Author + information
- Received July 9, 1998
- Revision received December 3, 1998
- Accepted January 21, 1999
- Published online May 1, 1999.
- Martial G Bourassa, MD, FACCa,* (, )
- Kevin E Kip, PhD†,
- Alice K Jacobs, MD, FACC‡,
- Robert H Jones, MD, FACC§,
- George Sopko, MD∥,
- Allan D Rosen, PhD†,
- Barry L Sharaf, MD, FACC¶,
- Leonard Schwartz, MD, FACC#,
- Bernard R Chaitman, MD, FACC∗∗,
- Edwin L Alderman, MD, FACC††,
- David R Holmes, MD, FACC‡‡,
- Gary S Roubin, MD, PhD, FACC§§,
- Katherine M Detre, MD, DR PH, FACC†,
- Robert L Frye, MD, FACC‡‡,
- for the BARI Investigators
- ↵*Reprint requests and correspondence: Dr. Martial G. Bourassa, Research Center, Montreal Heart Institute, 5000 Bélanger Street East, Montreal, Quebec, H1T 1C8 Canada
Our objective was to determine whether a strategy of intended incomplete percutaneous transluminal coronary angioplasty revascularization (IR) compromises long-term patient outcome.
Complete angioplasty revascularization (CR) is often not planned nor attempted in patients with multivessel coronary disease, and the extent to which this influences outcome is unclear.
Before randomization, in the Bypass Angioplasty Revascularization Investigation, all angiograms were assessed for intended CR or IR via angioplasty. Outcomes were compared among patients with IR intended if assigned to angioplasty, randomized to coronary artery bypass graft surgery (CABG) versus angioplasty; and within angioplasty patients only, among patients with IR versus CR intended.
At 5 years, there was a trend for higher overall (88.6% vs. 84.0%) and cardiac survival (94.5% vs. 92.1%) in CABG versus angioplasty patients with IR intended. The excess mortality in angioplasty patients occurred solely in diabetic subjects; overall and cardiac survival were similar among nondiabetic CABG and angioplasty patients. Freedom from myocardial infarction (MI) at 5 years was higher in nondiabetic CABG versus angioplasty patients (92.4% vs. 85.2%, p = 0.02), yet was similar to the rate observed (85%) in nondiabetic CABG and angioplasty patients with CR intended. Five-year rates of death, cardiac death, repeat revascularization and angina were similar in all angioplasty patients with IR versus CR intended. However, a trend for greater freedom from subsequent CABG was seen in CR patients (70.3% vs. 64.0%, p = 0.08).
Intended incomplete angioplasty revascularization in nondiabetic patients with multivessel disease who are candidates for both angioplasty and CABG does not compromise long-term survival; however, subsequent need for CABG may be increased with this strategy. Whether the risk of long-term MI is also increased remains uncertain.
Most patients undergoing coronary artery bypass graft surgery (CABG) receive complete or nearly complete revascularization, and this result appears to positively influence long-term prognosis (1–7). In contrast, complete revascularization may be planned and attempted in 50% or less of patients treated with percutaneous transluminal coronary angioplasty, and the extent to which lack of complete initial revascularization influences outcome is less clear (8–17). Late survival and survival free of myocardial infarction (MI) appear similar in patients with and without complete revascularization after angioplasty (8–10). However, the need for subsequent CABG is usually much higher in patients with incomplete compared with those with complete revascularization after angioplasty (11–18). This higher rate of CABG after incomplete angioplasty revascularization may be the result of lack of both initial procedural success and of long-term symptom relief.
The Bypass Angioplasty Revascularization Investigation (BARI) tested the hypothesis that an initial use of angioplasty compared with CABG in patients with multivessel disease and severe angina or ischemia requiring revascularization which could be approached by either procedure did not compromise survival during a five-year period (19,20). Angiographic eligibility required consent of both the angioplasty operator and the surgeon, but expectation of complete revascularization by either procedure was not a requirement for inclusion in BARI. Overall, five-year survival did not differ significantly between the two treatments despite the fact that 91% of significant lesions were bypassed in the CABG patients compared with only 54% of clinically important lesions successfully dilated in the angioplasty patients (20).
