| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2000;119:963-974
© 2000 The American Association for Thoracic Surgery
Surgery For Acquired Cardiovascular Disease |
From the Department of Thoracic and Cardiovascular Surgerya and the Department of Biostatistics and Epidemiology,b The Cleveland Clinic Foundation, Cleveland, Ohio.
Address for reprints: Bruce W. Lytle, MD, Department of Thoracic and Cardiovascular Surgery, 9500 Euclid Ave, Desk F25, Cleveland, OH 44195 (E-mail: lytleb{at}ccf.org ).
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
1 Small aortic prostheses may leave higher residual pressure gradients across the valve and are associated with less rapid and less complete regression of left ventricular hypertrophy. 2,3 Thus a large-sized prosthesis seems preferable. On the other hand, aortic root enlargement may complicate an operation for aortic valve replacement. 1,4-11 Furthermore, patients with small annular size may be small individuals, and the small valve size may be matched to their cardiac output needs. In addition to this uncertainty with respect to the small aortic root, there is also controversy about valve size and efficiency in general. Both the use of stentless aortic valve prostheses and the use of aortic root–enlarging procedures are strategies based in large part on the thesis that hemodynamic performance, and thus valve size, favorably influence late outcome. 12 The purpose of this study was to ascertain the relation of prosthesis size to survival after aortic valve replacement. |
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
18 years of age) operated on at The Cleveland Clinic Foundation from 1978 through 1996 with (1) aortic stenosis (with or without regurgitation); (2) no concomitant procedures, such as coronary artery bypass grafting; (3) no previous cardiac surgery; and (4) one of three different categories of devices with different expected orifice sizes with respect to the anulus. Through the Cardiovascular Information Registry (CVIR), a prospective ongoing registry of surgical and cardiologic procedures, we identified 892 patients with aortic stenosis who had undergone primary isolated aortic valve replacement with either (1) a mechanical prosthesis (St Jude Medical, standard, A101, n = 346; St Jude Medical, Inc, St Paul, Minn), (2) a bovine pericardial valve (Carpentier-Edwards Perimount RSR 2800, n = 463; Baxter Healthcare Corp, Edwards Division, Santa Ana, Calif), or (3) a cryopreserved allograft (CryoLife, n = 83; CryoLife, Inc, Kennesaw, Ga). Preoperative symptoms were classified according to New York Heart Association (NYHA) criteria. Patients underwent preoperative left heart catheterization with coronary arteriography, echocardiography, or both. Left ventricular function was evaluated by echocardiography and by the left ventriculogram, when available, and grouped by function as being normal or as having mild, moderate, or severe impairment. Other variables used in the data analyses were obtained from the concurrently abstracted data in the CVIR (see Appendix I).
Prostheses
Prosthesis selection was according to surgeon preference. The numbers of patients receiving each prosthesis type and size are presented in Table I. The relation of prosthesis size to survival was investigated by using 4 different expressions of prosthesis size. First, the manufacturer’s labeled size (see Table I) is commonly used to express prosthesis size. However, this was the external sewing ring size in the mechanical valve, the tissue mounting device (anulus) size in the pericardial valve, and the internal diameter in the allografts. Second, each manufacturer provided us with its official documentation of internal valve orifice diameter (Table I), and we used this measurement as a uniform valve size expression across valve types, as suggested by Christakis and colleagues. 13 Third, we then calculated the indexed internal valve orifice area as the internal area divided by the body surface area of the patient (in cm2/m2). Finally, the final expression of valve size was the standardized internal valve orifice size or Z-value based on body surface area. The Z-value is the number of SDs the internal orifice size differs from the predicted mean normal native aortic valve size on the basis of the patient’s body surface area (Appendix II). 14,15
|
Follow-up and end point
Patients were followed within the CVIR at 2-year intervals. Attempts were made to contact any patient for whom follow-up was not current by mail, telephone calls, or both during the summer of 1998. In all, patients were followed for 4415 patient-years. Mean follow-up among survivors was 5.0 ± 3.9 years; 25% were followed up for 2 years or less, and 25% were followed up for 6 years or more, with a maximum of 20 years. Twenty-seven patients had incomplete follow-up of less than 6 months’ duration (3%).
The end point for this study was death from any cause, with time zero being the time of operation.
