitro fatigue testing as an excellent tool for evaluating new designs. Another example is the prediction made in the same study relative to the Hancock pericardial valve (Vascor). This valve performed so badly in vitro that I had strong doubts about whether it should have gone to clinical trials. Needless to say, the valve was explanted by the hundreds 1 to 3 years after implantation. Even the Ionescu-Shiley valve, with some minor modifications,

10 1989; PELLETIER ET AL 361 can be considered a very good valve. Thousands of patients have had this valve for more than 10 years and are still doing very well. A good deal of understanding of the quality of pericardial tissue selected for cusp material is available today. Histological studies of clinical Ionescu-Shiley valves done in our laboratories show that of the three cusps of each valve, at least two cusps have substantial collagen degeneration before implantation. Rarely, two cusps of the same valve could be considered "good cusps" by the new standards. This technology was not available when these valves were made. We have good reason to believe that the quality of the tissue selected for cusp material is an important factor in the variability of durability of the same generic valve. We have seen that the pericardial sac has many anatomical as well as histological variations. Pericardium might look adequate to the naked eye but not in the microscope. With improvements in design as well as improved quality control of the biological material, and prolonged durability proven in a fatigue tester, pericardial valves can be made to last much longer than the present proven record. My belief is that the last chapter on pericardial valves has not been written yet. DR PELLETIER I thank the discussants for their nice comments. Dr Red, in regard to the choice of the various types of bioprostheses, I must stress that this was not a randomized study, and the choice of the prosthesis was left to the individual selection of each surgeon. Enlargement of the aortic root was rarely performed in this series, and this is a technique that we don't use very readily. We'd rather use a smaller aortic prosthesis, and this is probably one of the factors that explain the shift from porcine to pericardial valves in latter years, with the hope that these bioprostheses would do better in the small aortic size. In regard to the increased incidence of endocarditis in doublevalve replacement with pericardial bioprostheses, I have no clear explanation for that observation other than that it may reflect the tendency in recent years for early operation in complicated cases of endocarditis, which include particularly prosthetic valve endocarditis and double-valve endocarditis. As for the various types of pericardial prosthesis, you inquired about the number implanted in each group and whether a difference could be demonstrated. We have looked at that in a more detailed study which is actually going on, and the data that we've had so far indicate that indeed there are significant differences between Ionescu and Carpentier-Edwards valves, and that the numbers are sufficient to show such differences in valve survival, particularly in the mitral position. As for the Mitroflow prosthesis, it's a little more difficult to tell because the numbers are smaller, though actually it appears that in the mitral position, valve survival with the latter falls right in between the other two valves. You questioned whether there remains an indication for the use of pericardial valves today. I think this is a matter of individual choice. Actually, there are no reasons for rejecting all pericardial valves. We have shown that some pericardial valves do not perform adequately in the mid-term and long-term range, but that others, such as the Carpentier-Edwards, appear to perform as well as the porcine valves, at least up to 6 years. So I think this remains an individual choice, although there is no clear advantage of one over the other. Dr Frater, you stressed the point that the various pericardial valves are different, and I completely agree with you. This is clearly shown in our study, and our data support this conclusion. As major differences were found between porcine and pericardial valves, specific analysis of each type was necessary among the pericardial group, as shown in the last part of our data. Dr Gabbay, I agree with you that the fate of the various prostheses could be predicted based on an analysis of the design of the valve, because I am convinced that early failure of the Ionescu-Shiley valve, at least in the mitral position, is related to technical features in its design that differ markedly from other pericardial valves, particularly the Carpentier-Edwards.