As part of the screening process in BARI, an angiographic assessment was made to determine the suitability and operator willingness to attempt each ≥50% lesion with angioplasty revascularization. The same determination was also made to determine lesion suitability for CABG, although this is not the subject of the present report. Thus, before patient randomization, a determination of intended complete (CR) or intended incomplete (IR) angioplasty revascularization was made such that the two revascularization assignments (CABG and angioplasty) can be compared without bias within the IR intended and within the CR intended groups. In addition, the effect of IR intended on end points such as angina and need for repeat revascularization can be studied within angioplasty patients only, but not without bias, because the grouping of these patients reflects the selection made by the investigator. The purpose of this report was to assess whether angioplasty compromises outcome in patients where it cannot be expected to achieve complete revascularization, and whether such patients should always undergo CABG as their initial procedure.
Patients were eligible for BARI if they had clinically severe angina or objective evidence of ischemia requiring revascularization, multivessel coronary disease suitable for both angioplasty and CABG and informed consent for random assignment (19,20). Between August 1988 and August 1991, 1,829 patients were enrolled at 18 centers, 16 in the United States and two in Canada, of whom 914 were randomly assigned to undergo and 892 received CABG, and 915 were randomized to undergo angioplasty and 904 received this treatment. No significant difference was observed between the two groups in the primary end point of mortality at five years, 89.3% for patients assigned to CABG versus 86.3% for those assigned to angioplasty, p = 0.19 (20).
Figure 1shows the a priori (e.g., before randomization) determination of intended completeness of revascularization (if assigned to angioplasty) in the BARI population. Incomplete revascularization was planned in 624 patients (34.1%), CR in 1,196 (65.4%) and intended completeness of revascularization was not determined in 9 (0.5%). After random assignment, 612 patients with planned CR versus 301 with planned IR were assigned to and 600 versus 291 received CABG, whereas 584 versus 323 were assigned to and 579 versus 317 received angioplasty. Among IR patients, roughly one half had one or more lesions considered by the angioplasty operator to be clinically important but not intended for angioplasty; in the other half, one or more borderline lesions were not intended for angioplasty.
Bypass angioplasty revascularization investigation angiographic definitions
Angiographically significant lesions were defined as ≥50% stenosis in a vessel ≥1.5 mm as measured by electronic calipers (21). The number of diseased vessels was defined as the number of the three major coronary perfusion territories (anterior, lateral and inferoposterior) supplied by a vessel with an angiographically significant lesion. To estimate the amount of myocardium at risk, a global percent jeopardy index was calculated as the ratio of left ventricular territory subtended by terminal coronary segments compromised by significant lesions to the sum of all left ventricular territory supplied by major terminal coronary branches (21). Lesion complexity was categorized as type A, B and C using American Heart Association/American College of Cardiology consensus panel criteria (22). Flow distal to each stenosis was categorized by Thrombolysis in Myocardial Infarction flow criteria (23). A lesion was considered successfully dilated if there was a >20% reduction in stenosis with a residual stenosis <50% and TIMI flow grade 3.
Consecutive patients were screened for clinical eligibility and presence of multivessel disease at the time of angiography, and the angiograms were reviewed to determine suitability for both angioplasty and bypass surgery. Although the protocol called for anticipation of successful relief of the major areas of ischemia, the potential for complete revascularization was not a requirement. The BARI investigators recognized the limitations of trying to describe prospectively all the features that define technical suitability for angioplasty and CABG. Therefore, the final determination of eligibility involved the subjective judgment of different angioplasty operators and cardiac surgeons who were all highly experienced and BARI certified (19,24). Specific reasons for anticipated lack of efficacy or safety were recorded.
Before randomization, clinical site investigators specified the extent and severity of coronary disease and which coronary lesions would be selected for revascularization (if assigned to angioplasty). Lesion complexity and clinical relevance were also recorded. The angiograms of enrolled patients were also interpreted by a Central Radiographic Laboratory except for the clinical relevance and appropriateness of lesions for revascularization. Consequently, the results of the Central Radiographic Laboratory interpretation were not used in this report.
