| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2004;127:893-894
© 2004 The American Association for Thoracic Surgery
Letter to the editor |
Austin Hospital, Melbourne , Australia
The choice between the right internal thoracic artery (RITA) or radial artery (RA) for the second coronary artery bypass conduit remains hotly debated, as evidenced by 2 recent publications in the Journal, one a randomized controlled trial1 and the other a retrospective observational study.2
The interim 5-year result of our randomized controlled study suggested that the clinical outcomes of the RITA and the RA groups were similar. This differed from the conclusion drawn from the cohort study reported by Caputo and colleagues,2 which suggested a clinical benefit from using the RA compared with the RITA. It is important to consider possible reasons for such an apparent difference in conclusions.
Caputo and colleagues' study2 differed from the randomized trial. There was no assessment of graft patency, and they included additional end points. Obviously, cohort studies can never control for unrecognized factors. Therefore, an observational study should always be regarded as providing less reliable evidence than a randomized trial.3 This has recently been highlighted by the complete reversal of opinion as to the benefits of hormone replacement therapy once the results of randomized trials became available. One important factor that may not be well controlled in an observational study of this sort is the surgeon. The choice of graft was "entirely at the discretion of the surgeon." If some surgeons strongly preferred one type of operation, or even used it exclusively, this means that the comparison based on the observational data may suffer from bias because of confounding with the expertise of the surgeon.
Our comparison of cardiac event-free survival in the randomized controlled trial showed no significant differences, but this does not necessarily mean that the 2 studies are in conflict.
For our RA versus RITA comparison, the estimates and 95% confidence intervals (CIs) were 91% (76%-99%) for the RA group and 82% (63%-99%) for the RITA group. Because of the small numbers, these CIs are very wide and therefore consistent with a number of different possible underlying realities ("hypotheses"), namely, that the RA and RITA have the same cardiac event-free survival, that the RITA is better than the RA, and that the RA is better than the RITA. The latter is suggested by Caputo and coworkers.2 Our study is not yet in conflict with this because, for example, our findings are consistent with a "true" cardiac event-free survival of 90% for the RA group and 70% for the RITA group; both of these values are in the respective CIs, and if they were the true values, which they could be, that would be consistent with Caputo and colleagues' findings.
Biologically, the left internal thoracic artery and RITA are similar and when grafted to the left anterior descending and circumflex coronary arteries4 have almost identical results. However, when the RITA is grafted to the right coronary system, there is a significant reduction in patency. For example, in the first 5 years of our internal thoracic artery program from 1984 to 1989, patency of the in situ RITA graft to the distal right coronary artery or its branches was approximately 75%. Because of the high graft failure rate of in situ RITA grafts when used on the right side, most surgeons, including ourselves, have favored using the RITA graft to the left system,5 either as a free or in situ graft. This latter strategy results in better graft survival curves that separate by 6 years.6 Caputo and coworkers,2 on the other hand, used an in situ or pedicled graft in 94% of patients. Furthermore, 53% of the RITA grafts were attached to the right coronary or posterior descending artery where results could be less than optimal.2 Therefore, although some of their analyses adjusted for the target artery and the authors thus claim that the results were not compromised, it is not clear whether adjustment was made for the type of proximal anastomoses (in situ vs free).
The clinical end point used by Caputo and colleagues2 was a composite of death, myocardial infarction, repeat coronary bypass or percutaneous coronary intervention, and recurrent angina. Unfortunately, any composite outcome measure is only as good as its weakest end point because all outcomes are treated as having equal value. It is no surprise that the composite end point used by Caputo and colleagues is largely dependent on the presence of recurrent angina, which is responsible for 31 of 43 clinical events in the RITA group and 10 of 15 clinical events in the RA group. There were no objective criteria set for diagnosing "recurrent angina," which is always a difficult end point to assess in clinical trials. The clinical wisdom and validity of assigning the same statistical weight to angina and death defy common sense, because most patients would value the 2 quite differently.
The duration of follow-up in the study of Caputo and colleagues2 is short, a median of 1.8 years for the patients receiving a RITA graft and only 1.5 years for those receiving an RA graft. It is clear that this is a short-term study result and thus more likely to reflect perioperative factors rather than the development of graft disease.7 The study is too short to make any long-term conclusions. In our randomized trial,1 the number of patients who had reached 5 years follow-up was relatively small, and therefore longer-term conclusions are also limited. A more appropriate conclusion for the study of Caputo and colleagues would be that, in the short term, the RITA used as an in situ pedicle graft to the right coronary system seems to result in an inferior clinical outcome when compared with the RA.
Caputo and colleagues2 attempted to control analytically for confounding factors. In the Cox proportional hazards model for mortality, this means they were attempting to adjust for other explanatory variables with very small effective sample sizes (1 death in the RA group and 5 deaths in the RITA group). Such an analysis is ambitious, to say the least. An alternative form of analysis, which would probe the comparability of the groups in terms of their profiles, would be to use the propensity score approach.8 This would model the surgeon's choice of operation in terms of explanatory variables at the time of the operation. It could show that there are noncomparable subsets of the patient groups and would more generally permit an analysis that more closely mimicked a randomized trial than analyses based on modeling the outcomes in terms of all the explanatory variables.
In light of these observations, we believe that a more balanced conclusion should have highlighted the limitations of using composite outcomes to increase statistical power and raised the possibility that an in situ RITA graft placed in the right coronary system, at least in the short term, might result in inferior results. It should have also emphasized that these observations cannot be extrapolated to include free RITA grafts (used in only 6% of patients) or in situ RITA grafts placed in the left system, where many remain patent for more than 20 years.
There are satisfactory existing data to guide current surgical practice, but obviously the long-term results from trials of prospectively randomized patients are required.
 |
References |
|---|
 Top References |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. Reyes, A. Traba, L. Lopez, A. Pinto, J. Duarte, and J. L. Vallejo Neurological damage after radial artery harvesting in coronary surgery: a direct measure Interact CardioVasc Thorac Surg, August 1, 2006; 5(4): 433 - 438. [Abstract] [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 |
