Aortic Valve Replacement with Starr-Edwards Valves over 14 Years

Transcription

1 Aortic Valve Replacement with Starr-Edwards Valves over 4 Years W. H. Wain, B.Sc., Ph.D., P. J. Drury, B.Sc., Ph.D., andd. N. Ross, F.R.C.S. ABSTRACT Three hundred thirteen patients underwent aortic valve replacement with 39 Starr- Edwards caged-ball prostheses and have been followed for to 4 years. Hospital mortality (24.8%) and first-year mortality (4.8%) suggest that this is a high-risk group. Overall postoperative patient survival was 37.%, with 8.8% free from any event, at 4 years. Thromboembolism was the most significant single event (a probability of 8.7% at 4 years), and the Series 2300/2320 valves were associated with a probability of hemolysis of nearly 80% over 3 years. Patients who underwent valve replacement before 973 had a significantly greater probability of late death (3.%) and of complications (47.%) during the first 5 years. Those patients undergoing replacement after 973 had a significantly greater probability of thromboembolic episodes (5.4%). Starr-Edwards caged-ball prostheses provide an acceptable valve replacement for high-risk patients. However, the overall complication rate of 8.2 /~ over 4 years, compared with 78% for homograft valves for the same period, does not support the adoption of this prosthesis as the valve of first choice at this hospital. Aortic homograft valves have been the first choice for aortic valve replacement in isolated aortic valve disease at the National Heart Hospital since 964 [l]. However, there were some conditions for which a homograft was not considered to be suitable treatment, such as aortic regurgitation associated with Marfan's syndrome [2]. There was also a larger group of patients, predominantly in the higher age ranges, including 80 patients for valve rereplacement, who were regarded as high risk and for whom a From the Department of Surgery, Cardiothoracic nstitute, London, England. Accepted for publication Oct 8, 98. Address reprint requests to Mr. Ross, Department of Surgery, Cardiothoracic nstitute, 2 Beaumont St, London WN 2DX, England. caged-ball Starr-Edwards prosthesis offered effective palliation, especially as it is associated with a shorter, simpler, and potentially safer operation. Although Starr-Edwards valves require continuous anticoagulation and bear a risk of thromboembolism [3], such problems were acceptable hazards for this high-risk group of patients, hazards that were offset by the advantages of the simpler, safer operation. There have been a number of variations in the manufacture of Starr-Edwards valves since 964. Three of the variations have been used in the treatment of isolated aortic valve disease in 33 high-risk patients at the National Heart Hospital, and an analysis of the long-term performance of these valves forms the basis of this paper. Material and Methods Three hundred nineteen Starr-Edwards valves were implanted in 33 patients who have been followed for to 4 years. There were 258 male patients (82.4%) ranging in age from 6 to 75 years (50.7 f 3. years [average f standard deviation]) and 55 female patients (7.6%) ranging in age from 7 to 70 years (48.3 f 4.8 years). The different types of Starr-Edwards valves used at the National Heart Hospital between 964 and 980 are shown in Table. Two hundred eighty-two of the operations were isolated procedures (88.4%) while 37 were concomitant with other valve operations (.670). One hundred nineteen of the patients were seen with a complicated diagnosis (38.0 /o), as indicated in Table 2. Eighty of the valves (25.%) were implanted in patients who had a history of previous valve operation. Valve replacement with a Starr-Edwards valve because of infection occurred in nine of these operations and were associated with several conditions. Ninety-four additional cardiac surgical procedures were carried out in 79 patients (25.2%). They included the following: coronary artery operation, 37 patients; mitral valvotomy, 6; by The Society of Thoracic Surgeons

