P have been used for mitral and aortic valve replacement

A -Year Comparison of Mitral Valve Replacement With Carpentier-Edwards and Hancock Porcine Bioprostheses P. Perier, MD, A. Deloche, MD, S. Chauvaud, MD, J. C. Chachques, MD, J. Relland, MD, J. N. Fabiani, MD, Y. Stephan, MD, P. Blondeau, MD, and A. Carpentier, MD, PhD Department of Cardiovascular Surgery, HBpital Broussais, Paris, France Two hundred fifty-three patients who underwent isolated mitral valve replacement with a porcine bioprosthesis had long-term evaluation. One hundred fortyseven patients received a Carpentier-Edwards porcine bioprosthesis and 6, a Hancock valve. There were no significant differences in preoperative clinical characteristics between the two groups. Cumulative follow-up was,375 patient-years. At years, 93% 2 2.5% of the patients in the Carpentier-Edwards group and 85% f 7.8% of those in the Hancock group were free from valve-related death (not significant), and 95% f 2% and 9% f 3.8%, respectively, were free from thromboembolism (not significant). At years, 65% f 7.2% of the patients in the Carpentier-Edwards group and 66% f 7.2% of those in the Hancock group were free from structural valve deterioration (not significant), and 64% f 6% and 59% f 7.3%, respectively, were free from reoperation (not significant). We conclude that the first generation of Carpentier-Edwards and Hancock prostheses produce comparable long-term results in the mitral position. (Ann Thorac Surg 989;48:54-9) orcine bioprostheses treated with glutaraldehyde P have been used for mitral and aortic valve replacement since 968 [l]. However, only since 974 has the bioprosthetic valve been commercially available and widely implanted. The Carpentier-Edwards prosthesis and the Hancock prosthesis have been the most frequently used porcine bioprostheses. They differ somewhat in the glutaraldehyde concentration and the type of supporting stent. During the period January 975 to December 979, both types of porcine bioprostheses were commonly used in our institution without special selection or randomization; the choice depended mainly on the preference of the individual surgeon. The aim of this study was to compare long-term performance characteristics of Hancock or Carpentier-Edwards porcine mitral bioprostheses implanted in the same institution by the same team and over the same period. Material and Methods Two hundred fifty-three patients who underwent isolated mitral valve replacement with a porcine bioprosthesis between January 975 and December 979 constitute the material of this study. They were separated into two groups based on the type of valve implanted: 47 patients received a Carpentier-Edwards valve and 6, a Hancock Accepted for publication Jan 2, 989. Address reprint requests to Dr Perier, Clinique de Chirurgie Cardio- Vasculaire HBpital Broussais, 96, rue Didot, 754 Paris, France. valve. Age ranged from 8 to 77 years (mean age, 53.6 &.3 years and 5.2 * 3.3 years, respectively). Analysis of multiple preoperative variables-mean age, proportion of patients aged less than and more than 35 and 55 years, mean preoperative functional class, mean cardiothoracic ratio, incidence of atrial fibrillation, mitral valve lesion, and etiology-did not show any significant differences between the two cohorts (Table ). nformation on operative mortality and complications was collected as part of the follow-up. Long-term followup was completed over a 4-month interval (January through April 987) through questionnaires and telephone contacts with patients and their referring physicians. The average duration of follow-up of surviving patients was 7.2 2 2.3 years for the Carpentier-Edwards group and 6.8 2 2.9 years for the Hancock group (range, 7 to 2 years). The cumulative follow-up was 826 patientyears for the Carpentier-Edwards group and 549 patientyears for the Hancock group. Postoperative Anticoagulation All patients received immediate postoperative anticoagulant therapy with subcutaneous administration of heparin sodium beginning six hours after operation and continuing for ten days thereafter. Oral anticoagulation with warfarin sodium was initiated at that time and was eventually discontinued after 3 months depending on the decision of the referring cardiologist, rather than that of the surgeon, provided the patient was in sinus rhythm. Seventy-five percent Of the surviving patients in both groups are on a regimen of long-term anticoagulation. 989 by The Society of Thoracic Surgeons 3-4975/89/$3.5

