| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2006;82:1420-1429
© 2006 The Society of Thoracic Surgeons
a Division of Thoracic Surgery, Department of Surgery, Division of Cardiology, Department of Medicine, Duke University Medical Center, Durham, North Carolina
b Duke Clinical Research Institute, Durham, North Carolina
Accepted for publication April 13, 2006.
* Address correspondence to Dr Smith, Department of Thoracic Surgery, Duke University Medical Center, PO Box 3442, Durham, NC 27710 (Email: smith058{at}mc.duke.edu).
Presented at the Forty-second Annual Meeting of The Society of Thoracic Surgeons, Chicago, IL, Jan 30–Feb 1, 2006.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
METHODS: In this study, 18,481 patients with significant (>75% stenosis) CAD treated at a single center between 1986 and 2000 were assigned to one of three groups based on initial treatment strategy: medical therapy (MED) (n = 6862), PCI (n = 6292), or CABG (n = 5327). Each group was categorized into 3 groups according to baseline severity of CAD: low-severity (predominantly 1-vessel), intermediate-severity (predominantly 2-vessel), and high-severity (all 3-vessel), and prospectively evaluated in Cox models for all-cause mortality adjusted for cardiac risk, comorbidity, and propensity for selection of a specific treatment. Treatments were compared for the entire period and three eras (1: 1986 to 1990; 2: 1991 to 1995; 3: 1996 to 2000), the last encompassing widespread availability of PCI with stenting.
RESULTS: Survival significantly improved in all groups for all degrees of CAD, despite increasing severity of illness. Revascularization strategies provided significant survival over MED with 8.1, 10.6, and 23.6 additional months per 15 years of follow-up for low-severity, intermediate-severity, and high-severity CAD, respectively. Therapeutic improvements led to increased survival of 5.3 additional months per 7 years of follow-up (95% confidence interval, 0.2 to 10.2; p = 0.039) in era 3 for CABG compared with PCI for high-severity CAD.
CONCLUSIONS: Initial revascularization strategies result in significant survival advantage over MED for all CAD levels. Patients with high-severity CAD have reduced survival with PCI compared with those initially treated with CABG.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
The rapid development of PCI methodologies and substantial improvements in CABG have resulted in unresolved questions about optimal decision making in selecting the initial procedure most likely to prolong a patient's life. Before coronary stenting, CABG was shown to be superior to PCI in patients with diabetes, but opinions differ about the merits of current PCI technology [8, 9]. We undertook this study to (1) evaluate changes over time in survival of patients with CAD, (2) assess the overall association between revascularization and survival, and (3) specifically, to determine the relation between initial choice of treatment and survival in the modern era of PCI, CABG, and MED.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
Angiographically detected stenosis of at least 75% in one or more coronary arteries was recorded in 18,481 patients, after excluding patients with more than 50% left main coronary artery stenosis, PCI within one year of catheterization, previous CABG, more than 2+ mitral regurgitation, or cardiogenic shock.
The focus of the analysis was to evaluate treatment-related survival as a function of the initial treatment strategy. Within 30 days after the initial cardiac catheterization, patients were (1) assigned to the PCI group if they underwent PCI, (2) placed in the CABG group if they received CABG not preceded by PCI, or (3) defaulted to the MED group if they received neither. Patients who died within 5 days (median interval between catheterization and CABG at DUMC during study enrollment) of initial catheterization without receiving PCI or CABG were excluded from MED. This was to avoid bias associated with early medical deaths before the patient could undergo revascularization. For patients undergoing revascularization, follow-up was begun at the time of initial intervention.
Once patients were assigned to a treatment group, all outcome events remained with that group, regardless of whether the patient underwent subsequent revascularization. Revascularization techniques used in the CABG and PCI groups are as previously described [10, 11].
Because of widespread availability of stenting since 1996, patients undergoing catheterization that year or thereafter were assigned to the era 3 (1996 to 2000) group. The remainder was divided into two 5-year periods: era 1 (1986 to 1990), and era 2 (1991 to 1995). Follow-up was at 6 months and every year thereafter on the anniversary of initial presentation. Data were obtained by self-administered questionnaires and telephone contact and were 94% complete over all follow-up intervals. The National Death Index was used to determine vital status for nonresponders [12].