Data analysis
Adjustment for patient selection
Although the availability of several different categories of valves (mechanical, pericardial, and allografts) provided an opportunity to investigate a wide range of internal valve orifice sizes, the kind of patients receiving each were not the same (Table II), and the time frame for the use of each prosthesis was not uniform (Appendix Fig 1). To take into account these selection factors, the probability that a patient would receive one of the 3 types of valve was determined by logistic regression of each pair of valves. For this so-called propensity score (the propensity to receive one or another valve), demographic, symptom and cardiac comorbidity, and noncardiac comorbidity factors shown in Table II were entered into each analysis irrespective of P values. 16-18 Because the models incorporated identical variables, the sum of the conditional probabilities predicted for receiving each valve add to 100%. The pair-wise propensity scores in the format of logit units were incorporated into multivariable analyses of death as a way to adjust for selection factors.
|
19
Analysis of survival
Nonparametric estimates of survival were obtained by the method of Kaplan and Meier. 20 A parametric method was used to resolve the number of phases of instantaneous risk of death (hazard function) and to estimate its shaping parameters. 21
Exploratory analyses of the variables in Appendix I included correlation analysis, stratified life table analyses, and decile risk analysis of ordinal and continuous variables to determine the possible transformations of scale needed to properly calibrate the variables to survival.
The risk of small valve size might be prominent early after the operation or it may be a long-term risk. For this reason, we used a method that simultaneously but specifically interrogated the early, constant, and late hazard phases after the operation. 21 The analyses used a directed technique of entry of variables into the multivariable model. 22 The P value criterion for retention of variables in the final model was .05.
Reliability of the analysis
Because we were unable to identify an influence of prosthesis size on survival, we ascertained the reliability of this finding by using bootstrap resampling. 23 Patients were drawn at random from the 892 study members with replacement. This was repeated to construct a new data set of 892 observations, which could contain one or more duplicates of patients. This random sampling process was repeated 1000 times to create 1000 differently constituted data sets. We then inquired how often smaller valve size would be selected as a risk factor at a P value of less than .05 in each hazard phase adjusted for the other factors found by parsimonious modeling and the propensity scores. 24
The bootstrap variable selection process was carried out for indexed internal valve orifice area overall and repeated with the indexed areas for the individual valve types (as interaction terms). This entire process was repeated by using the standardized internal orifice size (Z-value). Thus 4000 separate data sets were actually used in these 4 investigations. Because of the small number of events in the early hazard phase, this investigation was only conducted for the constant and late hazard phases.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
. The complete spectrum of internal orifice area indexed to body surface area and its relation to the type of valve is shown in Fig 1. Only 7 patients received valves with indexed internal orifice areas below 0.85 cm2/m2. Mechanical prostheses had the smallest orifice size (1.3 ± 0.29 cm2/m2, Z = –2.2 ± 1.16), and allografts had the largest in relation to body size (2.1 ± 0.50 cm2/m2, Z = 0.24 ± 1.17), with pericardial valves being intermediate between the 2 (1.9 ± 0.36 cm2/m2, Z = –0.40 ± 1.01). A quarter of the patients, mostly those receiving a mechanical prosthesis, had internal valve orifice sizes more than 2 SDs below normal native aortic anulus size (Fig 2).
|
|
|
2, P < .0001; older: –1.38 ± 0.67·[50/age], P = .04).
Survival
The overall survival was 98%, 96%, 86%, 69%, and 49% at 30 days and 1, 5, 10, 15 years, respectively (Fig 3). There were only 13 hospital deaths (1.5%), precluding any meaningful examination of early risk factors. Thus we focused on deaths across time from the time of operation. The instantaneous risk of death (hazard function) was highest immediately after valve replacement but fell rapidly to its lowest level and gradually rose after about 3 months. This pattern of risk and the methods of analysis used allowed us to focus on the influence of valve size at various time intervals after operation.
|
|
|
, B ). Although not a statistically significant association (P = .6), the small differences were well accounted for by differences in prevalence of risk factors. These same findings were true of internal orifice size standardized to predicted normal native aortic valve size (P = .8; Fig 5, C ). Thus all differences related to prosthesis type and various expressions of valve size were well accounted for by differences in prevalence of risk factors.
|
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
We have used modern sophisticated methods to interpret the data in the face of nonuniform prosthesis selection and the nonrandomized nature of the study. We have then tested the reliability of our inferences by using computer-intensive sampling methodology.