Ann Thorac Surg 989:48:54-9 PERER ET AL 55 Table. Preoperative Clinical Characteristics of the Two Patient Groupsa4 Variable Age Mean (yr) Patients <35 yr Patients >35 yr Patients <55 yr Patients >55 yr Preoperative functional class Preoperative cardiothoracic ratio Cardiac rhythm Sinus rhythm Atrial fibrillation Predominant functional lesion Mitral stenosis Mitral regurgitation Mixed mitral lesions Etiology Rheumatic fever Degenerative disease schemic disease Endocarditis Previous prosthesis Carpentier- Edwards Hancock (n = 47) (n = 6) 53.6?.3 8 39 76 7 2.6 t.7.6?.7 52 (35) 95 (65) 27 (8) 48 (33) 72 (49) 3 (7) 37 (25) 3 (2) 2 () 2 () 5.2 rf: 3.3 7 99 56 5 2.5?.6.59?.6 3 (2) 93 (88) 7 (6) 3 (28) 59 (56) a There were no significant differences between the two groups. Where applicable, data are shown as the mean? the standard deviation. Numbers in parentheses are percentages. Definitions Stringent definitions for all valve-related complications derived from the Stanford framework were used in this study [2]. Structural valve deterioration was defined as an intrinsic abnormality of the prosthesis (leaflet disruption, calcification) causing stenosis or regurgitation found at autopsy or reoperation. - 5t6 d b 747 73 4 98 94 8 79 58 25 C-E 6 8 73 69 6 6 49 45 39 26 HCK ' 4 2 3 4 5 7 8 9 Fig. Actuarial survival curves. Operative mortality is included. (C-E = Carpentier-Edwards; HCK = Hancock.) Thromboembolism was defined as the occurrence of any new focal neurological deficit, whether transient or permanent, unless it could be proved unequivocally to be due to another cause. Peripheral emboli were also included in this category unless the embolus clearly did not arise from the heart. This category also included valve thrombosis. Anticoagulant-related hemorrhage included all episodes of internal or external bleeding resulting in hospitalization, transfusion, or death. Reoperation included all valve reoperations regardless of the indication, including re-replacement and suturing of a periprosthetic leak. Valve failure encompassed all valve-related complications prompting reoperation or causing death (structural valve deterioration, prosthetic valve endocarditis, periprosthetic leak, hemodynamic valvular dysfunction, thromboembolism, anticoagulant-related hemorrhage). Valve-related death was defined as valve failure leading to death. Statistical Analysis Standard actuarial and linearized techniques were used to describe survival and the incidence of valve-related complications [3-5. The log-rank test was used to compare actuarial curves [6]. Continuous data are presented as the mean? the standard error of the mean. Patients continued in the study until valve removal, death, or loss to follow-up. Patients were removed "alive" Table 2. Causes of Hospital Death" Cause Valve thrombosis Cardiac failure Myocardial infarction Aortic dissection Surgical hemorrhagic complication Sudden death Sepsis Total a Numbers in parentheses are percentages Carpentier-Edwards (n = 47) 9 2 3 (8.8) Hancock (n = 6) 3 2 9 (7.9) Table 3. Causes of Late Death Cause Carpentier-Edwards (n = 47) Valve-related death 7 Cardiac-related 2 death Sudden death 9 Noncardiac Cancer 4 Other 3 Unknown 4 Total 39 Hancock (n = 6) 3 9 8 3 2 26