The anatomic location of coronary artery stenoses and their severity have been shown to correlate with patient mortality after revascularization [2]. To categorize patients by disease severity, the Duke index of CAD severity was used. This index was derived from 7-year follow-up of 29,082 patients undergoing catheterization at DUMC between 1986 and 2000 who were treated without revascularization [13]. Patients within nine of the original 13 categories were included in this study. Those excluded had either insignificant coronary artery disease (<75% stenosis in any vessel) or significant left main coronary disease (
75%), which generally precluded medical or percutaneous therapy during the period. These nine categories were consolidated into 3 groups, with continuously increasing hazard of cardiovascular death: low-severity CAD (predominantly 1-vessel), intermediate-severity CAD (predominantly 2-vessel), and high-severity CAD (all 3-vessel). Anatomic definitions for the nine Duke CAD index categories and the three CAD severity groupings are in Table 1.
|
To develop the risk-adjustment model, the pool of all covariates shown to be both clinically relevant and statistically important was evaluated in univariable Cox models. Continuous and ordinal categoric variables were tested for linearity over the log hazard and were transformed as necessary to meet this modeling assumption. Stepwise selection of the covariates was used to pick those that were significant (
= 0.05) independent predictors of mortality in the multivariable setting. The three CAD severity levels were incorporated as two dichotomous covariates (low-severity CAD as reference group) in all Cox models. These two covariates were combined with treatment to determine differences in the effect of CAD severity on survival across therapies. These interaction tests were significant for revascularization versus MED (p < 0.0001) and CABG versus PCI (p = 0.031) models.
Logistic regression analysis was used to adjust for physician or patient propensity to choose a specific therapy, thus reducing some of the bias introduced by nonrandomized treatment selection. Two propensity scores, used as covariates in the multivariable Cox models, were created for each patient, one for revascularization (versus MED), the other for CABG versus PCI (no MED), representing that patient's probability of selection for the treatment being modeled.
Including treatment as a modeling covariate violated proportional hazards; therefore, the final Cox multivariable adjustment model consisted of baseline stepwise-selected covariates, appropriate propensity score, and CAD group variables plus the interaction terms of CAD severity by treatment, with treatment included as a model stratification variable.
Adjusted survival curves for each treatment were calculated by applying its estimated baseline hazard function plus covariate Cox model parameter estimates to all patients in the cohort and then averaging over all patients at each time point. The resulting curve represents an estimate of survival that would occur if all patients were assigned to that treatment. The difference between the estimated survivals of the two treatments was tested statistically by calculating areas under each of the curves and then bootstrapping this process to obtain 95% confidence intervals (CI), standard errors, and p values. The difference was expressed as additional survival time provided by the superior treatment during the relevant follow-up interval.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
Improved overall survival of patients after coronary angiography is gratifying and remains consistent with community-based epidemiologic [15] and medical cohort studies [15, 16]. Improvements during the past 15 years are almost certainly partially attributable to the remarkable improvements in acute [17] and chronic pharmacology [18], as well as risk factor modification [19]. Improvement in outcomes with PCI [20] and CABG [21] has also been demonstrated nationally.
In our study, an increasing proportion of patients with significant CAD were treated with revascularization, with most due almost exclusively to the increasing use of PCI. Whereas this trend was most evident in patients with low-severity and intermediate-severity CAD, treatment selection in high-severity CAD also changed slightly in favor of PCI. Our findings and those of other investigators [22, 23] raise concern about the recent trend toward recommending PCI as the initial treatment for high-severity CAD. These patients at greatest risk of premature cardiovascular death have the most to gain from revascularization, but a substantial and increasing proportion are not being treated initially with CABG—the therapy most likely to provide longevity benefit.
These results are consistent with other recent observational data showing a survival advantage of CABG over PCI in patients with severe CAD [22, 23], despite improvements in PCI related to the availability of stents. Our data confirm these findings and add new information, demonstrating an increased survival advantage for patients with severe CAD when CABG is chosen over PCI compared with the pre-stent era.
The advantage of CABG over PCI was initially surprising to many practitioners who assumed that stenting and better adjunctive pharmacology would give PCI an advantage over CABG. This was because of the ability to reduce abrupt closure and prevent periprocedural MI and restenosis and the need for an emergent operation. Our findings underscore the benefit of these advances, yet concomitant advances in surgery may have been less appreciated.