Study limitations
The patients represent a single institution’s experience and therefore may not be representative of other venues. This is not a randomized study and is retrospective, leaving the possibility that selection bias might cause differences in patient groups that might influence survival but are not accounted for, even with the extensive multivariable analyses conducted. Furthermore, this study does not examine all outcomes but only survival. For patients undergoing aortic valve replacement, meaningful survival data can only be obtained from studies with large patient numbers. It is quite difficult to examine indices of left ventricular mass regression over time and subtleties of late symptom status for this large number of patients followed for these postoperative intervals.
Importantly, the study is limited to patients with at most moderate patient-prosthesis mismatch 25 because no deliberate attempt was made to use an undersized prosthesis. Nevertheless, some 25% of patients had moderate mismatch.
The patients were not randomized with respect to either valve type or valve size. Indeed, deliberate selection of specific valve types for different patients is generally accepted in this mature era of valve replacement. We used 3 valve types to examine a wide spectrum of internal orifice sizes, even though this introduced a possible confounding variable. We addressed this by use of propensity scores to account for this selection. Of course, we were unable to take into account unrecorded selection factors. The fact that in no instances were the propensity scores associated with survival indicates that the factors in the multivariable analysis adjusted well for patient heterogeneity with respect to valve type selection.
We have used actual (in vitro) valve dimensions, even when normalized and standardized to body surface area, rather than effective valve orifice area. The effective valve orifice area is a dynamic, and not static, measurement that varies with temporal loading conditions and flow. Thus it is a poor variable for the kinds of comparison done in this study. Commonly, the labeled valve size has been used in discussions of small prostheses. This size clearly refers to different measurement points for each valve type. We have therefore used a uniform measurement for comparisons: the internal diameter of the prosthesis.
Finally, follow-up, although extending up to 20 years, averaged only 5 years. Thus the inferences apply mainly to intermediate-term survival.
Valve size
Both the use of stentless aortic valve prostheses and the use of aortic root–enlarging procedures are strategies based in large part on the thesis that favorable hemodynamics from a large and efficient prosthetic orifice size will improve late outcomes after aortic valve replacement.
In part, this logic is based on the exponential increase in the transprosthetic pressure gradient as indexed effective orifice size decreases, particularly below about 1 cm2/m2. 26,27 The magnitude of this gradient in a pulse duplicator system may be 20 mm Hg at a cardiac index of 4.5 L · min–1 · m–2 at an indexed effective orifice area of 1.0 cm2/m2. Because prostheses are not perfectly efficient in vivo, the effective orifice area is generally less than the calculated internal orifice area used in our study. Any component of left ventricular outflow obstruction must be added to the transprosthetic pressure gradient. 27
In part, this logic is based on the more rapid and complete regression of left ventricular hypertrophy in the first few years after aortic valve replacement, as assessed by echocardiography. 2,3,28 In part, it is also because others have found a relation between manufacturer-labeled prosthesis size and overall time-related survival. 29 The magnitude of this relation appeared to be modest and in part related to differences in mortality after the operations.
In the present study the consumption of space by sewing rings and stents is reflected in the indexed orifice area and standardized orifice size. However, we were unable to demonstrate an adverse relation between smaller valve size by any of the 4 different expressions and intermediate-term survival. This is a broad statement in the context of many older patients requiring aortic valve replacement for aortic stenosis; there may be individual patients for whom prosthesis size may be important.
This leaves unexplained the differences in intermediate-term survival in the case-matched study of the Toronto stentless prosthesis compared with a stented xenograft. 12 One can question how well matched that study was (eg, with more aortic regurgitation patients) in the stented group. Aortic enlargement was also used to allow larger stented valve sizes to be used, and this increases the early risk of operation. 4
This study has focused on survival. Survival is not the only outcome after aortic valve replacement, although it is the most important. There are situations where the use of complex forms of aortic valve replacement, such as aortic valve homografts, pulmonary valve autotransplantation, aortic root–enlarging procedures, and, perhaps, stentless heterograft valves, to achieve a more efficient aortic valve substitute are indicated by a desire for a high patient activity level. In those situations we have usually used aortic valve homografts or pulmonary valve autotransplantation. However, on the basis of our study, despite the arguments from logic, survival appears not to be adversely affected by aortic prosthesis size, even down to moderate patient-prosthesis mismatch. Thus except in selected circumstances, doing an operation, such as aortic root enlargement, that increases short-term risk to achieve a better long-term survival is not supported by these data.
