| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2006;81:1578-1585
© 2006 The Society of Thoracic Surgeons
a Imperial College of Science, Technology and Medicine, Department of Surgical Oncology and Technology, St. Mary's Hospital, London, United Kingdom
b Providence Health System, Portland, Oregon
Accepted for publication December 13, 2005.
* Address correspondence to Dr Athanasiou, 70 St Olaf's Road, Fulham, London SW6 7DN, United Kingdom (Email: tathan5253{at}aol.com).
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentReferences |
|---|
METHODS: A systematic review of the literature (1965–2005) reporting results after homograft implantation in the aortic position with emphasis in the implantation technique was performed. Random and fixed-effects models were used. Cumulative and influential meta-analysis, graphic exploration, and sensitivity analysis were carried out to explain the heterogeneity between studies and to investigate potential publication bias.
RESULTS: Eleven studies were included. None of the studies was randomized. There was no significant difference in early mortality between root replacement and root preserving groups (odds ratio = 2.57 with 95% confidence interval [CI] 0.68–9.72 in random-effects model) and the root replacement group had a significantly lower rate of reoperation during long-term follow-up (hazard ratio = 0.55 with 95% CI 0.38–0.80 in random-effects model). Subgroup analysis focusing on patients having the subcoronary technique for homograft implantation showed similar results. Significant heterogeneity between studies can be detected. No publication bias was found.
CONCLUSIONS: There was no significant difference in early mortality between root replacement and root preserving groups. However, the root replacement technique group had a significantly lower rate of reoperation during long-term follow-up. Excluding outliers identified in a funnel plot reduced the heterogeneity and reached a similar conclusion.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
The "Achilles' heel" of the homograft valve is the risk of incompetence and need for reoperation in the early and long-term postoperative period. Early incompetence is usually related to technical error and late incompetence is more often due to structural degeneration of the homograft.
Two main techniques can be used for implantation of homografts in the aortic position. (1) The non-root preserving or total root replacement (RR) technique; and (2) the root preserving (RP) techniques: (a) subcoronary (SC) technique, and (b) the inclusion of cylinder (IC) technique or mini-root (a term that should be abandoned since it is not anatomically descriptive and is certainly not "mini"). Several reports emphasize the significant role of the implantation technique as a potential mechanism of homograft failure [8–10].
Initially all homografts were inserted in a subcoronary position. Recent reports emphasized a lower incidence of aortic regurgitation as an advantage of the RR technique in comparison with the SC and preference for each technique has been based on surgeon's experience rather than available clinical results [11]. It is evident in the literature that there is discrepancy between studies with regard to the technique applied and its effects on early and late clinical outcome.
The present study uses meta-analytical techniques to compare RR versus RP homograft implantation techniques in the aortic position, with regard to the short-term and long-term outcomes (mortality and freedom from reoperation). The specific questions that our study aims to answer are: (1) Are there significant differences in the above endpoints between the RR and RP techniques? (2) Is there significant heterogeneity in the estimates of the outcomes of interest between the studies comparing RR and RP techniques? (3) Is there publication bias within the studies analyzed?
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
Data Collection
A systematic review of the literature concerning homograft implantation in the aortic position with emphasis in the implantation technique was performed. A literature search was performed in MEDLINE, Embase, and Ovid for published studies between 1965 and August 2005 and without language restrictions. The following terms and mesh headings were used: "long-term results and homografts," "homografts and aortic position," "allografts," "*heart valve prosthesis implantation/methods," and "follow-up studies." The "related articles" function was used to broaden the search, and all abstracts, studies, and citations scanned were reviewed. The Cochrane database was also searched using the terms "aortic homograft" and "implantation technique." Finally, we performed a manual search of bibliographic material collected by Dr. Donald N. Ross. When several articles were reported on the same patient material the most recent article was included. Studies that contained a zero in two cells of the critical cross-tabulation tables or no event for both groups were excluded. Studies in which the outcome of comparison for both techniques was not reported or was not possible to be calculated indirectly were excluded.
The quality of each study was evaluated by examining three items: patient selection, matching of RR and RP patient groups, and assessment of outcome using a modified Newcastle-Ottawa scale (according to the needs of this study as has been described in previous meta-analytic studies [12]) (Table 1). Studies were rated on an ordinal star scoring scale, with higher scores representing studies of higher quality. Using this system, a study could be awarded a maximum of one star for each numbered item within the selection and exposure categories, and a maximum of four stars could be given for the comparability of the two groups. The quality of each study was graded as either level 1 (0–4 stars) or level 2 (5–9 stars).
|
Literature searching and critical appraisal were conducted by two reviewers (TA, CJ) and data extraction was conducted by three reviewers (TA, CJ, RJ) independently, according to prespecified protocol. Decision was taken by consensus for discrepancy cases.
Outcomes of Interest and Definitions
We compared RR with RP (SC, IC) with special attention to the following outcomes of interest: (1) early (30-day) mortality; and (2) overall reoperation rate in maximum follow-up.
