- Email: [email protected]
A prospective controlled trial of st. jude versus starr edwards aortic and mitral valve prostheses
CARDIOVASCULAR
A Prospective Controlled Trial of St. Jude Versus Starr Edwards Aortic and Mitral Valve Prostheses Andrew J. Murday, FRCS, Andreas Hochstitzky, MRCS, Judith Mansfield, RSN, Julie Miles, RSN, Beverley Taylor, RSN, Eileen Whitley, RSN, and Tom Treasure, FRCS Department of Cardiothoracic Surgery, St. George’s Hospital, London, United Kingdom
Background. There is a paucity of controlled trials comparing the merits of different heart valve prostheses. In this prospective randomized trial we compared Starr Edwards and St. Jude prostheses in the aortic and mitral positions. Methods. Two hundred sixty-seven patients and 122 patients undergoing aortic and mitral valve replacement, respectively, were allocated by minimization to receive either St. Jude or Starr Edwards prostheses. Patients (2 patients were lost to follow-up) were followed up in a special clinic. Event definition, recording, and reporting were in accordance with published guidelines. Results. There were no demographic differences between patients receiving the two different valve models. With the exception of infective endocarditis, we found no differences in the rates of death or complication between patients receiving a Starr Edwards prosthesis or a St. Jude
prosthesis in either position. Neither were there any differences between the two valve models in either position, in terms of symptomatic relief 5 years after surgery. Conclusions. We found no differences in rates of complication or of symptomatic improvement between the Starr Edwards and St. Jude valve prostheses in either aortic or mitral position. Left ventricular function had such a marked effect on long-term survival that it overwhelmed any differences that might exist between different prosthetic designs. This confirms that historical comparisons are of limited value in deciding the respective merits of heart valve prostheses. The most reliable method of assessing surgical procedures is through prospective controlled trials. (Ann Thorac Surg 2003;76:66 –74) © 2003 by The Society of Thoracic Surgeons
W
The fate of the Bjork Shiley CC valve no doubt hastened a decline in the use of single-disc valves, and the bileaflet valves have taken over their market share. The low profile of the bileaflet valves is appealing to the implanting surgeon, but hemodynamic differences between well-established bileaflet valves are minimal [10]. Furthermore, there are greater similarities than differences in the behavior of St. Jude and Starr Edwards valves in terms of Doppler velocity and pressure recovery [11]. Evidence for the clinical performance of valves is based almost completely on retrospective case follow-up. For the St. Jude valve, which has been available for more than 20 years, there are large series with follow-up of 10 or more years [12–16]. However, the major determinants of survival in any series of patients with valve disease are left ventricular function and age at the time of implantation [17, 18] plus a number of other patient-related factors such as heart rhythm, pulmonary vascular resistance, and coexisting disease. Differences in thromboembolic events are highly dependent on time frame, probably because of a change in patient-related risk factors rather than change in valve design or anticoagulant management [19]. What randomized trial evidence there is on thromboembolic rates does not favor the bileaflet valve as opposed to a single-disc valve in a comparison of Edwards-Duromedics, Medtronic Hall, and Bjork-Shiley valves [20]. It is argu-
hat guides valve preference? In the late 1980s disappointment at the evident inevitability of tissue failure in frame-mounted porcine xenograft valves [1– 4] and bovine pericardial valves [5, 6] caused many surgeons to advise their patients to accept life-long anticoagulation rather than face the prospect of a second valve replacement. However, when the tissue versus mechanical debate was subjected to prospective randomized trials, survival and the overall incidence of valverelated complications was largely similar at 11 [7] and 12 years [8]. Nevertheless there was a clear choice between the need for reoperation for tissue failure and an increased risk of bleeding in those who had mechanical valves [7, 8]. If a mechanical valve is preferred to a tissue valve, how do the patient and surgeon choose which type of valve to implant? In the United Kingdom in 1986, of the 2,791 mechanical valves implanted, 55.2%, 21.5%, and 23.3% were single leaflet, bileaflet, and ball and cage, respectively. By the turn of the millennium the proportions had changed dramatically, so that in 2000 of the 4,049 implants, 91.8% were bileaflet, 6.5% single leaflet, and 1.7% ball and cage [9]. On what evidence was this dramatically changed use of mechanical valves based? Accepted for publication Jan 7, 2003. Address reprint requests to Dr Murday, Scottish Cardiopulmonary Transplant Unit, Glasgow Royal Infirmary, Alexandra Parade, Glasgow G31 2ER United Kingdom; e-mail: [email protected].
© 2003 by The Society of Thoracic Surgeons Published by Elsevier Inc
0003-4975/03/$30.00 PII S0003-4975(03)00118-8
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
able that the caged ball, with no hinge points and complete rinsing of the occluder, might be the best solution. In the absence of trial data comparing the performance of mechanical valves we sought to compare two mechanical valves, the St. Jude and the Starr Edwards valves. The St. Jude was chosen because it was seen to be “state of the art” whereas the Starr Edwards had become the robust workhorse valve, which surgeons turned to in difficult situations for its reliability and user friendliness. Neither of these propositions seemed to us to be evidence-based.
Anticoagulation
Patients and Methods
Follow-Up
Patients All patients undergoing aortic (AVR) or mitral (MVR) valve replacement between December 1991 and June 1997 were considered for entry into the trial. In general the policy in the unit is to consider patients 70 years of age or younger for prosthetic valve replacement, although there were no specific clinical features that were considered contraindicative. The trial was approved by the Local Research Ethics Committee of Wansdsworth Health Authority in October 1991. All patients for whom their surgeons was happy to randomize to either a Starr Edwards or a St. Jude valve prosthesis were put forward for consideration. There were no specific exclusion criteria for the trial. Patients with previous valve operations, endocarditis, and concomitant coronary artery disease were included. Patients were counseled about the trial and if they gave informed consent were considered for randomization. The number of patients undergoing concomitant AVR and MVR (27 in total) was insufficient to reach any worthwhile conclusions, and those patients are excluded from this report.
Randomization Patients were randomized during the operative procedure after valve excision at the point when the operating surgeon was satisfied that he or she would be prepared to put in either of the two prostheses. Allocation of valve was by minimization using a computer program [21]. Minimization has the advantage that differences in important patient variables that might otherwise occur by chance can be avoided. Factors used for minimization were as follows: age, 65 years or less or greater than 65 years; sex; intention to perform coronary artery bypass grafts; left ventricular function, good, impaired, or poor; and valve site, mitral or aortic.
Operative Technique The operations were performed according to the individual surgeons’ usual practice. Thus there were differences in surgical technique in terms of myocardial preservation and valve suture method. In particular, two suture techniques were used: one a continuous suture using 2-0 Prolene (Ethicon, Somerville, NJ), the other an interrupted method using pledgeted 2-0 Ethibond (Ethicon) sutures.
67
All patients received oral anticoagulation on the first operative day, or as soon as possible thereafter. It was recommended that the international normalized ratio should be maintained at between 3 and 4 for all patients, no matter which valve model, in whichever position, was implanted. Regular control of anticoagulation was left in the care of each patient’s general practitioner or local anticoagulation clinic. However, the current international normalized ratio was documented at each follow-up clinic visit and at the time of any thromboembolic or hemorrhagic event.