|
Appendix I: Variables |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
|
Appendix II: The standardized valve size |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
14,15 Generally, the normal structure size has been based on a logarithmic relation between structure size and body surface area. A standardized expression is particularly important in growing babies and children but has not been used extensively for adults. For this study, we used the following relation between aortic anulus diameter and body surface area (BSA), where ln is the natural logarithm:
|
|
Appendix: Discussion |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
For aortic valve replacement, the goal must be the achievement of optimal hemodynamics and regression of left ventricular hypertrophy. We thought it could be reached by using the largest possible prosthetic valves. In your study you could demonstrate that an effective prosthetic valve orifice area that is not less than 4 SDs from the native aortic anulus index is as efficient, and I think this is an important result. However, the literature demonstrates that especially in the early outcome operative mortality rate is higher in patients receiving very small aortic valves, and this is in contrast to your own experience.
Dr Medalion, I would like to ask you 3 questions. Do you think that, because this study was a nonrandomized trial and prosthesis selection was to the surgeon’s taste, this statistical analysis is really able to finally answer this question? You are well aware of other reports with high, early operative mortality rates with very small prostheses. On the basis of your results, what do you think about stentless aortic valves for patients with small aortic roots? The risk factor of left ventricular hypertrophy and the measured mass of left ventricular hypertrophy did not seem to be analyzed in your study.
In randomized prospective trials of stented valves versus stentless valves, as well as in Tirone David’s series and our series, it is clear that earlier and rapid regression of left ventricular hypertrophy occurs if you calculate this risk factor. Do you have any data on that?
Finally, please give us your current strategy from The Cleveland Clinic in patients with small aortic roots after you reviewed your data? Are you advising us not to hesitate to implant these small valves?
Dr Medalion. Thank you, Dr Mohr, for discussing the article. I will try to answer all 3 questions.
The findings of this study were somewhat surprising to us. We thought that there would be extremes of valve-patient mismatch in which we would see a negative effect on survival, and we used very extensive and advanced statistical analyses in attempts to define a level of mismatch that was important. However, we were unable to demonstrate this effect. There are, however, a couple of important points to keep in mind.
First, these operations were performed, in general, by surgeons with a bias in favor of larger valves, and therefore severe valve-patient mismatches were uncommon. Second, the end point of the study is survival. With this large number of patients being followed up, it is impossible to test exercise capacity and symptom status for them all. Valve size may be important for patients for whom a high activity level is important. In our current practice, we use aortic valve homografts, aortic root– enlarging procedures, and pulmonary valve autotransplantation for young patients with high activity level expectations. Thus we believe there are some indications for complex aortic root operations to achieve maximal valve efficiency.