56 PERER ET AL Ann Thorac Surg 989;48:549... 8.. 7-: 6.. 5.. i......... L 93f2 5 9 *-.a 85f78 6 3 4 98 94 89 79 58 25 C-E 47 8 73 69 67 6 49 4 38 27 HCK Table 5. Causes of Valve-Related Death Carpentier-Edwards Hancock Cause (n = 47) (n = 6) Valvular thrombotic 2 occlusion Thromboembolism Anticoagulant-related 4 hemorrhage Structural valve deterioration 3 Total 7 4 from the study when the original valve was replaced or when an additional valve replacement was required. Only the first event for each patient during the study was considered for the actuarial analysis, whereas all events for each patient were considered in the linear analysis. Results Operative Mortality Thirty-day operative mortality was 8.8% in the Carpentier-Edwards group and 7.9% in the Hancock group. The causes of death are listed in Table 2. One of the early deaths in each group was valve related (valve thrombosis). Late Survival Long-term actuarial survival is shown in Figure. Tenyear survival rates were 55% & 4.7% in the Carpentier- Edwards group and 5% & 6% in the Hancock group. The difference between the two groups was not significant. The causes of late death were grouped into five different categories: valve-related death, cardiac-related death, sudden death, noncardiac cause, and unknown (Table 3). Table 4. Summary of Linearized Rates of Valve-Related Valve- Related Deaths At years, 93% 2 2.5% of the patients in the Carpentier- Edwards group and 85% & 7.8% of those in the Hancock group were free from valve-related death (Fig 2). Statistical analysis showed no significant difference between the two groups. The linearized rates were.9% 2.3% per patient-year for the Carpentier-Edwards group and.9% f.4% per patient-year for the Hancock group (Table 4). The causes of valve-related death are listed in Table 5. ncluding sudden death in the causes of valve-related death, 83% & 4.8% of the patients in the Carpentier- Edwards group and 76% 2 8.2% of those in the Hancock group were free from valve-related death. Again, no significant difference was seen between the two groups. Throm boembolism On an actuarial basis, 95% & 2% of the patients in the Carpentier-Edwards group and 9% 2 3.8% of the patients in the Hancock group were free from thromboembolic complications at years (Fig 3). The linearized rates were.8% &.3% and.% 2.4% per patient-year, respectively (see Table 4). Differences between the two groups were not significant. Two fatal episodes of thrombotic valvular occlusion occurred in the Carpentier- Edwards group and one in the Hancock group. The other thromboembolic complications in both groups were transient and regressive. Carpentier-Edwards Hancock Complication (n = 47) (n = 6) Valve-related death Thromboembolism Structural valve deterioration Reoperation Valve failure.9 f.3.8 t.3 2.9 t.5.9 t.4.?.4 2.6 t.7 9 8 7.. 3.5?.6 4.?.8 6.. 4. 2.7 4. f.8 5..... ~...f...i...p... k-a 94 7t 2 W! 9'38.. m Anticoagulant-related?.3.3 t.4 ' hemorrhage Valve-related mortality and 4.8 t.7 5.6 f:. morbidity a Data are shown as percent per patient-year. Data are shown as the Fig 3. Actuarial curves of freedom frorn thromboe~bozisrn. C-E = Carpentier-Edwards; HCK = Hancock.) mean t the standard error of the mean. 47 5 99 96 9 79 6 25 C-E 6 77 7 68 67 59 47 42 36 25 HCK

Ann Thorac Surg 989;48:54-9 PERER ET AL 57. 9. 8. 7. 6. 5.. m 65t72 + - - 66?72 47 5 3 8 97 92 6 6 25 C-E *. Table 6. Causes of Reoperation Carpentier-Edwards Hancock Cause (n = 47) (n = 6) Structural Thrombosis valve deterioration 242 4 Periprosthetic leak 3 4 Endocarditis 3 6 79 74 7 69 62 5 44 39 26 HCK nadequate size.. ; i i i i j 6 7 8 9 Total 29 23 Fig 4. Actuarial representation of freedom from structural valve deterioration. (C-E = Carpentier-Edwards; HCK = Hancock.) Structural Valve Deterioration At years, 65% * 7.2% of the patients in the Carpentier- Edwards group and 66% * 7.2% of those in the Hancock group were free from structural valve deterioration (Fig 4). The difference was not significant. All of the patients with diagnosed structural valve deterioration were scheduled for an elective reoperation. Of the 24 patients with structural valve deterioration in the Carpentier-Edwards group and the 4 in the Hancock group who had reoperation, (4%) in the Carpentier-Edwards group died (postoperative heart failure) and 3 (2%) in the Hancock group died (2 of mediastinitis and of postoperative heart failure). Linearized rates are shown in Table 4. The prosthetic valves used at reoperation were as follows: 23 Starr-Edwards, porcine