| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2003;75:1482-1489
© 2003 The Society of Thoracic Surgeons
a Department of Cardio-Thoracic Surgery, Rotterdam, The Netherlands
b Department of Public Health, Erasmus University Medical Center Rotterdam, Rotterdam, The Netherlands
c Section of Thoracic and Cardiovascular Surgery, Department of Surgery, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma, USA
Accepted for publication November 1, 2002.
* Address reprint requests to Dr Takkenberg, Department of Cardio-Thoracic Surgery, Bd162, Erasmus University Medical Center Rotterdam, PO Box 2040, 3000CA Rotterdam, The Netherlands
e-mail: takkenberg{at}thch.azr.nl
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and methodsResultsAcknowledgmentsReferences |
|---|
METHODS: Our center’s experience with cryopreserved allograft aortic root replacement in 165 adult patients was combined in a meta-analysis with reported and individual results from four other hospitals. Using this information, the microsimulation model predicted age- and gender-specific total and reoperation-free and event-free life expectancy.
RESULTS: The pooled results comprised 629 patients with a total follow-up of 1860 patient-years (range 0 to 12.8 years). Annual risks were 0.6% for thromboembolism, 0.05% for bleeding, 0.5% for endocarditis, and 0.5% for nonstructural valve failure. Structural allograft failure requiring reoperation occurred in 15 patients, and a patient age–specific Weibull function was constructed accordingly. Calculated total life expectancy varied from 27 years in a 25-year-old to 12 years in a 65-year-old male; corresponding actual lifetime risk of reoperation was 89% and 35%, respectively.
CONCLUSIONS: Cryopreserved aortic allografts have an age-related limited durability. This results in a considerable lifetime risk of reoperation, especially in young patients. The combination of meta-analysis and microsimulation provides an appropriate tool for estimating individualized long-term outcome after aortic valve replacement and can be useful both for patient counseling and prognostic research purposes.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and methodsResultsAcknowledgmentsReferences |
|---|
Since the introduction of the allograft as an aortic valve substitute, several changes have been made with regard to preservation and surgical techniques. Initially, fresh or antibiotically sterilized valves were implanted using the subcoronary implantation technique. Long-term results after aortic valve replacement (AVR) with fresh or antibiotically sterilized allografts that were mainly implanted using the subcoronary or freehand technique show that after 20 years, 35% of patients are free from redo valve replacement and only 18% are still free from tissue failure [1]. Slightly better results were obtained in a large series of 804 patients who had AVR with either antibiotically sterilized (n = 124) or cryopreserved (n = 680) allografts that were implanted using predominantly the subcoronary implantation technique. Freedom from structural deterioration was 45% and 80%, respectively, after 15 years [4]. Both studies concluded that younger patient age is associated with increased structural valve failure rates. Most centers no longer use fresh or antibiotically sterilized allografts and do not use the subcoronary implantation technique. It is hypothesized that the use of cryopreservation methods and the root replacement technique will result in improved durability of the valve substitute.
Unfortunately, most reported series on ARR with cryopreserved allografts are small and have limited follow-up, which does not allow insight into long-term failure of allograft roots beyond the first 10 years after operation. In addition, because of the small size of the studies, it is difficult to identify potential risk factors for allograft failure and to assess the effect they have on prognosis. The aim of this study was to estimate age-specific prognosis after ARR with a cryopreserved aortic allograft. We performed a meta-analysis of reported and primary data [5] to quantify a microsimulation model that estimates life expectancy (LE) and actual risks of events and reoperation in the individual patient.
|
Patients and methods |
|---|
|
TopAbstractIntroductionPatients and methodsResultsAcknowledgmentsReferences |
|---|
Literature search
We performed a literature search of the MEDLINE database using the PubMed search engine for the period between January 1995 and June 2000. This was done to obtain the most recent reports with the longest follow-up. Terms used for the search were MeSH terms and the text words "allograft," "root," "aortic valve," and "homograft." All titles and abstracts were screened for study design (reports of clinical experience with cryopreserved allograft ARR), completeness of follow-up (> 90%), surgical technique (root replacement with reimplantation of the coronary arteries), study size (n > 40, reflecting the experience at that particular center), and patient age (age 
16 years at time of operation). The references of selected papers were cross-checked for other potentially relevant studies.