Our results suggest that the increasing use of PCI in patients with low-severity and intermediate-severity CAD may not have maximal effect on survival compared with CABG. The survival differences are small enough, however, that reduced periprocedural pain, shorter recovery time, and reduced procedural cost associated with PCI make it an attractive alternative to CABG in lower-risk patients. It can be argued that since initial PCI offers a modest advantage over MED for survival of Duke patients [2], its increasing use (>60% for low-severity CAD in Era 3) is an appropriate result. The more aggressive application of a superior therapy, principally in a group of patients who would previously have been treated medically, would be expected to reduce the overall survival advantage. The only available randomized clinical trial (RCT) data have not shown a benefit of PCI over MED in nonfatal events [24–27]. For patients in distress, PCI has been found to be superior to MED alone for symptom relief. The ongoing Clinical Outcomes Utilizing Percutaneous Coronary Revascularization and Aggressive Drug Evaluation (COURAGE) trial [28] is addressing this issue by comparing combined PCI and MED versus MED alone in patients with less severe CAD.
The advantage of CABG over other interventional methods for treating more severe coronary disease was surprisingly robust and more impressive in recent periods, even though metal stents were widely available. Whereas individual RCTs have shown no advantage for CABG [29–33], a recent overview of trials showed a benefit for CABG over PCI in patients with multivessel disease [34], similar to the observations in this study. The investigators found this difference to be consistent continuing into the era of coronary stenting. Several recent systematic overviews have demonstrated no effect of bare metal stenting on survival in more than 10,000 patients entered into RCTs [35, 36]. In the era of stenting, it is not surprising that results showed no advantage over balloon angioplasty alone. Whereas drug-eluting stents represent an advance over bare metal stents in preventing restenosis, recent systematic overviews [37, 38] show no benefit in prevention of death or MI. These results highlight the urgent need for an RCT in patients with multivessel CAD to compare PCI and CABG in the era of drug-eluting stents.
Limitations
This study has some important limitations. First, the treatments were not randomly allocated; rather, the treating physician and patient selected the initial treatment. Although methods were used to correct for the propensity to select a particular treatment, there are almost certainly some patient characteristics for which the model does not correct. For example, the increasing application of either CABG or PCI reflects a bias toward coronary intervention, leaving an increased proportion of the MED group who may not be suitable for intervention and may be at increased risk for cardiovascular death. While these uncertainties should not be underestimated, we found that propensity and risk factor adjustment in our data yielded results that were confirmed by later RCTs [39].
Second, all three treatment options have undergone considerable change during the enrollment period. We did not consider variation in treatment over time as a covariate, but simply considered year of entry as a surrogate variable for treatment improvement. The failure of PCI to obtain an advantage over CABG in more complex forms of CAD may be due to improvement in CABG therapy or to the selective application of angioplasty alone rather than angioplasty and stent placement in these patients. As with all observations of changing therapy, these data pertain only to the treatments used during the study period and should be extended to more current therapeutic methodologies with caution.
Third, the patients studied were treated at a referral center in the southeastern United States and may not be representative of the general patient population with CAD.
Conclusion
This analysis reveals improved survival in patients undergoing coronary angiography. Although better medical therapy is partially responsible, early revascularization after coronary angiography is associated with improved survival in virtually all patients with significant CAD. PCI provides benefits similar to those of CABG in patients with less severe CAD, but CABG is a superior procedure for severe multivessel disease. Pending the results of ongoing RCTs, our findings and those of recent reports support continued expansion of coronary revascularization, with a caveat regarding the choice of PCI as initial therapy in patients with severe multivessel disease. Practitioners and patients should be aware that CABG, compared with PCI, has been associated with a substantial improvement in survival in patients with multivessel disease in this and two other independent observational studies [22, 23].
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
Fundamentally, there are two types of studies that are taken seriously regarding bypass surgery versus percutaneous intervention for the treatment of coronary artery disease. One type is a randomized prospective trial, and there have been a whole bunch of these over the years, usually with an acronym like BARI, SOS, or the ARTS trials. In general those trials have either shown no difference in survival rate between PCT and surgery, or relatively small differences. Our cardiology colleagues have seized upon these trials as the embodiment of truth in part because of the conclusion that PTCA does not compromise survival and in part because it sounds like a good idea to do a trial where bias is not exerted at the point of treatment selection.
