DOI: 10.2337/dc06-0118 © 2006 by the American Diabetes Association
Diabetes Does Not Influence Treatment Decisions Regarding Revascularization in Patients With Stable Coronary Artery Disease
1 Department of Cardiology, Isala Klinieken, Zwolle, the Netherlands Address correspondence and reprint requests to Eric Boersma, PhD, Clinical Epidemiology Unit, Thoraxcenter Cardiology, Rm. Ba-563, Erasmus MC, Dr Molewaterplein 40, 3015 GD Rotterdam, Netherlands. E-mail: h.boersma{at}erasmusmc.nl
OBJECTIVETo evaluate whether in stable angina preference for coronary revascularization by either percutaneous coronary intervention (PCI) or coronary artery bypass surgery (CABG) is influenced by diabetes status and whether this has prognostic implications. RESEARCH DESIGN AND METHODSA total of 2,928 consecutive patients with stable angina who were enrolled in the prospective Euro Heart Survey on Coronary Revascularization were studied. Multivariable analyses were applied to evaluate the relation between diabetes, treatment decision, and 1-year outcome. RESULTSDiabetes was documented in 587 patients (20%) who had more extensive coronary disease. Revascularization was intended in 74% of patients with diabetes and in 77% of those without diabetes. In patients selected for revascularization, CABG was intended in 35% of diabetic and in 33% of nondiabetic patients. Multivariable analyses did not change these findings, but in some subgroups diabetes influenced treatment decisions. For example, diabetic subjects with mild heart failure had more often intended revascularization (91%) than those without diabetes (67%, P < 0.001). Treatment decisions in patients with more extensive (left main, multivessel, or proximal left anterior descending artery) disease were not influenced by diabetes status. Diabetes was not associated with an increased incidence of all-cause death, nonfatal cerebrovascular accident, or nonfatal myocardial infarction at 1 year, regardless of preferred treatment. The incidence of the combined end points was 7.3% in diabetic and 6.8% in nondiabetic patients (adjusted hazard ratio 1.0 [95% CI 0.71.4]). CONCLUSIONSIn stable angina, treatment decisions regarding revascularization or the choice for CABG or PCI were not influenced by the presence of diabetes. Diabetes was not associated with a poor prognosis.
Abbreviations: BARI, Bypass Angioplasty Revascularization Investigation CABG, coronary artery bypass surgery CAD, coronary artery disease CVA, cerebrovascular accident DES, drug-eluting stent EHS-CR, Euro Heart Survey on Coronary Revascularization LAD, left anterior descending artery MI, myocardial infarction NYHA, New York Heart Association PCI, percutaneous coronary intervention
Prevalence of diabetes in the general population is high, particularly in North America (7.9%) and Europe (7.8%) (1). In patients with established coronary artery disease (CAD) the prevalence of diabetes is even higher. In the Euro Heart Survey on Diabetes and the Heart, 14% of patients with stable CAD had newly detected diabetes, whereas 37% had impaired glucose regulation. Furthermore, patients with CAD and diabetes may have a worse prognosis (2,3). Whether coronary revascularization in patients with diabetes has comparable benefits as in patients without diabetes is not yet clear. There is also debate whether coronary artery bypass surgery (CABG) or percutaneous coronary intervention (PCI) should be preferred in diabetes (46). As a consequence, current treatment guidelines do not provide firm treatment advice for the general diabetic CAD population, although some detailed recommendations are given for specific subgroups (714). The Euro Heart Survey on Coronary Revascularization (EHS-CR) was developed to obtain quantitative information on the adherence to guidelines and prognosis in patients undergoing coronary angiography (15). The EHS-CR enrolled 3,006 consecutive patients with stable CAD, and this well-characterized study population provides a unique opportunity for a systematic analysis of the relation between patient characteristics (including diabetes), invasive treatment choices, and prognosis.
EHS-CR The EHS-CR was described in detail elsewhere (15). Briefly, the survey was designed to screen consecutive patients undergoing invasive procedures in the catheterization laboratory. Patients were enrolled if they had a diameter stenosis of at least 50% in at least one major epicardial coronary artery. Data were collected by dedicated data collecting officers and sent to a central database in the European Heart House (Sophia Antipolis, Valbonne, France) via the Internet using the MacroTM software (InferMed, London, U.K.). The collected data included demographics, comorbidity, diagnosis, and detailed information regarding diagnostic angiography and treatment modalities. Between 1 November 2001 and 1 March 2002, a total of 5,767 patients were enrolled.