Data extraction and analysis
The selected studies were reviewed, and patient characteristics and results of each study were tabulated in a spreadsheet. The authors of the papers were contacted for clarification and additional information, if necessary. Events and outcomes in all studies including our own were defined according to Edmund’s guidelines [6]. A combined estimate of outcome was obtained by means of weighted pooling [7], or in case of few events per study, direct pooling. Linearized annual occurrence rates were calculated for valvular thrombosis, thromboembolism, bleeding, endocarditis, and nonstructural valve failure. The incidence of structural valvular failure requiring replacement of the valve was described by a Weibull curve, which is a generalization of the exponential distribution that accommodates changing risk over time [8–10]. The Weibull model has been used previously to describe structural valve deterioration [9, 11]. The parameters of the Weibull model were estimated using the pooled structural valve failure life-table from the meta-analysis. The Oklahoma center [3] provided additional individual patient data on the relationship between patient age and structural valve failure in their dataset. These individual patient data were combined with the information on the relationship between patient age and structural valve failure from the Rotterdam dataset, and an age variable was estimated and added to the Weibull model, allowing patient age-specific calculations for structural valve failure [12, 13]. SAS 6.12 (SAS Inc., Cary, NC) was used to construct the Weibull model.
Microsimulation model
The basic assumption of the microsimulation model is that a disease follows a course in time that is characterized by a number of discrete health states and disease-related events. After AVR with a cryopreserved allograft root, the patient will either die as a result of the procedure or stay alive. If the patient stays alive, he or she remains at risk of developing valve-related events for the rest of his or her life. Eventually this patient will die of either valve-related or nonvalve-related causes. The AVR microsimulation model that was used for this study was constructed using Delphi IV software (Borland Corporation, Scotts Valley, CA).
The information on outcome after allograft ARR from the meta-analysis was entered into the microsimulation model. For a given patient, characterized by gender and age, 10,000 virtual life histories were calculated. Age at death from nonvalve-related causes was randomly drawn from a life table that was based on the Dutch general population (http://www.cbs.nl). However, there is a higher mortality rate in patients who have had AVR than in the general population, which cannot be explained solely by postoperative valve-related events. This higher mortality rate is caused by sudden, unexpected, unexplained death and cardiac death, related to valve disease, cardiomyopathy, and factors introduced by the type of valve substitute [14, 15]. We therefore multiplied the age- and gender-specific mortality hazard of the general population with an age- and gender-related hazard ratio for excess mortality [16]. Other assumptions we made were that bleeding risk increases with age (odds ratio 1.0345/year) [17] and operative mortality rate increases with age (2.6% at age 45 and increasing with odds ratio 1.022/year) and with each reoperation (odds ratio 1.7 with each reoperation) [18, 19]. The lethality of valve-related events was estimated from our previous work and from recent literature, resulting in mortality rates of 10% after thromboembolism, 7% after bleeding, and 25% after endocarditis [1, 20]. After structural or nonstructural valve failure, the allograft was assumed to be replaced by a mechanical prosthesis. If the patient survived the valve replacement with a mechanical prosthesis, the hazards of valve-related events were changed accordingly [18]. If valve replacement was necessary after endocarditis, another allograft was implanted.
For men at different ages (25, 35, 45, 55, and 65 years) LE, reoperation-free LE, actual lifetime reoperation risk, event-free LE, and actual lifetime event risk were calculated. The effect of valve-related events on mortality rate was assessed, and loss of LE was compared with healthy age-matched individuals.
To investigate validity, calculated survival rate was compared with reported survival rates in three of the studies in the meta-analysis (Rotterdam, Brisbane, and Nieuwegein [2, 21]). In addition, calculated survival rate was compared with reported survival rates after allograft aortic valve or root replacement in two large datasets from Australia and the United Kingdom. Those patients had aortic valve or root replacement with fresh, antibiotic stored, or cryopreserved allografts using several surgical techniques [22].