Patients were followed up annually in a specially designated outpatient clinic. At each attendance a clinical history was taken, from which New York Heart Association class was estimated, and a physical examination was performed by a doctor not blinded to the valve type. In addition an electrocardiogram and plain chest roentgenogram were undertaken. For a few patients for whom logistical problems made clinic attendance impossible, equivalent follow-up data were obtained from their local cardiologist. Definition of complications was in accordance with the guidelines of The Society of Thoracic Surgeons [22]. In addition each patient was registered at the time of operation with the Office of Population Census and Statistics (superseded in 1998 by the National Office of Statistics). This government organization provides a death certificate whenever any registered patient dies. When appropriate, information was obtained from postmortem examination. Two patients were lost to follow-up, of whom 1 emigrated and 1 remains unable to be contacted. Of those patients lost to follow-up, 1 underwent AVR with a St. Jude prosthesis and 1 underwent MVR with a Starr Edwards. These 2 patients have been included in the data analysis, and censored at their latest follow-up date.
Data Handling and Statistics All data were handled with care to maintain patient confidentiality. Records were maintained in both computer and paper formats. The closing point for any 1 patient was the time of their last visit to the annual follow-up clinic. Statistical analysis was carried out as appropriate: Kaplan-Meier survival analysis was used to examine the incidence of death and other relevant postoperative events. Kaplan-Meier curves were compared using the log-rank test. Continuous variables were compared using Student’s t test or Mann-Whitney U test as appropriate, and categorical variables were compared using a 2 test.
Results Patient Characteristics The patient characteristics are given in Table 1. Left ventricular function was estimated from left ventricular angiography, wherein ejection fraction less than 30% was
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74
CARDIOVASCULAR
68
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Ann Thorac Surg 2003;76:66 –74
Table 1. Demographic Data of Patients Aortic Valve Replacement Variable Number of patients Age (y)a Sex (male/female) BSAa (m2) NYHA I-II/III-IV Active endocarditis Sinus rhythm Predominant lesion stenosis/incompetence/ combination Previous surgery CABG LV function good/impaired/poor a
Mitral Valve Replacement
Starr Edwards
St. Jude
Starr Edwards
St. Jude
134 59.8 (9.7) 109/25 1.9 (0.2) 41/93 7 123 65/48/21
133 60.7 (8.4) 100/33 1.9 (0.2) 42/91 6 120 85/29/19
61 62.9 (9.1) 23/38 1.8 (0.2) 5/56 4 23 10/41/10
61 63.0 (8.8) 29/33 1.8 (0.2) 5/56 5 28 8/44/10
23 37 74/40/20
15 33 82/32/19
18 17 39/20/2
15 15 35/23/4
Values are mean (⫾ 1 standard deviation).
BSA ⫽ body surface area;
CABG ⫽ coronary artery bypass grafting;
categorized as poor, between 30% and 50% as impaired, and more than 50% as good. There were no statistically significant differences between the patients randomized to the two valve models in either the AVR or MVR groups. During the same time frame at this institution, 179 additional patients underwent AVR and 84 patients underwent MVR who were not entered into the trial. The recruitment rates were therefore 60% and 59%, respectively. The reasons for patients not being entered into the trial were not recorded. We do know that there was a marked variation among surgeons with respect to the proportion of their patients who were entered. It is likely that those surgeons who were most cautious in this regard were most likely to exclude from the trial patients with smaller aortic root sizes.
LV ⫽ left ventricular;
NYHA ⫽ New York Heart Association.
The distribution of prosthetic sizes is shown in Figure 1. Valve sizes are not directly comparable for any two different models of prosthetic valve. This is made particularly obvious with the Starr Edwards valve, as the size is a label rather than any particular dimension measurement.
Survival The actuarial survival curve after AVR is shown in Figure 2A. The 30-day survival after AVR was 97% (1 standard error [SE], ⫾2%) for St. Jude recipients and 95% (1 SE, ⫾2%) for Starr Edwards recipients. The 1-year survival after AVR was 92% (1 SE, ⫾2%) for St. Jude recipients and 89% (1 SE, ⫾3%) for Starr Edwards recipients. After 8 years the respective
AORTIC VALVE REPLACEMENT.
Fig 1. Aortic (A) and mitral (B) valve implant sizes. (Hatched bars ⫽ Starr Edwards; solid bars ⫽ St. Jude.)
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
69
Fig 2. Actuarial survival after aortic (A) and mitral (B) valve replacements. (Dashed line ⫽ St. Jude; solid line ⫽ Starr Edwards.)
survivals were 71% (1 SE, ⫾4%) and 65% (1 SE, ⫾5%). These differences are not statistically significant (logrank test, p ⫽ 0.6). The actuarial survival curve after MVR is shown in Figure 2B. The respective 30-day survivals were 88% (1 SE, ⫾4%) for St. Jude recipients and 95% (1 SE, ⫾3%) for Starr Edwards recipients. The respective 1-year survivals were 85% (1 SE, ⫾5%) for St. Jude recipients and 84% (1 SE, ⫾5%) for Starr Edwards recipients. After 8 years the respective survivals were 44% (1 SE, ⫾8%) and 57% (1 SE, ⫾ 7%). These differences do not reach statistical significance (p ⫽ 0.5).
MITRAL VALVE REPLACEMENT.
survival 1, 5, and 8 years after AVR was 95%, 80%, and 77% in patients with good left ventricular function, 87%, 76%, and 64% in patients with impaired ventricular function, and 79%, 51%, and 39% in patients with poor left ventricular function (p ⬍ 0.0002; Fig 3).
Thromboembolism The actuarial freedom from first thromboembolic event is given in Figure 4. The freedom from first embolus 1, 5, and 8 years after AVR was 95%, 89%, and 88% for St. Jude and 96%, 90%, and 86% for Starr Edwards. In the mitral position the freedoms at 1 and 5 years after surgery were 84% and
Causes of Death The causes of death are shown in Table 2. Only 10% of all the deaths in the whole trial were valve-related, although 54% of all deaths were owing to non–valve-related cardiac causes. When survival after AVR is analyzed with patients stratified into three groups according to preoperative left ventricular function, it is clear that ventricular function has a profound effect on survival. Actuarial Table 2. Causes of Death Aortic Valve Cause Valve-related Cardiac Other Unexplained Total
Mitral Valve
St. Jude
Starr Edwards
St. Jude
Starr Edwards
3 18 11 1 33
4 18 15 3 40
3 17 6 0 26
2 13 8 1 24
Fig 3. Actuarial survival (Kaplan-Meier) after aortic valve replacement stratified by left ventricular (LV) function. (Solid and dashed lines ⫽ poor left ventricular function; solid line ⫽ impaired ventricular function; dashed lines ⫽ good left ventricular function.)
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74
CARDIOVASCULAR
70
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Ann Thorac Surg 2003;76:66 –74
Fig 4. Actuarial freedom from first embolism after aortic (A) and mitral (B) valve replacements. (Dashed line ⫽ St. Jude; solid line ⫽ Starr Edwards.)