Statistical Analysis
For early mortality the effect measures estimated were odds ratio (OR) with 95% confidence intervals (CI). Aggregation of the overall rates of the outcomes was performed with the Mantel-Haenszel 
2 method.
For time to reoperation, the logarithm of the hazard ratio (HR) with 95% CI was used. Two ways of estimating the HR and its variance were used: (1) If the number of events (D) and the total patient follow-up years (Y) were given or could be calculated from the Kaplan-Meier curves, the hazard ratio for each group equals D/Y and its variance equals D/Y2. The log-HR is the logarithm of the ratio of the hazard rates for the two groups. The variance of the log-HR can be calculated as 1/D1+1/D2, where D1 and D2 are the numbers of events for the two groups [13]. (2) If only the event free rate (S) and its standard error (SE) for a certain time (T) were given, the hazard rate is –log(S)/T and the variance of S is the square of its standard error. The variance of the hazard rate can be calculated by using the variance of S by combining the formulas for the variance of a logarithm and the variance of a constant.
In this study, ORs and HRs are defined as RR vs RP: an OR or HR less than one favors the RR group. The point estimate is considered statistically significant at the p less than the 0.05 level if the 95% CI does not include the value 1.
Six strategies were employed to quantitatively assess and explain heterogeneity and publication bias.
5); year of publication (after 1997); sample size (each of the groups compared contained 50 cases); studies comparing RR versus SC techniques and studies not identified as outliers in the funnel plot. Analysis was conducted by using the statistical software Inter-cooled Stata version 7.0 for Windows (Stata Corp, College Station, TX), and for graphics we used S-PLUS 6.1 for Windows (Insightful Corp, Seattle, WA).
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
|
Early Mortality
Nine studies [18–23, 25–27] contained early mortality. The incidence of early mortality was 7.0% (59/846) in the RR group and 3.8% (63/1667) in the RP group. Figure 1
(upper panel) shows the ORs for each study. Two of the studies, by Kirklin and colleagues [18] and O'Brien and colleagues [26], showed a statistically significant difference between the two groups although they were in the opposite directions; the Begg funnel plot (Fig 2) showed that they were outliers. The meta-analysis revealed that RR had significantly higher early mortality than RP in the fixed-effect model (OR = 1.6, 95% CI 1.1–2.4) and no difference in the random-effects model (OR = 2.6, 95% CI 0.7–9.7). The 2 for heterogeneity was 43.9 (df = 8) with a p value less than 0.001.
|
|
|
|
Reoperation During Follow-Up
Eleven studies were included in the meta-analysis including 1,122 patients undergoing the RR technique and 1,858 patients undergoing the RP technique (25 patients with no follow-up excluded). The RR group had a significantly lower reoperation rate in comparison with the RP group, with overall HRs calculated by fixed-effects and random-effects models of 0.53 (95% CI 0.47–0.59) and 0.55 (95% CI 0.38–0.80), respectively (Fig 5). The 2 for heterogeneity was 29.1 (p = 0.001). Begg's funnel plot in Figure 6
showed that the studies by Doty and colleagues [24] and Kilian and colleagues [28] were outliers. The sensitivity analysis for subgroups showed results similar to the overall HR. The Egger plot is presented graphically in Figure 7
showing no detection of significant publication bias: the intercept of the regression was 0.27 (95% CI 1.3–1.8).
|
|
|
Sensitivity analysis was performed separately for each of the following subgroups: quality score, size of the study, year of publication, only SC technique in the RP group, and excluding outliers (Fig 5). They all had similar results for overall HR and when excluding the outliers, the heterogeneity was 10.58 (p = 0.226).
Cumulative meta-analysis and influential sensitivity analysis were presented in Figure 8. The confidence intervals of the overall HR became less wide over time, when more studies are added. None of the studies caused significant changes in the direction of the overall HR estimate by omitting one each time.
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
Heterogeneity is a measure of the diversity between different studies focusing on a similar subject. This may be methodologic (differences between study design or case mix) or statistical (when the true effects being evaluated differ between studies, and may be detectable if the variation between the results of the studies is above that expected by chance). Significant heterogeneity was evident between the studies included in this meta-analysis. We attempted to use, extensively, most of the available graphic and statistical techniques to describe and explain this variation. By sensitivity analysis the heterogeneity was significantly reduced in the subgroup excluding outliers in the funnel plot. Although some of the heterogeneity was explained, selection bias in terms of matching the two groups for predictors of clinical outcome remains an inherent problem of nonrandomized studies.
With the sensitivity analyses and especially by using fixed and random effects models the results do change in a way that might lead to different conclusions, which indicates a need for greater caution in drawing conclusions. Such differences can be explained by the fact that studies of lower quality also tend to show larger treatment effects [29]. All the above findings support the view that interpretation of the results should be very conservative.
We identified several differences among the published series, including the following: patient selection, definitions for key-variables, time and completeness of follow up, incidence of cointerventions, outcome assessment, urgency of operation, and incidence of high risk patients. Adding to the difficulties in a comparative assessment of homograft performance are differences in procurement, sterilization, preservation, surgical method, and patient selection [30].