Treatment decisions
Definitions
Follow-up
Data description and data analysis A number of multivariable logistic regression analyses were applied to further evaluate the relation between diabetes status and intended treatment (medical versus revascularization; CABG versus PCI in patients with intended revascularization) and the extent to which this relation was influenced by a range of clinical and angiographic characteristics. Therefore, each separate regression model included a diabetes status interaction term. If there was statistical evidence that this interaction term contributed to the model, it was concluded that the relation between diabetes and intended treatment was influenced by that specific characteristic. As suggested by previous trials (13), we studied the relation between diabetes status and intended treatment in three specific patient subgroups in more detail: patients with significant left main disease; patients with three-vessel disease, combined with those with two-vessel disease and a significant lesion in the proximal left anterior descending artery (LAD); and patients with two-vessel disease not involving the proximal LAD, combined with those with one-vessel disease. Kaplan-Meier survival analyses, as well as univariable and multivariable Cox proportional hazard regression, were applied to study patient outcome at 1-year follow-up. Since the number of outcome events was limited, in multivariable analysis we only adjusted for the EuroSCORE, a scoring system to estimate the operative mortality for patients undergoing cardiac surgery based on a broad range of clinical characteristics (16). We report adjusted hazard ratios and corresponding 95% CIs. All statistical tests were two sided, and significance was stated at the classical 0.05 probability level.
Patient characteristics A total of 3,006 patients were diagnosed with stable angina. Complete data on diabetes status and intended treatment was available in 2,928 (97%) patients who comprised our study population. A total of 587 (20%) patients had diabetes. There were important differences in clinical and angiographic baseline characteristics between patients with and without diabetes (Table 1). Patients with diabetes were more often women and had a higher prevalence of chronic renal insufficiency, peripheral vascular disease, and cerebrovascular disease. The size of the myocardium at risk was significantly increased in diabetic patients, as was the EuroSCORE and the number of diseased vessels and segments. Moreover, patients with diabetes more often had impaired left ventricular function than those without diabetes.
Univariable analyses In patients with diabetes, a PCI was intended in 282 (48%) patients and CABG in 155 (26%) patients, compared with 1,217 (52%) and 586 (25%) in patients without diabetes, respectively. There were, however, some specific patient subgroups in which the presence of diabetes seemed to influence choice of treatment (Table 2). The presence of diabetes was associated with an increased physicians preference for medical treatment in women, in patients without heart failure, in those with mitral valve insufficiency, and in patients with four or more diseased segments. In contrast, diabetes was associated with an increased preference for revascularization in patients with previous PCI and those with mild heart failure. In the diabetic patients, an increased physicians preference for CABG rather than PCI was observed in patients aged <60 years, in women, in patients with previous PCI or extensive antianginal medication, and in those with a EuroSCORE of less than three points. An increased physicians preference for intervention by PCI rather than CABG was observed in diabetic patients with a large area of jeopardized myocardium.
The extent of CAD was strongly associated with the physicians preference for revascularization (Table 2). Diabetes did not influence this association. Among patients selected for revascularization, CABG was preferred in those with more extensive disease, independent of the coexistence of diabetes.
Multivariable analyses
Outcome after 1 year The incidence of the composite end point of all-cause death, nonfatal CVA, or nonfatal MI at 1-year follow-up was 7.3% in patients with diabetes and 6.8% in patients without diabetes (adjusted hazard ratio 1.0 [95% CI 0.71.4]). Patients with more extensive disease had a higher incidence of death, CVA, or MI at 1-year follow-up than those with less extensive disease (Fig. 1; Table 3). However, importantly, in the subgroups according to the extent of CAD, there were no significant differences in the incidence of this composite end point between patients with and without diabetes.
Concerning the relation among diabetes status, treatment choice, and outcome, because of small numbers, patients with left main disease or three-vessel disease were considered as one group. Regardless of the extent of the disease, patients selected for PCI had a lower incidence of major adverse cardiovascular or cerebrovascular events at 1-year follow-up than those selected for medical treatment or CABG (Table 4), but again we observed no difference in the relation between intended treatment and outcome according to diabetes status.