73% for St. Jude and 97% and 68% for Starr Edwards. At 8 years there were insufficient numbers of MVR patients to draw any worthwhile conclusions. Neither in the aortic (p ⫽ 0.98) nor mitral (p ⫽ 0.59) position were the differences between the two prostheses statistically significant.
Postoperative Endocarditis There were 13 episodes of infective endocarditis in the follow-up period, the details of which are shown in Table 3. Excluding patients with preoperative infective endocarditis, the surprisingly large number of 8 patients who
received a Starr Edwards aortic valve prosthesis developed infective endocarditis compared with only 1 patient who received a St. Jude aortic valve.
Reoperation Table 4 shows the indication and the time since the original operation for the 14 patients who underwent reoperation after valve replacement. The single case of valve obstruction was the result of impingement of posterior leaflet tissue into the orifice of a St. Jude mitral valve.
Table 3. Details of Episodes of Infective Endocarditis
Valve Make Starr Edwards St. Jude St. Jude Starr Edwards Starr Edwards St. Jude Starr Edwards Starr Edwards Starr Edwards Starr Edwards Starr Edwards Starr Edwards Starr Edwards
Valve Position Aortic Aortic Mitral Aortic Aortic Mitral Aortic Aortic Aortic Aortic Aortic Aortic Aortic
Preoperative Endocarditis
yes
yes
Interval From Operation (mo) 1 1 2 3 3 3 7 13 13 14 22 32 49
Reoperation
Mortality
yes
yes yes yes yes yes yes yes
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Table 4. Details of Reoperations in Patients Undergoing Aortic Valve Replacement
Valve Make
Valve Site
Indication for Reoperation
Interval From First Operation (mo)
Starr Edwards St. Jude St. Jude Starr Edwards St. Jude St. Jude St. Jude Starr Edwards Starr Edwards Starr Edwards St. Jude St. Jude St. Jude St. Jude
Aortic Mitral Mitral Aortic Mitral Mitral Aortic Aortic Aortic Aortic Mitral Aortic Aortic Aortic
Endocarditis Endocarditis Paraprosthetic leak Endocarditis Paraprosthetic leak Valve obstruction Paraprosthetic leak Endocarditis Endocarditis Endocarditis Paraprosthetic leak Paraprosthetic leak Paraprosthetic leak Paraprosthetic leak
2 2 4 4 5 9 14 14 23 27 30 32 32 35
Anticoagulation The international normalized ratio results recorded nearest to the annual follow-up appointment are represented in Figure 5. The boxes represent the 25 to 75 percentile ranges, and the whiskers represent the range excluding outliers and extremes. There were no significant differences in international normalized ratio measurements recorded at annual follow-up among any of the four groups of patients. Not surprisingly, the incidence of hemorrhage was no different between the two valve models or the two valve positions. The freedom from hemorrhage at 5 years was 85% for St. Jude AVR, 82% for Starr Edwards AVR, 79% for St. Jude MVR, and 79% for Starr Edwards MVR.
71
Comment Many widely accepted surgical interventions have not been subjected to randomized prospective trials. There are many reasons for this. Such trials can be difficult to set up, often take a long time to reach any conclusion, may be expensive, and by the time the results appear may seem to be irrelevant. Despite the fact that large amounts of money are spent each year on mechanical heart valve implants there have been very few published randomized trials to inform decisions as to which prosthesis should be used. Instead a variety of less rigorous sources inform our decisions. These include durability testing on laboratory rigs and clinical cohort studies including, at best, comparisons with matched historical controls. It has been frequently stated that the Starr Edwards valve has a higher risk of thromboembolic complication than other more recently developed valve prostheses. Evidence for this is based on studies that compare series of patients who come from different eras. The problem with such analyses lies in the fact that so many other factors, which might well be expected to influence thromboembolic complication rates, change with time. One such factor might well be the nature of the patients themselves. During the past 30 years there has been a gradual change in the cause of valve disease, such that 20 years ago in the United Kingdom, the majority of patients undergoing heart valve replacement had rheumatic valve disease. Now the commonest surgical valve pathologic process precipitating surgical replacement in developed countries is degenerative. Similarly, it would seem reasonable to assume that anticoagulation control has improved with the passage of time. It might be supposed that such changes could influence the incidence of thromboembolic and hemorrhagic complications irrespective of the model of valve prosthesis implanted.
Symptomatic Status Symptomatic status was recorded at each outpatient visit. The average reduction in New York Heart Association grades from the preoperative status and to 5-year postoperative status was 1.4 for Starr Edwards AVR recipients, 1.5 for St. Jude AVR recipients, 1.8 for Starr Edwards MVR recipients, and 1.6 for St. Jude MVR recipients. The changes from preoperative New York Heart Association grade to 5-year postoperative New York Heart Association grade are shown in Figure 6.
Event-Free Survival Actuarial event-free survival curves for patients receiving AVR or MVR are shown in Figure 7, A and B, respectively. The event-free survivals at 1, 5, and 8 years after AVR were 84%, 56%, and 45% for St. Jude, and 81%, 54%, and 43% for Starr Edwards valves, respectively (p ⫽ 0.65). The event-free survivals at 1, 5, and 8 years after MVR were 64%, 37%, and 25% for St. Jude, and 79%, 35%, and 29% for Starr Edwards valves, respectively (p ⫽ 0.46).
Fig 5. Box and whisker plot of international normalized ratio (INR) at annual follow-up. (AVR ⫽ aortic valve replacement; MVR ⫽ mitral valve replacement.)
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74
CARDIOVASCULAR
72
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Ann Thorac Surg 2003;76:66 –74
Fig 6. Change in New York Heart Association status after aortic (A) and mitral (B) valve replacements. (postop ⫽ postoperative.)
One group of patients, those who underwent AVR with a Starr Edwards prosthesis, had a higher incidence of infective endocarditis than others. With the exception of this rather strange and seemingly inexplicable finding we have demonstrated that the Starr Edwards valve has an indistinguishable risk profile from the St. Jude valve in either the mitral or aortic position. It is possible that the trial allowed a systematic exclusion of patients with very small aortic roots. The timing of randomization, after visualization of the valve and only when the surgeon was happy to insert either valve model, allowed for such systematic exclusion but ensured equipoise among the groups within the trial. By 5 years after operation there was no difference in symptomatic status between patients with a St. Jude or a Starr Edwards valve in the aortic or mitral position. At this stage in this trial superior hemodynamics claimed for the bileaflet over the ball and cage
models has made no difference in the patients’ well-being. We believe that it is equally important that we have demonstrated that direct comparison between valve prostheses by means of randomized trials is not only possible, but that other indirect means of comparison are inadequate. Not surprisingly we have shown that factors other than the valve model itself, particularly left ventricular function, have a great effect on long-term survival after valve replacement. With so many relatively new valve prostheses, both mechanical and biologic, currently being introduced, we believe that it is essential that whenever appropriate, the proper way to examine a new device is through controlled trials. In an age when new surgical techniques are arriving with increased frequency, it seems pertinent to remind ourselves that each new treatment should be adequately tested against the existing modalities.