It is important to remember, however, that randomized trials are biased; they are just not biased at the point of treatment selection. They are biased at the point of inclusion into the trial. Therefore, the randomized trials have been made up of relatively low risk patient subsets. If we look at overall survival rates in the randomized trials, all patient subsets have better survival rates than the corresponding groups in observational studies such as that which Dr Smith and his colleagues have presented.
The second type of study, and that which Dr Smith has presented, is a retrospective analysis of large patient groups using risk adjustment to allow comparison of patients treated with percutaneous therapies and those receiving surgery. These tend to be very inclusive studies where treatment selection was biased but inclusion into the study was not biased. In general, these types of studies have shown an improvement in survival rate associated with surgery for high-risk groups. Where along the spectrum from low risk to high risk the survival advantage for surgery has occurred has varied from trial to trial. Previously published studies of this type include a single center trial from The Cleveland Clinic Foundation, and a study derived from the New York State-wide Database published in the New England Journal of Medicine earlier this year. These studies have a lot of patients in them, a higher risk group of patients and more events. They also show an advantage for surgery.
A unique feature of Dr Smith's study is that it compares percutaneous therapy and surgery to patients treated medically. Dr. Smith's conclusion is that revascularization of both types enhances survival rate and this is the first major study that I am aware of that shows an improvement in survival for PTCA in regard to medical treatment. But, what the weight of the data has shown is that the higher the risk profile of the patients examined, the more the survival advantage of bypass surgery appears to be demonstrated.
I have a couple of questions for Dr Smith. One, how they risk-adjusted for acute myocardial infarction. Stenting has become the preferred treatment for acute myocardial infarction, and the risk adjustment strategies for that issue are difficult. How did they accomplish that?
The second question is, how do he and his colleagues feel that the advent of drug-coated stents pertains to their conclusion?
DR SMITH: Thank you very much for your kind discussion, Dr Lytle. Regarding acute myocardial infarction, it was apparent in analyzing our data that acute myocardial infarction was more the province of percutaneous intervention than surgery, although it occurs in of both. The timing between myocardial infarction and intervention was one of the risk factors included in the multivariable model. And I should mention, and I did not have it on the slide, that a small number of acute myocardial infarctions that were salvage type patients in the cath lab, with no other alternative, were excluded from the analysis because it was clear that all three options of therapy were not available to the patient.
The second question regarding drug-eluting stents gives me an opportunity to speculate a bit. Relevant to the discussion is the fact that the drug-eluting stent has been shown to have no advantage over the bare metal stents in terms of survival, although it was anticipated that it might. So whether or not these data can be extended into the drug-eluting stent era are unknown, but there are some features of these data that suggest that the results will be similar in the drug-eluting era, the similar features being that unlike a randomized controlled trial, the percutaneous intervention here represents real world decision-making where there has been extension of intervention percutaneously into more and more difficult coronary artery disease patients. Therefore, the results may not be quite as optimal as in patients preselected for inclusion in a randomized controlled trial. We know that the initial era of drug-eluting stents involved large stents being placed in smaller vessels, due to lack of availability of smaller stents, and obviously the incidence of abrupt closure, which is about 3% at a year with a 50% mortality, basically is equal to most center's operative mortality. So we would expect that the results would be similar. We now have 3 years of follow-up from the drug-eluting stent era and are carefully looking at that.
DR ROBERT A. GUYTON (Atlanta, GA): Thank you. That was a beautiful presentation, Peter. I wish to make a comment and ask Peter an ethical question. I believe there is compelling evidence now that coronary bypass is superior to percutaneous intervention for patients with multivessel disease. Randomized controlled trials have shown a survival advantage for coronary bypass versus PCI, but more compelling is registry data. I agree with Bruce that data from entire captive populations or data from the real world are equally important or perhaps more important. The New York State registry reported in 2000 in the pre-stent era a highly significant survival advantage for coronary bypass at 3 years, a 43% relative survival advantage for triple-vessel disease, including the proximal LAD. But the cardiologists responded, now we have stents, and again, they told their patients that there was no mortality difference between stenting and coronary bypass. Last year the data were presented from New York for the stent era. There is still the same highly significant survival advantage for coronary bypass, now a 46% survival advantage compared to stented patients. Cardiologists respond, now we have drug-eluting stents.
(Slide) But as this meta-analysis from Lancet showed, there is no survival advantage for drug-eluting stents over bare metal stents. Restenosis does not cause death. Sudden coronary occlusion causes death. Coronary bypass, revascularizing the distal vessels, protects against coronary occlusion better than PCI, which revascularizes the proximal vessels. And I think that mortality data with drug-eluting stents will be the same as bare metal stents.