A significant proportion of studies were conducted from 1981 through the early 1990s, well before the widespread knowledge of the importance of the geometric relationships of the various components of the aortic root, and in particular that between the aortic annulus and the sinotubular junction. The results of the non-root-preserving technique have been published in more recent years when compared with the RP technique, and also in some large series the time period of different implantation techniques was not the same. This may result in a temporal bias, as it is well-known that mortality and morbidity for cardiac surgery decreases over time.
There are several factors that can explain the finding of increased reoperation with the RP technique in comparison with the RR technique: the surgeon's learning curve, the sizing of the homograft, and age-related anatomic changes of the aortic root which may cause progressive dilatation of the sinotubular junction with the time leading to valve failure. Willems and colleagues [31] explored the influence of learning process on the incidence of reoperation and severity of postoperative aortic regurgitation and suggested that the risk of reoperation and the echocardiographic follow-up data are significantly influenced by the surgeon's learning curve. The cutoff for learning curve was defined as 30 cases and jet diameter ratio is preferred over grading as a more valid measure to compare implantation techniques. Sizing of the aortic homograft is an important parameter. Thubrikar and colleagues [32], in an experimental study, described that the most common practice for sizing the homograft valve is to measure the inside diameter of the aortic valve annulus and a 3 to 4 mm smaller valve when implanted in the SC position will be more durable and hemodynamically beneficial. There is no doubt that homograft durability is a multifactorial issue which is primarily dependent on precise implantation techniques related to surgical training.
Study Limitations and Advantages
The application of a meta-analytic approach to calculate a single estimate of intervention effect can be misleading because of a number of reasons. First, the design of the studies included (lack of a random allocation), and the fact that no description of the criteria considered by the individual surgeons to allocate patients to each group were included. Selection bias is a particular problem due to the tendency to perform a RR in the emergency setting or for more extensive disease involving the root and ascending aorta rather than the aortic valve. Second, the two groups were not comparable for all the risk factors that can alter the outcome of interest. In fact, none of the studies were matched for the extent of disease distribution. Third, it is important to keep publication bias in mind when meta-analysis relies on previously published studies, because positive results (statistically significant results) were more likely to be published than negative results. Finally, freedom from reoperation after valve surgery is not equal to freedom from valve failure. Surgeons are less likely to reoperate on patients who had an aortic root replacement than in those who had the homograft implanted in the subcoronary position because of the complexity and increased associated risk of the reoperation.
We would like also to mention that we used various ways of estimating the HR and its variance for time-to-event as described by Parmar and colleagues [33], especially the survival curve method. Since the incidence rate of reoperation in some studies was really low, only a small part of the information was used to calculate the final HR and also some studies had quite different total follow-up years for the two groups; this method may not be optimal and can affect the results.
From a different view, valid comparison between RR and RP surgical techniques requires accounting for the hierarchical structure of the data and particularly how variables at different levels can be responsible for the variation in outcome (four levels of variation, homograft valve-patient-surgeon-institution). We attempted to evaluate and appraise the literature rather than calculate an accurate overall estimate for the outcomes of interest by using a meta-analytic technique that is in reality a hierarchical model (two levels of variation, effect size-study).
In our view there is a clear need for a controlled trial comparing the SC and RR techniques to be carried out randomly either by the same surgeon or by cooperating surgeons equally skilled in both techniques. This may be difficult to apply now in patients with informed consent if they are determined to have the conventional homograft root replacement.
Conclusions
Analysis of the data suggested that there was no significant difference in early mortality between RR and RP groups. However, the RR technique group had a significant lower rate of reoperation during long-term follow-up. Excluding the outliers identified in the funnel plot can significantly reduce the heterogeneity and get a similar overall effect.
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  A. Ali, Y. Abu-Omar, A. Patel, Z. Ali, A. Y. Sheikh, A. Akhtar, A. Pavlovic, P. Theodorou, T. Athanasiou, and J. Pepper Valve failure following homograft aortic valve replacement: does implantation technique have an effect? Eur. Heart J., June 1, 2008; 29(11): 1454 - 1462. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Athanasiou, A. Chow, C. Rao, O. Aziz, F. Siannis, A. Ali, A. Darzi, and F. Wells Preservation of the mitral valve apparatus: evidence synthesis and critical reappraisal of surgical techniques Eur. J. Cardiothorac. Surg., March 1, 2008; 33(3): 391 - 401. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 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]
|
|
|
|
 |
|
 R.-U. Kuehnel, U. A. Stock, M. O. Wendt, I. Degenkolbe, U. Jainski, M. Hartrumpf, M. Pohl, and J. M. Albes Physiological function of stentless aortic valves is altered by trimming and removal of aortic wall components Interactive CardioVascular and Thoracic Surgery, April 1, 2007; 6(2): 182 - 187. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. E. David Invited commentary. Ann. Thorac. Surg., May 1, 2006; 81(5): 1585 - 1586. [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 |