Fig 7. Event-free survival after aortic (A) and mitral (B) valve replacements. (Dashed line ⫽ St. Jude; solid line ⫽ Starr Edwards.)
We are grateful to both St. Jude Medical UK, Ltd, and Edwards Lifesciences, Ltd, for providing financial support for this study.
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
11.
References 1. Glower DD, White WD, Hatton AC, et al. Determinants of reoperation after 960 valve replacements with Carpentier Edwards prostheses. J Thorac Cardiovasc Surgery 1994;107: 381–93. 2. Jamieson WRE, Burr LH, Tyers GFO, Munro AI. CarpentierEdwards standard and supra-annular porcine bioprostheses: 10 year comparison of structural valve deterioration. J Heart Valve Dis 1994;3:59 –65. 3. Barratt-Boyes BG, Jaffe WM, Ko PH, Whitlock RMI. The zero pressure fixed Medtronic Intact porcine valve: an 8.5 year review. J Heart Valve Dis 1993;2:604 –11. 4. Sarris GE, Robbins RC, Miller C, et al. Randomized, prospective assessment of bioprosthetic valve durability: Hancock versus Carpentier-Edwards valves. Circulation 1993;88:55–64. 5. Moggio RA, Pooley RW, Sarabu MR, Christiana J, Ho AW, Reed GE. Experience with the Mitroflow aortic bioprosthesis. J Thorac Cardiovasc Surg 1994;108:215–20. 6. Grabenwoger M, Grimm M, Leukauf C, et al. Failure mode of a new pericardial valve prosthesis (Sorin Pericarbon). Eur J Cardiothorac Surg 1994;8:470 –7. 7. Hammermeister KE, Sethi G, Henderson W, Oprian C, Kim T, Rahimtoola S. A comparison of outcomes in men 11 years after heart-valve replacement with a mechanical valve or bioprosthesis. N Engl J Med 1993;328:1289 –96. 8. Bloomfield P, Wheatley DJ, Prescott RJ, Miller HC. Twelveyear comparison of a Bjork-Shiley mechanical heart valve with porcine bioprostheses. N Engl J Med 1991;324:573–9. 9. The United Kingdom Heart Valve Registry Report 2000. The Department of Cardiac Surgery, Hammersmith Hospital. UK Heart Valve Registry Copyright 2002, page 10. 10. Ruel H, Van Son JAM, Steinseifer U, et al. In vitro comparison of bileaflet aortic heart valve prostheses. St. Jude Medical,
12. 13. 14. 15. 16.
17. 18. 19.
20.
21. 22.
73
CarboMedics, modified Edwards-Duromedics, and SorinBicarbon valves. J Thorac Cardiovasc Surg 1993;1063:412–20. Baumgartner H, Khan S, DeRobertis M, Czer L, Maurer G. Effect of prosthetic aortic valve design on the Dopplercatheter gradient correlation: an in-vitro study of normal St. Jude, Medtronic Hall, Starr-Edwards and Hancock valves. J Am Coll Cardiol 1992;19:324 –32. Avron KV. St. Jude Medical prosthesis: another 10-year follow up report. Ann Thorac Surg 1993;56:403–4. Kratz JM, Crawford FA, Sade RM, Crumbley AJ, Stoud MR. St. Jude prosthesis for aortic and mitral valve replacement: a ten-year experience. Ann Thorac Surg 1993;56:462–8. Ibrahim M, O’Kane H, Cleland J, Gladstone D, Sarsam M, Patterson C. The St. Jude Medical prosthesis: a thirteen-year experience. J Thorac Cardiovasc Surg 1994;108:221–30. Nakano K, Koyanagi H, Hashimoto A, et al. Twelve years’ experience with the St. Jude Medical valve prosthesis. Ann Thorac Surg 1994;57:697–703. Fernandez J, Laub GW, Adkins MS, et al. Early and latephase events after valve replacement with the St. Jude Medical prosthesis in 1200 patients. J Thorac Cardiovasc Surg 1994;107:394 –407. Lytle BW, Cosgrove DM, Taylor PC, et al. Primary isolated aortic valve replacement. J Thorac Cardiovasc Surg 1989;97: 675–94. Treasure T. Which heart valves should we use? Lancet 1990;336:1115–7. Macmanus Q, Grunkemeier GL, Lambert LE, Teply JF, Harlan BJ, Starr A. Year of operation as a risk factor in the late results of valve replacement. J Thorac Cardiovasc Surg 1980;80:834 –41. Kuntze CEE, Ebels T, Eijgelaar A, Homan van der Heide JN. Rates of thromboembolism with three different mechanical heart valve prostheses: a randomised study. Lancet 1989;1: 514 –7. Treasure T, MacRae KD. Minimisation: the platinum standard for trials? BMJ 1998;80:7–8. Edmunds LH, Clark RE, Cohn LH, Grunkmeier GL, Miller DC, Weisel RD. Guidelines for reporting morbidity and mortality after cardiac valvular operations. J Thorac Cardiovasc Surg 1996;112:708 –11.