(Slide) Data from the New York registry show that in every category of multivessel disease there is an excess 3-year mortality for stents compared to coronary bypass. It varied from 24% in simple two-vessel disease to 46% in complex triple-vessel disease.
My question is this, Peter. With this data, can we continue to passively let the interventionalists be the only physicians to counsel the multivessel disease patients? With the data you presented and the data from New York, is there not now an ethical imperative to confront cardiologists with these data and to educate primary care physicians and our patients?
Thank you.
DR SMITH: Thank you, Dr Guyton. I think the short answer is yes. Obviously there is an ethical dilemma with the cardiologists serving as the primary advisor to our patients with multi-vessel disease and also the provider of the interventional therapy. They have, I think, an obligation to reveal to the patient the mortality disadvantage when it exists so that the patient can make an informed decision related to whether to undergo an operation, which is arguably more painful and costly, as opposed to getting a survival-inferior procedure that may have benefit only at the procedural time. So that is how I would recommend the stance be taken.
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. T. Turer, K. W. Mahaffey, E. Honeycutt, R. H. Tuttle, L. K. Shaw, M. H. Sketch Jr., P. K. Smith, R. M. Califf, and J. H. Alexander Influence of body mass index on the efficacy of revascularization in patients with coronary artery disease. J. Thorac. Cardiovasc. Surg., June 1, 2009; 137(6): 1468 - 1474. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. K. Smith Treatment selection for coronary artery disease: the collision of a belief system with evidence. Ann. Thorac. Surg., May 1, 2009; 87(5): 1328 - 1331. [Full Text] [PDF]
|
|
|
|
|
|
P. K. Smith Treatment selection for coronary artery disease: The collision of a belief system with evidence. J. Thorac. Cardiovasc. Surg., May 1, 2009; 137(5): 1050 - 1053. [Full Text] [PDF]
|
|
|
|
|
|
W. A. Ahmed, P. J. Tully, R. A. Baker, and J. L. Knight Survival After Isolated Coronary Artery Bypass Grafting in Patients With Severe Left Ventricular Dysfunction Ann. Thorac. Surg., April 1, 2009; 87(4): 1106 - 1112. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. J. Shemin Coronary Artery Bypass Grafting Versus Stenting for Unprotected Left Main Coronary Artery Disease: Where Lies the Body of Proof? Circulation, December 2, 2008; 118(23): 2326 - 2329. [Full Text] [PDF]
|
|
|
|
|
|
M. J. Mack, S. L. Prince, M. Herbert, P. P. Brown, M. Katz, G. Palmer, J. R. Edgerton, E. Eichhorn, M. J. Magee, and T. M. Dewey Current Clinical Outcomes of Percutaneous Coronary Intervention and Coronary Artery Bypass Grafting Ann. Thorac. Surg., August 1, 2008; 86(2): 496 - 503. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. B. Ferguson Jr On the Evaluation of Intervention Outcome Risks for Patients With Ischemic Heart Disease Circulation, January 22, 2008; 117(3): 333 - 335. [Full Text] [PDF]
|
|
|
|
 |
|
 D. M. Bravata, A. L. Gienger, K. M. McDonald, V. Sundaram, M. V. Perez, R. Varghese, J. R. Kapoor, R. Ardehali, D. K. Owens, and M. A. Hlatky Systematic Review: The Comparative Effectiveness of Percutaneous Coronary Interventions and Coronary Artery Bypass Graft Surgery Ann Intern Med, November 20, 2007; 147(10): 703 - 716. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Management of Stable Coronary Disease N. Engl. J. Med., October 25, 2007; 357(17): 1762 - 1766. [Full Text] [PDF]
|
|
|
|
 |
|
 R. H. Jones The Year in Cardiovascular Surgery J. Am. Coll. Cardiol., May 8, 2007; 49(18): 1887 - 1898. [Full Text] [PDF]
|
|
|
|
|
|
J. S. Rankin, R. H. Tuttle, A. S. Wechsler, T. L. Teichmann, D. D. Glower, and R. M. Califf Techniques and Benefits of Multiple Internal Mammary Artery Bypass at 20 Years of Follow-Up Ann. Thorac. Surg., March 1, 2007; 83(3): 1008 - 1015. [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 |