2 563 Wain, Drury, and Ross: Aortic Valve Replacement with Starr-Edwards Valves Table. Types of Starr-Edwards Valves Used at the National Heart Hospital, No. Model Period mplanted Percent 000 (Silastic ball) (Silastic ball) i2320 (Metal ball) " Total "One metal ball was implanted in 974. Table 2. Diagnoses of Patients Receiving Starr-Edwards Aortic Valves Table 3. Causes of Late Death in Patients with Starr-Edwards Aortic Valve Replacement Diagnosis Aortic stenosis Calcified aortic stenosis Aortic regurgitation Aortic stenosis + regurgitation Calcified aortic stenosis + regurgitation Previous valve operation Marfan's syndrome Aneurysm (not Marfan's syndrome) Emergency nfection, native Total No. of Valves (5 infections) 2 7 ( infection) (9 infections) Cause Unknown Thromboemboli nfection Hemolysis Not valve-related Perivalvular leaks Obstruction Aneurysm Congestive cardiac failure Left bundle-branch block Gas embolism Postop cardiac arrest after mitral valve replacement No. of Patients (2 HD) ( HD) HD = death within thirty days of operation. mitral repair, 2; relief of subaortic stenosis myectomy, 5; tricuspid annuloplasty, 6; repair of atrial septal defect or ventricular septal defect, 5; correction of coarctation, ; correction of pulmonary atresia, ; and left ventricular imbrication,. There was a maximum follow-up of 4 years, with a mean of 3.6 years. Valve malfunction was diagnosed if one of the following criteria was evident: mechanical failure; infection; thromboembolic episode; or hemolysis or hemolytic anemia. Actuarial probabilities of survival and eventfree survival were calculated according to Bodnar and colleagues [4]. An examination was also made of the 5-year postoperative survival probabilities for the first 5-year (966 to 97) and the most recent 5-year (975 to 980) experience with the Series valve in order to assess the year of operation as a risk factor [5]. Results There were 79 early deaths in the first thirty days, a hospital mortality of 24.8%. This includes deaths associated with further perioperative surgical intervention. Twenty-one of the 80 patients who had undergone previous valve operation died at reoperation (26.25%). There were 58 late deaths (8.2%), 25 of which were valve related (Table 3). Eight of the valve-related deaths were directly associated with reoperation, 5 of which occurred within a year of the initial operation and the others, and 2 years later. Seven other valve-related deaths occurred in the first year. Three of 0 non-valve-related deaths of unknown cause also occurred during the first year. Among the 6 patients experiencing valve failure (Table 4), 25 died of valve-related causes

3 564 The Annals of Thoracic Surgery Vol 33 No 6 June 982 Table 4. Causes of Valve Failure in Patients with Starr-Edwards Aortic Valve Replacements Cause No. of Valves No. of Deaths Thromboemboli 7 9 Paravalvular leak 6 3 nfection 2 8 Hemolysis 2 4 Cloth wear 2 0 Encapsulation 0 Obstruction Total 6 25 (4%) (see Table 3) and 4 have died subsequently, the remaining 32 surviving with or in spite of their problems. The most common causes of valve failure were embolic events (27.9%) and paravalvular leaks (26.2%). Valves used to replace failed valves were Starr-Edwards (4), xenografts (3), homograft (l), autograft (l), fascia lata (l), and Bjork-Shiley (). One was unknown. Actuarial analysis shows an overall probability of survival of 37.% at 4 years (Fig ) and Fig. Actuarial curves of probabilities of patient survival and event-free survival after isolated aortic valve replacement with Starr-Edwards valves (964 to 980). Entering numbers of patients for each year are shown along the abscissa. event-free survival for the same period of 8.8%. The overall probability of survival for the Series was 4.0% at 4 years compared with 34.3% for the Series 000 and 36.3% for the Series , both at 3 years (Fig 2). While these differences were not significant, the results at 0 years show the Series 000 and to be significantly better (57.2% and 52.8% survival) than the Series (36.3% survival) (p < 0.0). The average ages for the patients receiving the three variations of Starr-Edwards valves were not significantly different. The average ages were as follows: Series 000, 47.0 f 0.2 years; Series , 50.7 f 4.3 years; and Series , 53.5 f 9.6 years. The probability of freedom from valverelated death for each type of valve at 3 or 4 years is shown in Table 5, together with similar statistics for reoperation to change the valve (Fig 3), thromboembolism, infection, paravalvular leak, hemolysis (Fig 4), ball variance, and cloth wear. The actual percentages at 3 or 4 years are for each individual complication, and it is possible for any one valve to be represented by more than one complication-free value, e.g., paravalvular leak and ball variance. Although the Series has the worst 3-year record for valve-related deaths and for reoperations to change the valve (see Fig 3), that valve was not associated with any infection or paravalvular leaks. However, the Series i Y e a r s % Patient 8.8 _+ 5.2% Survival Event - Free Survival