To investigate the effect of uncertainty in the variable estimates on life expectancy, one-way sensitivity analyses were performed as follows. For variables that were estimated using linearized annual occurrence rates (thromboembolism, bleeding, endocarditis, and nonstructural valve failure) 95% confidence limits were calculated [23]. The 95% confidence interval (CI) for the time to structural valve deterioration was calculated using the variance-covariance matrix of the fitted Weibull model. The estimates for operative mortality rate were ranged from half to double the baseline values. Next, the individual variable estimates were changed to the lower and upper confidence limits, while the other variables in the microsimulation model remained constant. Thus the effect of varying each variable on life expectancy and event-free life expectancy could be calculated.
|
Results |
|---|
|
TopAbstractIntroductionPatients and methodsResultsAcknowledgmentsReferences |
|---|
|
|
Figure 1
shows the age-specific Weibull function derived from the pooled data on structural valve failure requiring reoperation. The formula for freedom from structural valve failure was S(t) = e-(t/
)
, where S(t) indicates the probability of being free from structural valve failure at time t, and and ß indicate the scale and shape variables of the Weibull model. The value of depended on age ( = e2.234 + 0.0112*age), and the value of ß was 3.669 (p < 0.001). The age variable was estimated using combined information on the relationship between patient age and structural valve failure requiring reoperation from the Rotterdam and Oklahoma centers [3]. In the Oklahoma dataset, 5 patients required reoperation for structural valve failure at 4.7, 8.1, 8.5, 8.9, and 10.2 years after the initial operation. The mean age at the time of primary operation of these patients was 47 years (standard deviation, 6.8; range, 37 to 55 years). The estimated value of the age variable in the Weibull model (0.0112; p = 0.078) resulted in an increase in the median time to structural allograft failure from 11.1 years in a 25-year old man to 17.5 years in a 65-year-old man.
|
|
|
One-way sensitivity analyses showed that varying the individual variables had little effect on the total LE in all age groups. There is, however, a major effect of varying structural valve deterioration on event-free LE. These results are shown in Table 2. For example, by ranging the estimates for median time to structural valve deterioration for a 25-year-old patient from its lower to upper confidence limit (9.1 to 13.7 years) total LE hardly changed (26.9 to 27.1 years), whereas event-free life expectancy changed from 7.8 to 11.6 years.
|
Increased rates of allograft structural valvular deterioration are associated with younger patient age [1, 4, 25]. This could be explained by a stronger immune response against the allograft [26], greater strain on the allograft because of the more active lifestyle in young adults, or to nonviability of cryopreserved allografts resulting in the failure of repair processes [27]. Although allografts are overall more durable than stented porcine bioprostheses, the effect of patient age on structural valve failure after allograft ARR closely resembles the previously reported effect of patient age on structural valve failure of stented porcine bioprostheses [19]. This suggests that the dominant mechanisms underlying valve deterioration might be similar in these two valve types.
According to the microsimulation model, structural valvular deterioration is the most important cause of reoperation. In a 25-year-old male patient it results in a large lifetime reoperation risk of 89%, whereas in a 65-year-old male patient this risk is 35%. This finding illustrates the need for improvement of the available valve substitutes, as well as the need for an objective clinical decision support system that allows selection of the most appropriate aortic valve substitute for the individual patient. It is obvious from our findings that in young adults other aortic valve substitutes should be seriously considered. The estimates of age-dependent freedom from structural valve failure in the model are based on 15 cases from studies with a mean follow-up of less than 4 years. This is an important limitation of the model. It is necessary to obtain more information on the relationship between patient age and structural valve deterioration rates in order to estimate this relationship more accurately. Note that the 35% lifetime risk of reoperation in a 65-year-old patient is probably an overestimate of the true risk of reoperation. This is a result of the assumption in the microsimulation model that all patients with structural valve deterioration are reoperated on. In actuality, many older patients with structural valve deterioration will probably be less often reoperated on than younger allograft recipients.
Although valve-related events are common after ARR with a cryopreserved aortic allograft, they do not play a major role in reduction of total LE. Death is caused mainly by nonvalve-related events, and loss of LE compared with healthy age-matched individuals is determined mainly by excess mortality caused by cardiac disease secondary to the heart valve disease. The reduction of LE due to excess mortality is most pronounced in younger patients and decreases with age. This could be caused by a more aggressive form of aortic valve disease in younger patients, which requires replacement of the valve relatively early in life. Conversely, for older patients, a selection process takes place, and only the most vital patients with a relatively long LE are considered for AVR.