INVITED COMMENTARY Murday et al have documented the only prospective randomized trial comparing the St. Jude Medical and the Starr-Edwards aortic and mitral prostheses in the English literature. The investigators are congratulated for contributing to evidence based management. This study has been long overdue, for the Starr-Edwards prosthesis has been used for 40 years and the St. Jude Medical prosthesis for 25 years. The clinical performance of these two prostheses has been designated the “gold standard” in their respective eras. The authors, although, have not provided an hypothesis for their study, nor a power calculation to prove their hypothesis. It is assumed the study was conducted only to evaluate the comparative clinical performance of the two prostheses. The two prostheses have enjoyed extended durability with only sporadic reports of structural failure, particularly with the poppet of the Starr-Edwards prosthesis. There have been three models of the Starr-Edwards inclusive of the poppet ball material cloth-covered prosthesis, and the track prosthesis. The investigators are documenting similar performances between the two prostheses with regard to sur© 2003 by The Society of Thoracic Surgeons Published by Elsevier Inc
vival, and valve-related complications of thromboembolism and anti-thrombotic hemorrhage. The study has confirmed that the degrees of ventricular dysfunction, as well as age, are the predominant predictors of survival. These results do not mean that the two prostheses have similar clinical performance. The investigators have documented only overall thromboembolism with no differentiation for major, reversible ischemic neurological deficits (RIND) and minor events. There may be a differentiation between the two prostheses. The major issue with the study is related to the anticoagulation management. The target INR for the study was 3.0 — 4.0 for both aortic and mitral prostheses with both valve types. The anticoagulant level is reflected in five years freedom from hemorrhage of approximately 80% for both types and positions. The current recommended INR targets for the St Jude Medical prostheses are 2.0 — 3.0 for aortic prostheses and 2.5—3.5 for mitral prostheses. There has always been a general opinion that the StarrEdwards prosthesis in both positions requires higher INR levels to prevent thromboembolic events. The target INR may be necessary of the Starr-Edwards, but not for 0003-4975/03/$30.00 PII S0003-4975(03)00698-2
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74
A Prospective Controlled Trial of St. Jude Versus Starr Edwards Aortic and Mitral Valve Prostheses Andrew J. Murday, FRCS, Andreas Hochstitzky, MRCS, Judith Mansfield, RSN, Julie Miles, RSN, Beverley Taylor, RSN, Eileen Whitley, RSN, and Tom Treasure, FRCS Department of Cardiothoracic Surgery, St. George’s Hospital, London, United Kingdom
Background. There is a paucity of controlled trials comparing the merits of different heart valve prostheses. In this prospective randomized trial we compared Starr Edwards and St. Jude prostheses in the aortic and mitral positions. Methods. Two hundred sixty-seven patients and 122 patients undergoing aortic and mitral valve replacement, respectively, were allocated by minimization to receive either St. Jude or Starr Edwards prostheses. Patients (2 patients were lost to follow-up) were followed up in a special clinic. Event definition, recording, and reporting were in accordance with published guidelines. Results. There were no demographic differences between patients receiving the two different valve models. With the exception of infective endocarditis, we found no differences in the rates of death or complication between patients receiving a Starr Edwards prosthesis or a St. Jude
prosthesis in either position. Neither were there any differences between the two valve models in either position, in terms of symptomatic relief 5 years after surgery. Conclusions. We found no differences in rates of complication or of symptomatic improvement between the Starr Edwards and St. Jude valve prostheses in either aortic or mitral position. Left ventricular function had such a marked effect on long-term survival that it overwhelmed any differences that might exist between different prosthetic designs. This confirms that historical comparisons are of limited value in deciding the respective merits of heart valve prostheses. The most reliable method of assessing surgical procedures is through prospective controlled trials. (Ann Thorac Surg 2003;76:66 –74) © 2003 by The Society of Thoracic Surgeons
W
The fate of the Bjork Shiley CC valve no doubt hastened a decline in the use of single-disc valves, and the bileaflet valves have taken over their market share. The low profile of the bileaflet valves is appealing to the implanting surgeon, but hemodynamic differences between well-established bileaflet valves are minimal [10]. Furthermore, there are greater similarities than differences in the behavior of St. Jude and Starr Edwards valves in terms of Doppler velocity and pressure recovery [11]. Evidence for the clinical performance of valves is based almost completely on retrospective case follow-up. For the St. Jude valve, which has been available for more than 20 years, there are large series with follow-up of 10 or more years [12–16]. However, the major determinants of survival in any series of patients with valve disease are left ventricular function and age at the time of implantation [17, 18] plus a number of other patient-related factors such as heart rhythm, pulmonary vascular resistance, and coexisting disease. Differences in thromboembolic events are highly dependent on time frame, probably because of a change in patient-related risk factors rather than change in valve design or anticoagulant management [19]. What randomized trial evidence there is on thromboembolic rates does not favor the bileaflet valve as opposed to a single-disc valve in a comparison of Edwards-Duromedics, Medtronic Hall, and Bjork-Shiley valves [20]. It is argu-
hat guides valve preference? In the late 1980s disappointment at the evident inevitability of tissue failure in frame-mounted porcine xenograft valves [1– 4] and bovine pericardial valves [5, 6] caused many surgeons to advise their patients to accept life-long anticoagulation rather than face the prospect of a second valve replacement. However, when the tissue versus mechanical debate was subjected to prospective randomized trials, survival and the overall incidence of valverelated complications was largely similar at 11 [7] and 12 years [8]. Nevertheless there was a clear choice between the need for reoperation for tissue failure and an increased risk of bleeding in those who had mechanical valves [7, 8]. If a mechanical valve is preferred to a tissue valve, how do the patient and surgeon choose which type of valve to implant? In the United Kingdom in 1986, of the 2,791 mechanical valves implanted, 55.2%, 21.5%, and 23.3% were single leaflet, bileaflet, and ball and cage, respectively. By the turn of the millennium the proportions had changed dramatically, so that in 2000 of the 4,049 implants, 91.8% were bileaflet, 6.5% single leaflet, and 1.7% ball and cage [9]. On what evidence was this dramatically changed use of mechanical valves based? Accepted for publication Jan 7, 2003. Address reprint requests to Dr Murday, Scottish Cardiopulmonary Transplant Unit, Glasgow Royal Infirmary, Alexandra Parade, Glasgow G31 2ER United Kingdom; e-mail: [email protected].
© 2003 by The Society of Thoracic Surgeons Published by Elsevier Inc
0003-4975/03/$30.00 PII S0003-4975(03)00118-8
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
able that the caged ball, with no hinge points and complete rinsing of the occluder, might be the best solution. In the absence of trial data comparing the performance of mechanical valves we sought to compare two mechanical valves, the St. Jude and the Starr Edwards valves. The St. Jude was chosen because it was seen to be “state of the art” whereas the Starr Edwards had become the robust workhorse valve, which surgeons turned to in difficult situations for its reliability and user friendliness. Neither of these propositions seemed to us to be evidence-based.
Anticoagulation
Patients and Methods
Follow-Up
Patients All patients undergoing aortic (AVR) or mitral (MVR) valve replacement between December 1991 and June 1997 were considered for entry into the trial. In general the policy in the unit is to consider patients 70 years of age or younger for prosthetic valve replacement, although there were no specific clinical features that were considered contraindicative. The trial was approved by the Local Research Ethics Committee of Wansdsworth Health Authority in October 1991. All patients for whom their surgeons was happy to randomize to either a Starr Edwards or a St. Jude valve prosthesis were put forward for consideration. There were no specific exclusion criteria for the trial. Patients with previous valve operations, endocarditis, and concomitant coronary artery disease were included. Patients were counseled about the trial and if they gave informed consent were considered for randomization. The number of patients undergoing concomitant AVR and MVR (27 in total) was insufficient to reach any worthwhile conclusions, and those patients are excluded from this report.
Randomization Patients were randomized during the operative procedure after valve excision at the point when the operating surgeon was satisfied that he or she would be prepared to put in either of the two prostheses. Allocation of valve was by minimization using a computer program [21]. Minimization has the advantage that differences in important patient variables that might otherwise occur by chance can be avoided. Factors used for minimization were as follows: age, 65 years or less or greater than 65 years; sex; intention to perform coronary artery bypass grafts; left ventricular function, good, impaired, or poor; and valve site, mitral or aortic.
Operative Technique The operations were performed according to the individual surgeons’ usual practice. Thus there were differences in surgical technique in terms of myocardial preservation and valve suture method. In particular, two suture techniques were used: one a continuous suture using 2-0 Prolene (Ethicon, Somerville, NJ), the other an interrupted method using pledgeted 2-0 Ethibond (Ethicon) sutures.
67
All patients received oral anticoagulation on the first operative day, or as soon as possible thereafter. It was recommended that the international normalized ratio should be maintained at between 3 and 4 for all patients, no matter which valve model, in whichever position, was implanted. Regular control of anticoagulation was left in the care of each patient’s general practitioner or local anticoagulation clinic. However, the current international normalized ratio was documented at each follow-up clinic visit and at the time of any thromboembolic or hemorrhagic event.