4 565 Wain, Drury, and Ross: Aortic Valve Replacement with Star-Edwards Valves A c.= 60._ n m n 0 a 40 c l d Y e a r s % Model lo00 Fig 2. Actuarial curves of probabilities of patient survival for individual models of Starr-Edwards aortic valves used for isolated aortic valve replacement (964 to 980). Entering numbers of patients for each year are shown along the abscissa particularly was associated with hemolysis and hemolytic anemia, with only a 20.8% probability of freedom from hemolysis over 2 years. Along with this was a probability of freedom from cloth wear of only 53.3% over 2 years, a significantly worse statistic than the freedom from ball variance on the Silastic ball Series (96.0%) or the early 000 Series (75.0%). Paravalvular leak and ball variance were the worst problems with the Series 000 valve (25% at 3 years). Thromboembolism was the most serious problem with the Series (46.9% at 4 years). The influence of the year of operation was assessed over two 5-year periods, 966 to 97 and 975 to 980 for the Series valve. The earlier group of 9 patients had an average age of 50. f 2.6 years and the later group of 05 patients, 5.4 k.3 years. There was no significant difference between these two groups. The probabilities of patient survival, embolus-free survival, and event-free survival are expressed in Table 6. t is clear that the patients undergoing operation more recently have a better survival and a lower morbidity, but the Series 200/260 valves used more recently have been associated with a significantly higher rate of thromboembolic events (5.4%) than earlier implantations (7.%) (p < 0.00). Table 5. Actuarial Probabilities of Freedom from Principal Value-Related Failures for lndiuidual Models of Starr-Edwards Values Used for solated Aortic Value Replacement Series Series Series Cause of Failure (Yo) (Yo) ( O h Valve-related deaths Reoperation (valve change) Thromboembolism nfection Paravalvular leak Hemolysis Ball variance Cloth wear

5 566 The Annals of Thoracic Surgery Vol 33 No 6 June y-7 *-T(,\ ~ L ep Am ~5.6% Model % Model 080 \ : 8.3% Model ~ Fig 3. Actuarial curves of probabilities of freedom from valve change for individual models of Starr-Edwards aortic valves used for isolated aortic valve replacement (964 to 2980). Entering numbers of valves for each year are shown along the abscissa. - 00% Model loo ~5.4% \ Model 200 / 260 '\ \ o : 8.% Model 2300 ) m w n m ~ ' t e a r s Fig 4. Actuarial curves of probabilities of freedom from hemolusis for individual models of StarriEdwards valves used f;)r isolated aortic valve replacement (964 to 980). Entering numbers of valves for each year are shown along the abscissa.

6 567 Wain, Drury, and Ross: Aortic Valve Replacement with Stan-Edwards Valves Table 6. Comparison of Age, Postoperative Survival, Late Morbidity, and Late Embolism for Series Starr-Edwards Valves during Two Time-Frames P Variable 966 to 97 Value 975 to 980 Average age at operation (yr) 50. NS year survival (YO) 68.9 < year freedom from all complications (%) 52.9 < year freedom from embolic events ("/o) 92.9 < NS = not significant. Comment The Starr-Edwards caged-ball prosthesis has been used at the National Heart Hospital as the valve of second choice. That is, it has been used for high-risk patients for whom the possible disadvantages of long-term anticoagulation are more than offset by the advantages of a short cardiopulmonary bypass time. The average age of the patients in this high-risk group was 50.3 k 3.4 years, which is greater than the average age of 530 patients in an equivalent series who received homograft aortic valve replacements (39. k 4.6 years) over the same time period [l]. n fact, the 200 patients with diagnoses of aortic stenosis or aortic regurgitation, for example, had an even higher average age of 52.5 f 2.2 years, and the remaining 9 valves were required for a variety of complicating diagnoses (see Table 2). More importantly, the effect of the high-risk nature of this group of patients undergoing aortic valve replacement is reflected in the overall hospital mortality of 24.8%. A further effect of the high-risk nature of this group of patients is seen in the 5 deaths within a year of operation, including 3 of the 24 deaths for which there was no recorded information other than the date of death. The current model of the Starr-Edwards caged-ball prosthesis with a Silastic ball, Series , constituted 84.3% of the valves in this study, and this major group, in place for to 4 years, has shown the greatest absolute number of valve problems. The small numbers of valves with the early Silastic ball (Series 000, 24 valves) have had particular problems with paravalvular leak and ball variance (see Table 5). This is in broad agreement with a report from Edwards Laboratory on ball variance in the Series 000 valves.* Although there was only valve-related death (90.9% free from valve-related deaths), only 62.0% of patients had a possibility of freedom from any event over 3 years. However, only 4 of the Series 000 valves have been followed-up for more than 8 years and none beyond 3 years, even though the first valve in this series was implanted in January, 964. n contrast, there have been no paravalvular leaks and no infections associated with the few metal-ball valves, Series 2300/2320 (26 valves) (see Table 5). However, this valve, with clothcovered struts, has been associated with hemolysis and with cloth wear as a special problem. Only 53.3% of the valves have a probability of freedom from cloth wear over 3 years, and the probability of freedom from hemolysis is even lower, 20.8% at 3 years. Again, only 5 patients with a Series valve have been followed for more than 8 years and none beyond 3 years. This incidence of hemolysis (79.2% over 3 years) is similar to that reported in 976 by Santinga and colleagues [6], and in 979 by Falk [7, Anderson [B], and their colleagues. The Series valve has also been associated with the highest probability of valve-related death over 3 years (32.%). n spite of the absence of paravalvular leaks and infections, the high incidences of hemolysis, cloth wear, and valve-related death all indicate that the overall performance of the Series 2300/2320 valve was markedly inferior to that of the Series valve. Thromboembolic phenomena are the principal complications associated with the mechani- "Edwards Laboratories: Personal communication, 98.