Comparison of survival rate, as calculated by the microsimulation model, showed good agreement with reported short-term and long-term outcomes after allograft aortic valve or root replacement [1, 2, 21, 22]. This finding indicates that survival rates calculated using the model are an accurate reflection of true LE after AVR. It is more difficult to compare the outcome of structural valve failure as calculated with the model with reported long-term durability of aortic allografts. Because surgical technique can affect outcome, we chose to include in the meta-analysis only studies of ARR, which is now the most commonly used surgical technique. An overview of different reports on allograft AVR using a variety of preservation and surgical techniques showed an actuarial freedom from allograft failure varying from less than 20% to more than 80% at 15 years postoperatively [28]. This wide variation reflects the numerous factors that affect the process of allograft structural failure. It is therefore difficult to make a straightforward comparison between the structural valve failure outcome of the microsimulation model and reported results with allografts that were implanted with different preservation and operative techniques. However, the Weibull model estimates of 15-year freedom from structural valve failure with cryopreserved aortic allografts are in good agreement with the reported 15-year freedom from structural valve failure in a subset of patients who had ARR with fresh or antibiotic-sterilized allografts [1]. It suggests that the rate of structural valvular deterioration might not be very different between homovital or antibiotic-sterilized versus cryopreserved allograft roots. This is supported by observations in a recent update of a large Australian series, where it was shown that long-term durability of allografts implanted with the subcoronary technique was similar in antibiotic-stored and cryopreserved valves [22].
Microsimulation allows investigators to have detailed insight into the outcome of patients according to age and gender. It also provides not only actuarial but also actual outcome after AVR. When the outcome is death, there is no difference between actuarial and actual estimates. However, when the outcome measure is a nonfatal complication, actual estimates are considered more meaningful than actuarial ones [29]. In addition, microsimulation allows us to have insight into repeatedly occurring events, enables hazards to vary over time, and deals adequately with competing risks. The standard statistical techniques of outcome analysis also deal with all the issues mentioned above, albeit separately. The major strength of microsimulation is that it addresses these issues simultaneously, and it can provide a helpful adjunct to the standard statistical methods. However, the microsimulation model is a simplification of real life, requires several structural assumptions, and depends highly on the quality of the input. These issues are considered below.
Limitations of this methodology
By structuring the clinical problem using microsimulation, simplification of reality cannot be avoided. To date, the microsimulation model only considers age and gender when calculating prognosis, whereas several other factors are also important determinants of outcome, for example, the need for concomitant coronary artery operation, etiology of the aortic valve disease, heart rhythm, and left ventricular function [15, 30]. Therefore, it is still unable to make predictions that account for all of these additional risk factors.
We estimated the operative mortality rate in the microsimulation model to be 2.6% for a 45-year-old patient undergoing first-time elective cardiac operation, whereas in the meta-analysis, the pooled operative mortality rate was substantially higher because of the high percentage of patients with active endocarditis, higher New York Heart Association class, and prior cardiac operation in some of the studies in the meta-analysis.
To take into account the excess mortality after AVR that is not a result of valve-related events, a multiplicative mortality hazard was assumed. Birkmeyer and associates [31] assumed an additive mortality hazard in a simulation study of the choice of mechanical valve or bioprosthesis for replacement of the aortic valve. Further studies will have to determine the optimal model to take into account excess mortality after AVR. Currently, the microsimulation model allows the use of both additive and multiplicative mortality hazards.
Other assumptions included a constant hazard for thromboembolism, bleeding, endocarditis, and nonstructural valve failure, which might in fact depend on age and time since implantation. Finally, if a patient required reoperation after structural or nonstructural valve deterioration, the allograft was by definition replaced with a mechanical valve.