bioprostheses, and 5 St. Jude Medical. t is of note that in the Carpentier-Edwards group, four of the cases of structural valve deterioration occurred in the 8 patients younger than 35 years. n this same group, 6 cases occurred in the 76 patients younger than 55 years, and only eight cases were found in the 7 patients older than 55 years. The same situation prevailed in the Hancock group, with three cases of structural valve deterioration in the 7 patients younger than 35 years old. The 56 patients younger than 55 years experienced ten cases of structural valve deterioration, whereas the 5 patients older than 55 experienced four cases. Reoperation On an actuarial basis, at years, 64% & 6% in the Carpentier-Edwards group and 59% +- 7.3% in the Hancock group have undergone reoperation (Fig 5). The difference was not significant. The linearized rates of reoperation are shown in Table 4. The causes of reoperation are listed in Table 6. The mortality for reoperation was % in the Carpentier-Edwards group (3/29) and 7% in the Hancock group (4/23). Valve Failure At years, 6% * 6.5% of the patients in the Carpentier- Edwards group and 59% 5 7.3% of those in the Hancock group were free from valve failure (Fig 6). The difference was not significant. Causes of valve failure are listed in Table 7. t is noteworthy that in the Carpentier-Edwards group, 29 of the 34 patients with valve failure were able to have a reoperation and all the patients with valve failure in the Hancock group underwent reoperation. Seven patients in the Carpentier-Edwards group died as a result of valve failure (3, reoperation; 4, anticoagulant-related complication) and 4 in the Hancock group (4, reoperation). Anticoagulant-Related Hemorrhage At years, 92% 5 2.7% of the patients in the Carpentier- Edwards group and 85% * 5.5% of those in the Hancock 8.. 5 59: 73 47 9 3 99 94 89 79 58 25 C-E 6 78 74 7 68 6 49 44 39 26 HCK Fig 5. Actuarial curves of freedom from reoperation. (C-E = Carpentier-Edwards; HCK = Hancock.) 7.. 5 k 59f73 47 ) 9 3 97 93 8 S 55 25 C-E 6 7 i.4 7 6 6 49 44 38 2 HCK >.. Fig 6. Actuarial curves of freedom from valve failure. (C-E = Carpentier-Edwards; HCK = Hancock.)

58 PERER ET AL Ann Thorac Surg 989;48:54-9 Table 7. Causes of Valve Failure @ Cause Carpentier-Edwards (n = 47) Hancock (n = 6) Structural valve 24 deterioration Thromboembolic 3 complications Periprosthetic leak 3 Endocarditis Anticoagulant-related 4 hemorrhage nadequate size Total 34 4 3 4 23 5 4 49'75 47 5 6 99 9 87 97 58 25 C-E 6 78 72 68 65 57 45 39 32 2 HCK 2 3 4 5 6 7 8 9 Fig 8. Actuarial curves showing percentage of patients free from value-related mortality and morbidity. (C-E = Carpentier-Edwards; HCK = Hancock.) group were free from anticoagulant-related hemorrhage (Fig 7). The difference was not significant. Linearized rates are shown in Table 4. n the Carpentier-Edwards group, 4 patients died as a result of an anticoagulantrelated complication; none in the Hancock group died of this complication. Valve-Related Mortality and Morbidity Combining the incidence of thromboembolic and anticoagulation complications not resulting in reoperation or death with the incidence of valve failure yielded a composite function describing the incidence of valve-related mortality and morbidity. On an actuarial basis, 5% f 7.2% of the patients in the Carpentier-Edwards group and 49% * 7.5% of those in the Hancock group were free from valve-related complications at years. This difference was not significant (Fig 8). The linearized rates are shown in Table 4. Comment n spite of the extensive use since 974 of porcine bioprostheses, mainly the Carpentier-Edwards and the Hancock, few reports are available that compare the long-term results between these two valves [7-9. Using the methods described in a previous report from our institution [lo], this study seems appropriate because it compares two series of patients operated on in the same institution by the same team over a limited period of 5 years with a 8.. 7.. 