7 568 The Annals of Thoracic Surgery Vol 33 No 6 June 982 cal cardiac prostheses, such as the Starr- Edwards caged-ball valves [3]. There were, in fact, only seventeen recorded embolic events (see Table 4), an actuarial probability of 8.7% over 4 years, or.% per annum. These thromboembolic episodes were fatal in 9 patients and had major, permanent sequelae in a further 8 patients. Macmanus and colleagues [5] analyzed late deaths and embolic events with reference to the year of operation as an additional risk factor. The Series valves in our presentation are comparable with those results, but differ from them. Specifically, the late mortality, reported by Macmanus and associates to be not significantly different between the pre-973 and post-973 groups (9% and 23%, respectively), was in fact significantly greater in the pre-973 group reported here (3.% versus.6%; p < 0.0). The probability of freedom from embolic episodes was significantly greater for our pre-973 group (see Table 6) (92.9% versus 84.6%; p < 0.00), in marked contrast to the results of Macmanus and colleagues (77% versus 87%; p < 0.0) in whose study the more recent implantations had a greater freedom from embolic episodes. However, a multiple decrement analysis for all postoperative deaths and all events for the two time-frames in Table 6 showed that the more recent implantations had a significantly greater freedom from all complications over the first 5 years (75.6% versus 52.9%; p < 0.00). Mechanical prostheses have a predicted function period well in excess of 20 years." Nevertheless, apart from further operations to resuture paravalvular leaks (2) or to replace defective balls (3) (see Table 4), reoperation to replace a defective valve has been necessary on twelve occasions over 4 years in this group of patients, an actuarial probability of 37.6% (see Table 5). An examination of these different problems as events in a multiple decrement analysis has shown that only 8.8% of patients may be expected to survive without any problem for 4 years (see Fig ). At the end of the first decade, *Edwards Laboratories: Personal communication, 98. the probability of event-free survival is 23% (see Fig l), and these figures may be compared with reports of event-free survival for biological valves of potentially lower durability [l, 9. Several studies published during 977 through 980 and concerning long-term follow-up after valve replacements referred to valves implanted between 965 and 969, implying more than a decade of follow-up information. Careful examination of these studies led us to select only ten in which adequate information is provided for a follow-up survey of 0 years or more [l, 5, 9-6]. This list is not exhaustive but illustrates the paucity of genuine long-term follow-up information. The Starr-Edwards Series 000 ball valve has the longest history of implantation [4], but the results in our paper show a longer follow-up period for the more recent Series 260 valve. Nevertheless, a patient survival of 5% at 0 years (see Fig ) compared favorably with the results from Portland with Starr-Edwards valves [5, 4, 5, and results for homograft valves [l] and from Auckland [6], despite the fact that the patients in this series form a high-risk group with an average age of 50 years. Again by analogy with other long-term results of valves free from any complication, the valves in this series are similar to homografts [l, 6 and other Starr valves [3, 4, and it is clear that although structural failure, such as poppet fracture, is now relatively uncommon, the probability of a complication-free decade is no more likely with a structurally sound mechanical valve than with a biological valve such as a homograft. n this comparison of complication-free analyses, the different populations receiving the valve are of less significance than for the overall patient survival.[4]. The use of porcine xenograft valves in this hospital during the past 4 years has reduced the number of Starr-Edwards valves used, since the xenografts may be inserted as conveniently as the Starr-Edwards valves and have a much lower incidence of thromboembolic phenomena [7, 8. The policy is still to regard mechanical prostheses as a second choice for aortic valve replacement, but the Starr-Edwards 260 valve continues to be the valve of second choice.