Patients were followed up annually in a specially designated outpatient clinic. At each attendance a clinical history was taken, from which New York Heart Association class was estimated, and a physical examination was performed by a doctor not blinded to the valve type. In addition an electrocardiogram and plain chest roentgenogram were undertaken. For a few patients for whom logistical problems made clinic attendance impossible, equivalent follow-up data were obtained from their local cardiologist. Definition of complications was in accordance with the guidelines of The Society of Thoracic Surgeons [22]. In addition each patient was registered at the time of operation with the Office of Population Census and Statistics (superseded in 1998 by the National Office of Statistics). This government organization provides a death certificate whenever any registered patient dies. When appropriate, information was obtained from postmortem examination. Two patients were lost to follow-up, of whom 1 emigrated and 1 remains unable to be contacted. Of those patients lost to follow-up, 1 underwent AVR with a St. Jude prosthesis and 1 underwent MVR with a Starr Edwards. These 2 patients have been included in the data analysis, and censored at their latest follow-up date.
Data Handling and Statistics All data were handled with care to maintain patient confidentiality. Records were maintained in both computer and paper formats. The closing point for any 1 patient was the time of their last visit to the annual follow-up clinic. Statistical analysis was carried out as appropriate: Kaplan-Meier survival analysis was used to examine the incidence of death and other relevant postoperative events. Kaplan-Meier curves were compared using the log-rank test. Continuous variables were compared using Student’s t test or Mann-Whitney U test as appropriate, and categorical variables were compared using a 2 test.
Results Patient Characteristics The patient characteristics are given in Table 1. Left ventricular function was estimated from left ventricular angiography, wherein ejection fraction less than 30% was
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74
CARDIOVASCULAR
68
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Ann Thorac Surg 2003;76:66 –74
Table 1. Demographic Data of Patients Aortic Valve Replacement Variable Number of patients Age (y)a Sex (male/female) BSAa (m2) NYHA I-II/III-IV Active endocarditis Sinus rhythm Predominant lesion stenosis/incompetence/ combination Previous surgery CABG LV function good/impaired/poor a
Mitral Valve Replacement
Starr Edwards
St. Jude
Starr Edwards
St. Jude
134 59.8 (9.7) 109/25 1.9 (0.2) 41/93 7 123 65/48/21
133 60.7 (8.4) 100/33 1.9 (0.2) 42/91 6 120 85/29/19
61 62.9 (9.1) 23/38 1.8 (0.2) 5/56 4 23 10/41/10
61 63.0 (8.8) 29/33 1.8 (0.2) 5/56 5 28 8/44/10
23 37 74/40/20
15 33 82/32/19
18 17 39/20/2
15 15 35/23/4
Values are mean (⫾ 1 standard deviation).
BSA ⫽ body surface area;
CABG ⫽ coronary artery bypass grafting;
categorized as poor, between 30% and 50% as impaired, and more than 50% as good. There were no statistically significant differences between the patients randomized to the two valve models in either the AVR or MVR groups. During the same time frame at this institution, 179 additional patients underwent AVR and 84 patients underwent MVR who were not entered into the trial. The recruitment rates were therefore 60% and 59%, respectively. The reasons for patients not being entered into the trial were not recorded. We do know that there was a marked variation among surgeons with respect to the proportion of their patients who were entered. It is likely that those surgeons who were most cautious in this regard were most likely to exclude from the trial patients with smaller aortic root sizes.
LV ⫽ left ventricular;
NYHA ⫽ New York Heart Association.
The distribution of prosthetic sizes is shown in Figure 1. Valve sizes are not directly comparable for any two different models of prosthetic valve. This is made particularly obvious with the Starr Edwards valve, as the size is a label rather than any particular dimension measurement.
Survival The actuarial survival curve after AVR is shown in Figure 2A. The 30-day survival after AVR was 97% (1 standard error [SE], ⫾2%) for St. Jude recipients and 95% (1 SE, ⫾2%) for Starr Edwards recipients. The 1-year survival after AVR was 92% (1 SE, ⫾2%) for St. Jude recipients and 89% (1 SE, ⫾3%) for Starr Edwards recipients. After 8 years the respective
AORTIC VALVE REPLACEMENT.
Fig 1. Aortic (A) and mitral (B) valve implant sizes. (Hatched bars ⫽ Starr Edwards; solid bars ⫽ St. Jude.)
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
69
Fig 2. Actuarial survival after aortic (A) and mitral (B) valve replacements. (Dashed line ⫽ St. Jude; solid line ⫽ Starr Edwards.)
survivals were 71% (1 SE, ⫾4%) and 65% (1 SE, ⫾5%). These differences are not statistically significant (logrank test, p ⫽ 0.6). The actuarial survival curve after MVR is shown in Figure 2B. The respective 30-day survivals were 88% (1 SE, ⫾4%) for St. Jude recipients and 95% (1 SE, ⫾3%) for Starr Edwards recipients. The respective 1-year survivals were 85% (1 SE, ⫾5%) for St. Jude recipients and 84% (1 SE, ⫾5%) for Starr Edwards recipients. After 8 years the respective survivals were 44% (1 SE, ⫾8%) and 57% (1 SE, ⫾ 7%). These differences do not reach statistical significance (p ⫽ 0.5).
MITRAL VALVE REPLACEMENT.
survival 1, 5, and 8 years after AVR was 95%, 80%, and 77% in patients with good left ventricular function, 87%, 76%, and 64% in patients with impaired ventricular function, and 79%, 51%, and 39% in patients with poor left ventricular function (p ⬍ 0.0002; Fig 3).
Thromboembolism The actuarial freedom from first thromboembolic event is given in Figure 4. The freedom from first embolus 1, 5, and 8 years after AVR was 95%, 89%, and 88% for St. Jude and 96%, 90%, and 86% for Starr Edwards. In the mitral position the freedoms at 1 and 5 years after surgery were 84% and
Causes of Death The causes of death are shown in Table 2. Only 10% of all the deaths in the whole trial were valve-related, although 54% of all deaths were owing to non–valve-related cardiac causes. When survival after AVR is analyzed with patients stratified into three groups according to preoperative left ventricular function, it is clear that ventricular function has a profound effect on survival. Actuarial Table 2. Causes of Death Aortic Valve Cause Valve-related Cardiac Other Unexplained Total
Mitral Valve
St. Jude
Starr Edwards
St. Jude
Starr Edwards
3 18 11 1 33
4 18 15 3 40
3 17 6 0 26
2 13 8 1 24
Fig 3. Actuarial survival (Kaplan-Meier) after aortic valve replacement stratified by left ventricular (LV) function. (Solid and dashed lines ⫽ poor left ventricular function; solid line ⫽ impaired ventricular function; dashed lines ⫽ good left ventricular function.)
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74
CARDIOVASCULAR
70
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Ann Thorac Surg 2003;76:66 –74
Fig 4. Actuarial freedom from first embolism after aortic (A) and mitral (B) valve replacements. (Dashed line ⫽ St. Jude; solid line ⫽ Starr Edwards.)