Limitations of this particular study
In a perfect world we would use a super data set with detailed and complete information on each patient, including long-term follow-up [14], for input into the microsimulation model. However, such data are not available for allograft ARR; therefore, information on patients with different characteristics who were operated on by different surgeons in different institutions was pooled. Although no clear heterogeneity was noted, patient age and the percentage of patients with preoperative active endocarditis and early death varied among the reports in the meta-analysis. Although meta-analysis was used to increase the number of patients, this study is still limited by the relatively small amount of clinical data that was pooled from only five centers. In particular, the age-specific Weibull estimates for structural valve deterioration (the most important valve-related event) were obtained from only two centers and based on just 12 cases of structural valve deterioration. Consequently, they carry a considerable amount of uncertainty, as evidenced by the sensitivity analyses. Finally, publication bias may have caused underestimation of the true occurrence of valve-related events and their consequences in this study.
In conclusion, applying microsimulation to pooled results of outcome after ARR with cryopreserved aortic allografts enabled us to have detailed insight into the factors that cause postoperative morbidity and mortality. Joint effort will be necessary to improve and regularly update the input of the microsimulation model in order to provide valid estimates of prognosis after ARR with cryopreserved aortic allografts. By adding other valve substitutes to the model, we would be able to compare the performance of different valve substitutes in the individual patient; this might be a useful tool for obtaining improved insight into the factors that determine outcome after AVR. An internet-based version of the model will become available in the near future for easy access by clinicians.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionPatients and methodsResultsAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionPatients and methodsResultsAcknowledgmentsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J.-F. Avierinos, F. Thuny, V. Chalvignac, R. Giorgi, L. Tafanelli, J.-P. Casalta, D. Raoult, T. Mesana, F. Collart, D. Metras, et al. Surgical Treatment of Active Aortic Endocarditis: Homografts Are Not the Cornerstone of Outcome Ann. Thorac. Surg., December 1, 2007; 84(6): 1935 - 1942. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. J.M. Takkenberg, L. M.A. Klieverik, J. A. Bekkers, A. P. Kappetein, J. W. Roos, M. J.C. Eijkemans, and A. J.J.C. Bogers Allografts for aortic valve or root replacement: insights from an 18-year single-center prospective follow-up study Eur. J. Cardiothorac. Surg., May 1, 2007; 31(5): 851 - 859. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Lima, G. C. Hughes, A. Lemaire, J. Jaggers, D. D. Glower, and W. G. Wolfe Short-Term and Intermediate-Term Outcomes of Aortic Root Replacement with St. Jude Mechanical Conduits and Aortic Allografts Ann. Thorac. Surg., August 1, 2006; 82(2): 579 - 585. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. P. Kappetein, J. J.M. Takkenberg, J. P.A. Puvimanasinghe, W.R. E. Jamieson, M. Eijkemans, and Ad.J.J.C. Bogers Does the type of biological valve affect patient outcome? Interactive CardioVascular and Thoracic Surgery, August 1, 2006; 5(4): 398 - 402. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Are allografts the biologic valve of choice for aortic valve replacement in nonelderly patients? Comparison of explantation for structural valve deterioration of allograft and pericardial prostheses. J. Thorac. Cardiovasc. Surg., March 1, 2006; 131(3): 558 - 564.e4.
|
|
|
|
|
|
B. G. Leshnower and T. G. Gleason Reoperative innominate arterial, ascending aortic, and root replacement for extensive fungal endocarditis J. Thorac. Cardiovasc. Surg., April 1, 2005; 129(4): 941 - 942. [Full Text] [PDF]
|
|
|
|
 |
|
 E. M. Isselbacher Thoracic and Abdominal Aortic Aneurysms Circulation, February 15, 2005; 111(6): 816 - 828. [Full Text] [PDF]
|
|
|
|
|
|
T. G. Gleason, T. E. David, J. S. Coselli, J. W. Hammon Jr, and J. E. Bavaria St. Jude Medical Toronto biologic aortic root prosthesis: Early FDA phase II IDE study results Ann. Thorac. Surg., September 1, 2004; 78(3): 786 - 793. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J.M. Takkenberg, J. P.A. Puvimanasinghe, and G. L. Grunkemeier Simulation models to predict outcome after aortic valve replacement Ann. Thorac. Surg., May 1, 2003; 75(5): 1372 - 1376. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