8 569 Wain, Drury, and Ross: Aortic Valve Replacement with Starr-Edwards Valves References 0. Shigenobu M, Mendez MA, Zubiate P, Kay JH:. Ross DN, Martelli V, Wain WH: Allograft and Thirteen years experience with the Kay-Shiley autograft valves used for aortic valve replace- disc valve for tricuspid replacement in Ebstein s ment. n onescu M (ed): Biological Tissue anomaly. Ann Thorac Surg 29: , 980 Valves. London, Butterworth, 979. Bjork VO, Henze A: Ten years experience with 2. Gallotti R, Ross DN: The Marfan syndrome: sur- the Bjork-Shiley tilting disc valve. J Thorac Cargical technique and follow-up in 50 patients. diovasc Surg 78:33-342, 979 Ann Thorac Surg 29: , Smith JM, Dernbach TA, Cooley DA: Treatment 3. Barnhorst DA, Oxman HA, Connolly DC, et al. of aortic insufficiency with a descending aortic solated replacement of the aortic valve with the valvular homograft: 9 year follow-up and his- Starr-Edwards prosthesis. J Thorac Cardiovasc torical perspective. Cardiovasc Dis, Bull Tex Surg 70:3-8, 975 Heart nst 7:206-23, Bodnar E, Haberman S, Wain WH: Comparative 3. Lefrak EA, Starr A: Cardiac Valve Prostheses. method for actuarial analysis of cardiac valve re- New York, Appleton-Century-Crofts, 979, p 86 placements. Br Heart J 42:54-552, Teply JF, Grunkemeier GL, Sutherland HD, et al: 5. Macmanus Q, Grunkemeier GL, Lambert LE, et The ultimate prognosis after valve replacement: al: Year of operation as a risk factor in the late an assessment at twenty years. Ann Thorac Surg results of valve replacement. J Thorac Cardiovasc 32~-9, 98 Surg 80:834-84, Santinga JT, Batsakis JT, Flora JD, Kirsh MM: Hemolysis in the aortic prosthetic valve. Chest 69~56-6, Falk RH, MacKinnon J, Wainscoat J, et al: ntramuscular haemolysis after valve replacement: comparative study between Starr-Edwards (ball valve) and Bjork-Shiley (disc valve) prostheses. Thorax 34: , Andersen JD, Lyngborg K, Wennervole A, et al: 9. Hyposideraemia and haemolysis in patients with Lillehei-Kaster or Starr-Edwards heart valve prostheses. Scand J Thorac Cardiovasc Surg 3~27-275, 979 Wain WH, Greco R, ngegneri A, et al: 5 years experience with 65 homograft and autograft aortic valve replacements. nt J Artif Organs 3~69-72, Starr A, Grunkemeier GL, Lambert LE, et al: Aortic valve replacement: a ten-year follow-up of nonclothcovered vs. clothcovered caged-ball prosthesis. Circulation 56:Suppl 2: , Barratt-Boyes BG: Cardiothoracic surgery in the Antipodes. J Thorac Cardiovasc Surg 78: , Oyer PE, Stinson EB, Griepp RB, Shumway NE: Valve replacement with the Starr-Edwards and Hancock prostheses: comparative analysis of late morbidity and mortality. Ann Surg 86:30-309, Cohn LH: Bioprosthetic cardiac valvesanticoagulation or not? n Sebening F, Klovekorn WP, Meisner E, Struck E (eds): Bioprosthetic Cardiac Valves. Munich, Deutsches Herzzentrum, 979, p 07 Notice from the American Board of Thoracic Surgery The American Board of Thoracic Surgery now requires that candidates pass both the written and oral portions of the certifying examination. n 980 and thereafter, a written examination will be given prior to the oral examination. t will be necessary to pass the written examination before the oral examination can be taken. The closing date for registration for 983 is August, 982. The exact time and places of these examinations will be announced later. A candidate applying for admission to the 983 certifying examination must fulfill all the requirements for the Board in force at the time the application is received. Please address all communications to the American Board of Thoracic Surgery, 4640 E Seven Mile Road, Detroit, M