73% for St. Jude and 97% and 68% for Starr Edwards. At 8 years there were insufficient numbers of MVR patients to draw any worthwhile conclusions. Neither in the aortic (p ⫽ 0.98) nor mitral (p ⫽ 0.59) position were the differences between the two prostheses statistically significant.
Postoperative Endocarditis There were 13 episodes of infective endocarditis in the follow-up period, the details of which are shown in Table 3. Excluding patients with preoperative infective endocarditis, the surprisingly large number of 8 patients who
received a Starr Edwards aortic valve prosthesis developed infective endocarditis compared with only 1 patient who received a St. Jude aortic valve.
Reoperation Table 4 shows the indication and the time since the original operation for the 14 patients who underwent reoperation after valve replacement. The single case of valve obstruction was the result of impingement of posterior leaflet tissue into the orifice of a St. Jude mitral valve.
Table 3. Details of Episodes of Infective Endocarditis
Valve Make Starr Edwards St. Jude St. Jude Starr Edwards Starr Edwards St. Jude Starr Edwards Starr Edwards Starr Edwards Starr Edwards Starr Edwards Starr Edwards Starr Edwards
Valve Position Aortic Aortic Mitral Aortic Aortic Mitral Aortic Aortic Aortic Aortic Aortic Aortic Aortic
Preoperative Endocarditis
yes
yes
Interval From Operation (mo) 1 1 2 3 3 3 7 13 13 14 22 32 49
Reoperation
Mortality
yes
yes yes yes yes yes yes yes
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Table 4. Details of Reoperations in Patients Undergoing Aortic Valve Replacement
Valve Make
Valve Site
Indication for Reoperation
Interval From First Operation (mo)
Starr Edwards St. Jude St. Jude Starr Edwards St. Jude St. Jude St. Jude Starr Edwards Starr Edwards Starr Edwards St. Jude St. Jude St. Jude St. Jude
Aortic Mitral Mitral Aortic Mitral Mitral Aortic Aortic Aortic Aortic Mitral Aortic Aortic Aortic
Endocarditis Endocarditis Paraprosthetic leak Endocarditis Paraprosthetic leak Valve obstruction Paraprosthetic leak Endocarditis Endocarditis Endocarditis Paraprosthetic leak Paraprosthetic leak Paraprosthetic leak Paraprosthetic leak
2 2 4 4 5 9 14 14 23 27 30 32 32 35
Anticoagulation The international normalized ratio results recorded nearest to the annual follow-up appointment are represented in Figure 5. The boxes represent the 25 to 75 percentile ranges, and the whiskers represent the range excluding outliers and extremes. There were no significant differences in international normalized ratio measurements recorded at annual follow-up among any of the four groups of patients. Not surprisingly, the incidence of hemorrhage was no different between the two valve models or the two valve positions. The freedom from hemorrhage at 5 years was 85% for St. Jude AVR, 82% for Starr Edwards AVR, 79% for St. Jude MVR, and 79% for Starr Edwards MVR.
71
Comment Many widely accepted surgical interventions have not been subjected to randomized prospective trials. There are many reasons for this. Such trials can be difficult to set up, often take a long time to reach any conclusion, may be expensive, and by the time the results appear may seem to be irrelevant. Despite the fact that large amounts of money are spent each year on mechanical heart valve implants there have been very few published randomized trials to inform decisions as to which prosthesis should be used. Instead a variety of less rigorous sources inform our decisions. These include durability testing on laboratory rigs and clinical cohort studies including, at best, comparisons with matched historical controls. It has been frequently stated that the Starr Edwards valve has a higher risk of thromboembolic complication than other more recently developed valve prostheses. Evidence for this is based on studies that compare series of patients who come from different eras. The problem with such analyses lies in the fact that so many other factors, which might well be expected to influence thromboembolic complication rates, change with time. One such factor might well be the nature of the patients themselves. During the past 30 years there has been a gradual change in the cause of valve disease, such that 20 years ago in the United Kingdom, the majority of patients undergoing heart valve replacement had rheumatic valve disease. Now the commonest surgical valve pathologic process precipitating surgical replacement in developed countries is degenerative. Similarly, it would seem reasonable to assume that anticoagulation control has improved with the passage of time. It might be supposed that such changes could influence the incidence of thromboembolic and hemorrhagic complications irrespective of the model of valve prosthesis implanted.
Symptomatic Status Symptomatic status was recorded at each outpatient visit. The average reduction in New York Heart Association grades from the preoperative status and to 5-year postoperative status was 1.4 for Starr Edwards AVR recipients, 1.5 for St. Jude AVR recipients, 1.8 for Starr Edwards MVR recipients, and 1.6 for St. Jude MVR recipients. The changes from preoperative New York Heart Association grade to 5-year postoperative New York Heart Association grade are shown in Figure 6.
Event-Free Survival Actuarial event-free survival curves for patients receiving AVR or MVR are shown in Figure 7, A and B, respectively. The event-free survivals at 1, 5, and 8 years after AVR were 84%, 56%, and 45% for St. Jude, and 81%, 54%, and 43% for Starr Edwards valves, respectively (p ⫽ 0.65). The event-free survivals at 1, 5, and 8 years after MVR were 64%, 37%, and 25% for St. Jude, and 79%, 35%, and 29% for Starr Edwards valves, respectively (p ⫽ 0.46).
Fig 5. Box and whisker plot of international normalized ratio (INR) at annual follow-up. (AVR ⫽ aortic valve replacement; MVR ⫽ mitral valve replacement.)
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74
CARDIOVASCULAR
72
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
Ann Thorac Surg 2003;76:66 –74
Fig 6. Change in New York Heart Association status after aortic (A) and mitral (B) valve replacements. (postop ⫽ postoperative.)
One group of patients, those who underwent AVR with a Starr Edwards prosthesis, had a higher incidence of infective endocarditis than others. With the exception of this rather strange and seemingly inexplicable finding we have demonstrated that the Starr Edwards valve has an indistinguishable risk profile from the St. Jude valve in either the mitral or aortic position. It is possible that the trial allowed a systematic exclusion of patients with very small aortic roots. The timing of randomization, after visualization of the valve and only when the surgeon was happy to insert either valve model, allowed for such systematic exclusion but ensured equipoise among the groups within the trial. By 5 years after operation there was no difference in symptomatic status between patients with a St. Jude or a Starr Edwards valve in the aortic or mitral position. At this stage in this trial superior hemodynamics claimed for the bileaflet over the ball and cage
models has made no difference in the patients’ well-being. We believe that it is equally important that we have demonstrated that direct comparison between valve prostheses by means of randomized trials is not only possible, but that other indirect means of comparison are inadequate. Not surprisingly we have shown that factors other than the valve model itself, particularly left ventricular function, have a great effect on long-term survival after valve replacement. With so many relatively new valve prostheses, both mechanical and biologic, currently being introduced, we believe that it is essential that whenever appropriate, the proper way to examine a new device is through controlled trials. In an age when new surgical techniques are arriving with increased frequency, it seems pertinent to remind ourselves that each new treatment should be adequately tested against the existing modalities.
Fig 7. Event-free survival after aortic (A) and mitral (B) valve replacements. (Dashed line ⫽ St. Jude; solid line ⫽ Starr Edwards.)