6.. w."c'.l 5.., 92f27 6 8 74 7 69 6 49 44 37 23 C-E.7 5 33 lo7 99 95 89 77 59 23 HCK La. follow-up of 7 to 2 years. The homogeneity of the two groups has been demonstrated only on the basis of clinical findings (age, preoperative functional class, preoperative cardiothoracic ratio, predominant functional lesion, etiology), as hernodynamic data were not available on all patients. One criticism that could be leveled is that the selection of the type of bioprosthesis placed was not randomized. Rather it was left to the discretion of each individual surgeon who habitually used the same prosthesis during the study period. The operative mortality was high. However, the operative risk has improved in the past years because of better myocardial protection. The current risk for isolated mitral valve replacement is 3%. Our -year survival of 55% and 5% for the Carpentier- Edwards and the Hancock groups, respectively, is similar to that in other reports on bioprostheses in the mitral position [ll, 2. However, the overall late survival, which is determined by patient-related factors and mostly by the condition of the heart at the time of operation, is not a sensitive indicator of the performance of a given prosthesis. The incidence of valve-related death may be a more accurate indicator of the risks associated with a given device. The actuarial rates of freedom from valve-related death (93% in the Carpentier-Edwards group and 85% in the Hancock group) revealed no significant difference. However, in a nonrandomized retrospective study, assessing the cause of death in a population who lived outside the hospital and who often died unobserved leads to a sustained possibility of error, as underlined by Hammond and colleagues [3]. As a result, we separated the patients who died suddenly. Most certainly, many of these patients died of an arrhythmia, but it is ikely that some patients died of a valve-related cause. The computation of valve-related mortality incorporating sudden death did not yield different results. One of the prime reasons for using porcine bioprosthetic valves is to decrease the incidence of thromboembolic complications compared with mechanical valves. As in other reports [4, 5, the rate of thromboembolic complications was low in the two groups (.8% 2.3% per patient-year in the Carpentier-Edwards group and

Ann Thorac Surg 989;4854-9 PERER ET AL 59.% 5.4% per patient-year in the Hancock group). AS other studies [5, 6 have shown, most of the thromboembolic complications occurred within the first postoperative year. The incidence of anticoagulant-related hemorrhage, %?.3% per patient-year in the Carpentier-Edwards group and.3% 2.4% per patient-year in the Hancock group, was relatively high in this report, especially considering that 4 patients in the Carpentier-Edwards group died as a result of anticoagulation complications. Durability remains the major problem of tissue valves, and represents a serious concern for the patient and the surgeon. At the end of years, 35% of the patients in both groups have undergone reoperation for structural valve deterioration, an incidence frequently reported in the literature [ll, 7-9]. As in most other reports [7, 9, no difference was observed in the rate of structural valve deterioration between the Carpentier-Edwards prosthesis and the Hancock prosthesis. One exception was the study of Hartz and co-workers [2], which suggested that the deterioration rate of the Carpentier-Edwards bioprosthesis was less than that of the Hancock valve, but the follow-up was shorter. n both groups, the first cases of structural valve deterioration began occurring in the third postoperative year. One of the major issues with bioprostheses is the operative mortality associated with reoperation for structural valve deterioration. The operative mortality for valve re-replacement was 4% in the Carpentier-Edwards group and 2% in the Hancock group. n a larger series of 9 reoperations for structural valve deterioration of bioprostheses, we [2] found an operative mortality of 5.5%, a risk similar to that of primary operations. t is of note that in 25% of the reoperations for structural valve deterioration, a