This analysis revealed that in stable angina, subsequent treatment decisions regarding revascularization (and the choice for either CABG or PCI) were not influenced by the presence of diabetes. Importantly, diabetic patients with left main disease, proximal LAD disease, or more extensive, multivessel disease were not more likely to undergo coronary revascularization (or CABG) than their nondiabetic counterparts. Diabetes was also not associated with a poor prognosis. Previous (1999) and current (2002) guidelines recommend CABG rather than PCI in patients with diabetes and multivessel disease (10,11). This treatment advice is mainly based on post hoc analyses from randomized trials that were conducted before the use of stents and glycoprotein IIb/IIIa inhibitors. In the Bypass Angioplasty Revascularization Investigation (BARI), patients with treated diabetes who were randomized to CABG had significantly better survival after 7-year follow-up than those randomized to balloon angioplasty (4). To some extent, this finding was confirmed by a meta-analysis of 13 randomized trials of CABG versus balloon angioplasty (including BARI), which demonstrated improved survival in favor of CABG at 4-year follow-up but no longer at 6.5-year follow-up (17). How can we understand the discrepancy between guideline recommendations and the clinical practice patterns that we observed? Possibly, those involved in the care of patients with stable angina are not convinced that patients with diabetes should be treated differently than those without diabetes, especially since large-scale randomized trials are lacking. The ongoing BARI 2 Diabetes, in which diabetic patients will be randomly allocated to aggressive medical management targeting at optimal glycemic and metabolic control or revascularization, may help to solve this "burgeoning dilemma," as the investigators call it (18). Furthermore, it is well known that patients enrolled in clinical trials form a selected population, particularly in randomized trials comparing PCI and CABG (19). In this respect, observational studies, including the BARI registry (5), reported similar outcomes after CABG and PCI in patients with diabetes and symptomatic coronary disease (6). Another interesting aspect is emphasized by McGuire et al. (20), who evaluated the effects of reporting the BARI results (which were made public together with a "clinical alert") on decision making. It was reported that the rapid advancement of health care technology was the major factor of the lack of influence of the clinical alert and the BARI on current clinical practice. In recent (19962000) randomized trials comparing CABG against PCI, patients with diabetes who were allocated to stenting with liberal use of glycoprotein IIb/IIIa inhibitors had much more favorable outcomes than the patients allocated to balloon angioplasty in the earlier (19851995) trials (21). Furthermore, no difference in irreversible clinical end points was observed between CABG and PCI with stenting. It is true that these trials are limited by a relatively small sample size and a short (1-year) follow-up duration. Still, these observations may have influenced treatment decisions. In our survey, diabetes was not associated with an increased incidence of major adverse cardiac and cerebral events. Diabetic patients scheduled for PCI had similar prognosis as nondiabetic subjects. It is true that the limited number of patients and the short duration of follow-up might have masked significant and clinically relevant differences. Furthermore, we realize that this is a purely observational study, and patients were not randomized to the different treatment regimens. Consequently, estimates of treatment effect might be seriously biased due to differential patient selection. Still, based on these observations, we are confident that large differences in clinical outcome can be excluded with sufficient certainty. Hence, the position of PCI with stenting as a safe treatment for patients with diabetes and stable CAD seems justified.
Limitations Second, as the EHS-CR is a survey of routine clinical practice, the treating physician was not enforced to use specific laboratory tests in order to establish the diagnosis of diabetes. Thus, we cannot exclude the possibility that misclassification of diabetes had occurred. However, the clinical and angiographic characteristics of patients with diabetes corresponded quite well with other datasets of diabetic subjects with stable coronary disease (23). Third, no core lab analysis was performed of the qualifying coronary angiogram. Consequently, detailed anatomic information that may have influenced treatment decisions was lacking. This is especially relevant for the 366 patients with a prior history of CABG. However, sensitivity analyses that excluded these patients showed consistent results. Fourth, the present data have been acquired before clinical availability of antirestenotic, drug-eluting stents (DESs). Randomized clinical trials comparing DESs with bare metal stents demonstrated lower repeat revascularization rates in favor of DESs (2426). However, these trials did not provide evidence for a reduction in irreversible adverse cardiac events, including death and MI. Since the recommendation to conduct CABG instead of PCI is mainly based on the long-term results of the BARI, demonstrating a mortality reduction in favor of CABG (4) based on currently available evidence, it seems unlikely that the physicians preference for CABG or stenting will be different in the DES than in the bare metal stents era. At the other hand, the ongoing CARDia, FREEDOM, and BARI 2D trials, which enroll patients with coronary disease and diabetes who are randomized to either CABG or PCI with DES and modern antiplatelet therapy, including glycoprotein IIb/IIIa inhibitors and clopidogrel, may shed a vivid light on the PCI-versus-CABG debate. New surveys of clinical practice are warranted after the results of these trials are available. Diabetes is not among the factors that determine treatment decisions regarding revascularization in patients with stable coronary disease. Diabetes was not associated with a worse prognosis, independent of invasive treatment preference. In expectation of the results of ongoing clinical trials comparing several medical and more invasive treatment strategies, guidelines for the management of CAD patients with diabetes should be updated more systematically than is currently the case.
Organization of the survey Survey Expert Committee. W. Wijns (survey chairman), Belgium; N. Mercado (research fellow), the Netherlands; M. Bertrand, France; W. Maier, Switzerland; B. Meier, Switzerland; C. Moris, Spain; F. Piscione, Italy; U. Sechtem, Germany; P. Sergeant, Belgium; E. Stahle, Sweden; J. Vos, the Netherlands; P. Widimsky, Czech Republic; F. Unger, Austria.
Euro Heart Survey Team (European Heart House, France).
Industry sponsor.
Sponsoring institutions.
* Members of the Euro Heart Survey on Coronary Revascularization can be found in the APPENDIX. A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C Section 1734 solely to indicate this fact. Received for publication January 17, 2006. Accepted for publication May 8, 2006.
This article has been cited by other articles:
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||