We are grateful to both St. Jude Medical UK, Ltd, and Edwards Lifesciences, Ltd, for providing financial support for this study.
MURDAY ET AL CONTROLLED TRIAL OF MITRAL VALVE PROSTHESES
11.
References 1. Glower DD, White WD, Hatton AC, et al. Determinants of reoperation after 960 valve replacements with Carpentier Edwards prostheses. J Thorac Cardiovasc Surgery 1994;107: 381–93. 2. Jamieson WRE, Burr LH, Tyers GFO, Munro AI. CarpentierEdwards standard and supra-annular porcine bioprostheses: 10 year comparison of structural valve deterioration. J Heart Valve Dis 1994;3:59 –65. 3. Barratt-Boyes BG, Jaffe WM, Ko PH, Whitlock RMI. The zero pressure fixed Medtronic Intact porcine valve: an 8.5 year review. J Heart Valve Dis 1993;2:604 –11. 4. Sarris GE, Robbins RC, Miller C, et al. Randomized, prospective assessment of bioprosthetic valve durability: Hancock versus Carpentier-Edwards valves. Circulation 1993;88:55–64. 5. Moggio RA, Pooley RW, Sarabu MR, Christiana J, Ho AW, Reed GE. Experience with the Mitroflow aortic bioprosthesis. J Thorac Cardiovasc Surg 1994;108:215–20. 6. Grabenwoger M, Grimm M, Leukauf C, et al. Failure mode of a new pericardial valve prosthesis (Sorin Pericarbon). Eur J Cardiothorac Surg 1994;8:470 –7. 7. Hammermeister KE, Sethi G, Henderson W, Oprian C, Kim T, Rahimtoola S. A comparison of outcomes in men 11 years after heart-valve replacement with a mechanical valve or bioprosthesis. N Engl J Med 1993;328:1289 –96. 8. Bloomfield P, Wheatley DJ, Prescott RJ, Miller HC. Twelveyear comparison of a Bjork-Shiley mechanical heart valve with porcine bioprostheses. N Engl J Med 1991;324:573–9. 9. The United Kingdom Heart Valve Registry Report 2000. The Department of Cardiac Surgery, Hammersmith Hospital. UK Heart Valve Registry Copyright 2002, page 10. 10. Ruel H, Van Son JAM, Steinseifer U, et al. In vitro comparison of bileaflet aortic heart valve prostheses. St. Jude Medical,
12. 13. 14. 15. 16.
17. 18. 19.
20.
21. 22.
73
CarboMedics, modified Edwards-Duromedics, and SorinBicarbon valves. J Thorac Cardiovasc Surg 1993;1063:412–20. Baumgartner H, Khan S, DeRobertis M, Czer L, Maurer G. Effect of prosthetic aortic valve design on the Dopplercatheter gradient correlation: an in-vitro study of normal St. Jude, Medtronic Hall, Starr-Edwards and Hancock valves. J Am Coll Cardiol 1992;19:324 –32. Avron KV. St. Jude Medical prosthesis: another 10-year follow up report. Ann Thorac Surg 1993;56:403–4. Kratz JM, Crawford FA, Sade RM, Crumbley AJ, Stoud MR. St. Jude prosthesis for aortic and mitral valve replacement: a ten-year experience. Ann Thorac Surg 1993;56:462–8. Ibrahim M, O’Kane H, Cleland J, Gladstone D, Sarsam M, Patterson C. The St. Jude Medical prosthesis: a thirteen-year experience. J Thorac Cardiovasc Surg 1994;108:221–30. Nakano K, Koyanagi H, Hashimoto A, et al. Twelve years’ experience with the St. Jude Medical valve prosthesis. Ann Thorac Surg 1994;57:697–703. Fernandez J, Laub GW, Adkins MS, et al. Early and latephase events after valve replacement with the St. Jude Medical prosthesis in 1200 patients. J Thorac Cardiovasc Surg 1994;107:394 –407. Lytle BW, Cosgrove DM, Taylor PC, et al. Primary isolated aortic valve replacement. J Thorac Cardiovasc Surg 1989;97: 675–94. Treasure T. Which heart valves should we use? Lancet 1990;336:1115–7. Macmanus Q, Grunkemeier GL, Lambert LE, Teply JF, Harlan BJ, Starr A. Year of operation as a risk factor in the late results of valve replacement. J Thorac Cardiovasc Surg 1980;80:834 –41. Kuntze CEE, Ebels T, Eijgelaar A, Homan van der Heide JN. Rates of thromboembolism with three different mechanical heart valve prostheses: a randomised study. Lancet 1989;1: 514 –7. Treasure T, MacRae KD. Minimisation: the platinum standard for trials? BMJ 1998;80:7–8. Edmunds LH, Clark RE, Cohn LH, Grunkmeier GL, Miller DC, Weisel RD. Guidelines for reporting morbidity and mortality after cardiac valvular operations. J Thorac Cardiovasc Surg 1996;112:708 –11.
INVITED COMMENTARY Murday et al have documented the only prospective randomized trial comparing the St. Jude Medical and the Starr-Edwards aortic and mitral prostheses in the English literature. The investigators are congratulated for contributing to evidence based management. This study has been long overdue, for the Starr-Edwards prosthesis has been used for 40 years and the St. Jude Medical prosthesis for 25 years. The clinical performance of these two prostheses has been designated the “gold standard” in their respective eras. The authors, although, have not provided an hypothesis for their study, nor a power calculation to prove their hypothesis. It is assumed the study was conducted only to evaluate the comparative clinical performance of the two prostheses. The two prostheses have enjoyed extended durability with only sporadic reports of structural failure, particularly with the poppet of the Starr-Edwards prosthesis. There have been three models of the Starr-Edwards inclusive of the poppet ball material cloth-covered prosthesis, and the track prosthesis. The investigators are documenting similar performances between the two prostheses with regard to sur© 2003 by The Society of Thoracic Surgeons Published by Elsevier Inc
vival, and valve-related complications of thromboembolism and anti-thrombotic hemorrhage. The study has confirmed that the degrees of ventricular dysfunction, as well as age, are the predominant predictors of survival. These results do not mean that the two prostheses have similar clinical performance. The investigators have documented only overall thromboembolism with no differentiation for major, reversible ischemic neurological deficits (RIND) and minor events. There may be a differentiation between the two prostheses. The major issue with the study is related to the anticoagulation management. The target INR for the study was 3.0 — 4.0 for both aortic and mitral prostheses with both valve types. The anticoagulant level is reflected in five years freedom from hemorrhage of approximately 80% for both types and positions. The current recommended INR targets for the St Jude Medical prostheses are 2.0 — 3.0 for aortic prostheses and 2.5—3.5 for mitral prostheses. There has always been a general opinion that the StarrEdwards prosthesis in both positions requires higher INR levels to prevent thromboembolic events. The target INR may be necessary of the Starr-Edwards, but not for 0003-4975/03/$30.00 PII S0003-4975(03)00698-2
CARDIOVASCULAR
Ann Thorac Surg 2003;76:66 –74