second kind of bioprosthesis was used (usually at the request of the patient) to avoid anticoagulation, provided failure did not occur within the first 8 years after implantation. This percentage is low compared with that of Magilligan and colleagues [7], who replaced 77% of degenerated valves in the mitral position with a different bioprosthesis. n summary, our study shows that the first-generation Carpentier-Edwards porcine bioprosthesis and the Hancock prosthesis produce comparable long-term results in the mitral position. The problem of structural valve failure in tissue valves may be eased in the recent products by the introduction of new methods of tissue preservation, lowpressure fixation, and improvements in calcium mitigation. References. Carpentier A, Lemaigre G, Ladislas R, et al. Biological factors affecting long term results of valvular heterografts. J Thorac Cardiovasc Surg 969;58:467-83. 2. Miller DC, Oyer PE, Mitchel RS, et al. Performance characteristics of the Starr Edwards model 26 aortic valve pros- thesis beyond ten years. J Thorac Cardiovasc Surg 984; 88:93-27. 3. Fabiani JN, Carpentier A. La mkthode actuarielle pour analyse statistique des rksultats cliniques et expknmentaux. Nouv Presse Med 977;6357-6. 4. Grunkemeier GL, Lambert LE, Bonchek L, Starr A. An improved statistical method for assessing the results of operation. Ann Thorac Surg 975;2:289-98. 5. Grunkemeier GL, Thomas DR, Starr A. Statistical considerations in the analysis and reporting of time related events. Am J Cardiol 977;39:257-8. 6. Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. Br J Cancer 977;35:-39. 7. Cobanoglu A, Jamieson WRE, Miller DC, et al. A triinstitutional comparison of tissue and mechanical valves using a patient-oriented definition of treatment failure. Ann Thorac Surg 987;43:245-53. 8. Bolooki H, Kaiser GA, Mallon SM, Palatianos GM. Comparison of long-term results of Carpentier-Edwards and Hancock bioprosthetic valves. Ann Thorac Surg 986;42:494-9. 9. Nistal F, Artiitano E, Gallo. Primary tissue valve degeneration in glutaraldehyde-preserved porcine bioprostheses: Hancock versus Carpentier-Edwards at 4- to 7-year followup. Ann Thorac Surg 986;42:568-72.. Perier P, Bessou JP, Swanson JS, et al. Comparative evaluation of aortic valve replacement with Starr, Bjork and porcine valve prostheses. Circulation 985;72(Suppl 2):4-5.. Magilligan DJ, Lewis JW, Tilley 8, et al. The porcine bioprosthetic valve: twelve years later. J Thorac Cardiovasc Surg 985;89:499-57. 2. Oyer PE, Stinson EB, Miller DC, et al. Clinical analysis of the Hancock porcine bioprosthesis. n: Cohn LH, Gallucci V, eds. Cardiac bioprostheses. New York: Yorke Medical Books, 982:539-5. 3. Hammond GL, Geha AS, Hashim SW. Biological versus mechanical valves. J Thorac Cardiovasc Surg 987;93:82-98. 4. Jamieson WRE, Pelletier CL, Janusz MT, et al. Five year evaluation of the Carpentier-Edwards porcine bioprosthesis. J Thorac Cardiovasc Surg 984;88:324-33. 5. Oyer PE, Stinson EB, Griepp RB, et al. Long term evaluation of the porcine xenograft bioprosthesis. J Thorac Cardiovasc Surg 979;78:343-5. 6. Perier P, Deloche A, Chauvaud S, et al. Comparative evaluation of mitral valve repair and replacement using Starr, Bjork and porcine valve prostheses. Circulation 984; 7O(Suppl ):87-92. 7. Magilligan DJ, Lewis JW, Heinzerling RH, et al. Fate of a second porcine bioprosthetic valve. J Thorac Cardiovasc Surg 983;85:362-7. 8. Schoen FJ, Collins JJ, Cohn LH. Long term failure rate and morphologic correlations in porcine bioprosthetic heart valves. Am J Cardiol 983;5:957-64. 9. Gallo, Nistal F, Artiitano E. Six to ten year follow-up with patients with the Hancock cardiac bioprosthesis. J Thorac Cardiovasc Surg 986;92:4-2. 2. Hartz RS, Fisher EB, Finkelmeier B, et al. An eight-year experience with porcine bioprosthetic cardiac valves. J Thorac Cardiovasc Surg 986;9:9-7. 2. Perier P, Swanson J, Takriti A, et al. Decreasing operative risk in isolated valve re-replacement. n: Bodnar E, Yacoub M, eds. Biologic bioprosthetic valves. New York: Yorke Medical Books, 986:333-8.