However, the thought that no patient should ever receive a size 19 or size 21 valve because that strategy will lead to an increased risk of late death cannot be supported by these data. For most patients, the in-hospital risk of aortic valve replacement with standard prostheses is extremely low. Performing more complex operations makes sense only if they do not increase the risk of in-hospital mortality because the idea that a moderate valve-patient mismatch will lead to a shortened long-term survival is a concept that has no data to support it at the present time.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and methods Results Discussion Appendix I: Variables Appendix II:
The standardized...Appendix: Discussion References |
|---|
This article has been cited by other articles:
|
|
 |
|
  J Mascherbauer, R Rosenhek, C Fuchs, E Pernicka, U Klaar, C Scholten, M Heger, G Wollenek, G Maurer, and H Baumgartner Moderate patient-prosthesis mismatch after valve replacement for severe aortic stenosis has no impact on short-term and long-term mortality Heart, December 1, 2008; 94(12): 1639 - 1645. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. C. Waisbren, L.-M. Stevens, E. G. Avery, M. H. Picard, G. J. Vlahakes, and A. K. Agnihotri Changes in mitral regurgitation after replacement of the stenotic aortic valve. Ann. Thorac. Surg., July 1, 2008; 86(1): 56 - 62. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Casali, G. Luzi, M. Vicchio, P. Lilla della Monica, G. Minardi, and F. Musumeci Echocardiographic Follow-Up after Implanting 17-mm Regent Mechanical Prostheses Asian Cardiovasc Thorac Ann, June 1, 2008; 16(3): 208 - 211. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Mihaljevic, E. R. Nowicki, J. Rajeswaran, E. H. Blackstone, L. Lagazzi, J. Thomas, B. W. Lytle, and D. M. Cosgrove Survival after valve replacement for aortic stenosis: Implications for decision making. J. Thorac. Cardiovasc. Surg., June 1, 2008; 135(6): 1270 - 1279.e12. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Kohsaka, S. Mohan, S. Virani, V.-V. Lee, A. Contreras, G. J. Reul, and S. A. Coulter Prosthesis-patient mismatch affects long-term survival after mechanical valve replacement. J. Thorac. Cardiovasc. Surg., May 1, 2008; 135(5): 1076 - 1080. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S Bleiziffer, W B Eichinger, I Hettich, D Ruzicka, M Wottke, R Bauernschmitt, and R Lange Impact of patient-prosthesis mismatch on exercise capacity in patients after bioprosthetic aortic valve replacement Heart, May 1, 2008; 94(5): 637 - 641. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Yoshikawa, S. Fukunaga, K. Arinaga, H. Hori, E. Nakamura, T. Ueda, E. Tayama, and S. Aoyagi Long-Term Results of Aortic Valve Replacement With a Small St. Jude Medical Valve in Japanese Patients Ann. Thorac. Surg., April 1, 2008; 85(4): 1303 - 1308. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Rao Invited Commentary Ann. Thorac. Surg., April 1, 2008; 85(4): 1317 - 1318. [Full Text] [PDF]
|
|
|
|
 |
|
 N. D. Desai and G. T. Christakis Bioprosthetic Aortic Valve Replacement: Stented Pericardial and Porcine Valves Card. Surg. Adult, January 1, 2008; 3(2008): 857 - 894. [Full Text]
|
|
|
|
|
|
B. A. Youdelman, H. Hirose, H. Jain, J. Y. Kresh, J. W.C. Entwistle III, and A. S. Wechsler Comparison of eight prosthetic aortic valves in a cadaver model. J. Thorac. Cardiovasc. Surg., December 1, 2007; 134(6): 1526 - 1532. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Taniguchi, T. Takahashi, K. Toda, H. Matsue, Y. Shudo, H. Shintani, M. Mitsuno, and Y. Sawa Left ventricular mass: impact on left ventricular contractile function and its reversibility in patients undergoing aortic valve replacement Eur. J. Cardiothorac. Surg., October 1, 2007; 32(4): 588 - 595. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Dhareshwar, T. M. Sundt III, J. A. Dearani, H. V. Schaff, D. J. Cook, and T. A. Orszulak Aortic root enlargement: What are the operative risks? J. Thorac. Cardiovasc. Surg., October 1, 2007; 134(4): 916 - 924. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Garcia Fuster, V. Estevez, I. Rodriguez, S. Canovas, O. Gil, F. Hornero, and J. Martinez-Leon Prosthesis patient mismatch with latest generation supra-annular prostheses. The beginning of the end? Interactive CardioVascular and Thoracic Surgery, August 1, 2007; 6(4): 462 - 469. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. B. Vanky, E. Hakanson, and R. Svedjeholm Long-Term Consequences of Postoperative Heart Failure After Surgery for Aortic Stenosis Compared With Coronary Surgery Ann. Thorac. Surg., June 1, 2007; 83(6): 2036 - 2043. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Pibarot and J. G. Dumesnil Prosthesis-patient mismatch in the mitral position: Old concept, new evidences J. Thorac. Cardiovasc. Surg., June 1, 2007; 133(6): 1405 - 1408. [Full Text] [PDF]
|
|
|
|
|
|
S. Bleiziffer, W. B Eichinger, I. Hettich, R. Guenzinger, D. Ruzicka, R. Bauernschmitt, and R. Lange Prediction of valve prosthesis-patient mismatch prior to aortic valve replacement: which is the best method? Heart, May 1, 2007; 93(5): 615 - 620. [Abstract] [Full Text] [PDF]
|
|
|