Details of the angioplasty and bypass surgery procedures and follow-up clinical events including procedural and long-term mortality rates, recurrence of angina and repeat revascularization were recorded on standardized forms. Rest electrocardiograms were routinely collected at study entry, before and after all coronary revascularization procedures, at scheduled follow-up and for all suspected myocardial infarct events. A central electrocardiographic laboratory coded all Q-wave events blinded by initial treatment assignment (25). According to protocol, cardiac enzymes were not used to define myocardial infarction within 96 h of a revascularization procedure (19,25). Cause of death was classified by an independent Mortality and Morbidity Classification Committee. Cardiac death was defined as: death less than 1 h after onset of cardiac symptoms, or within 1 h to 30 days after a documented or probable myocardial infarction, or death from intractable congestive heart failure, cardiogenic shock or other documented cardiac causes.
Assessment of lesion clinical relevance and intention for angioplasty
According to the clinical judgment of the angioplasty operator and surgeon, each significant lesion was rated before randomization as clinically important, borderline or not relevant (small territory or nonviable myocardium). A lesion was rated as clinically important if it was felt to be responsible for or to contribute significantly to the patient’s ischemic syndrome and if the myocardial territory distal to the lesion was viable and large enough to warrant revascularization. A borderline lesion was a lesion for which revascularization was not required or necessary but was often performed incidentally during treatment of significant lesions (26). Diffuse lesions, which constituted 4% of all significant lesions, were generally not rated a priori for clinical relevance, and were considered to be clinically important in this report. Only approximately 2% of all significant lesions were total occlusions supplying a small territory/nonviable myocardium, and they were excluded from the analysis. Intended completeness of revascularization was determined on the basis of all lesions considered clinically important or borderline and whether or not these lesions were declared as both suitable and intended to be dilated with angioplasty (26).
This determination yielded 624 patients with ≥1 clinically important or borderline lesions considered not intended for angioplasty and hence representing a strategy of IR intended (Fig. 1). For subanalyses, patients were further categorized in a nonrandomized fashion, according to whether clinically important lesions were intended or not intended for angioplasty.
Groups of patients were compared in two ways. First, among patients with IR intended if assigned to angioplasty, those randomized to CABG versus angioplasty were compared. Second, among angioplasty patients only, those with a strategy of CR intended versus IR intended were compared (nonrandomized comparison). The second nonrandomized comparison was performed because some end points (e.g., angina, repeat revascularization) may be most informative when compared among angioplasty patients only. Differences in proportions of baseline clinical and angiographic characteristics, and the prevalence of angina during follow-up were assessed by use of chi-square tests. Similarly, differences in continuous measures of these factors were assessed by Student ttests. The Kaplan–Meier estimate (27)was used to calculate five-year freedom from untoward event rates which included total mortality, cardiac mortality, MI, repeat CABG, repeat angioplasty, and any repeat revascularization. These estimates were restricted to patients who received their assigned treatment, and were compared by the use of log-rank statistic.
Power calculations were made assuming a one-sided hypothesis with a type I error of 0.05. We postulated that: a) CABG patients have a greater freedom from events than angioplasty patients with IR intended and b) angioplasty patients with CR intended have a greater freedom from events than angioplasty patients with IR intended. Risk ratios and percent power were calculated given the fixed sample size and actual differences in event rates. Next, we calculated the between-group differences and risk ratios that would be necessary to achieve 80% power given our sample size and the event rates observed in the two lower risk reference groups.
Baseline clinical and angiographic characteristics
Table 1presents the baseline clinical and angiographic characteristics of the study patients. Overall, patients with IR intended if assigned to angioplasty had a worse baseline profile than patients with CR intended. Clinically, they had a significantly higher prevalence of prior MI, congestive heart failure, treated diabetes and peripheral vascular disease. However, among patients with IR intended, those randomized to CABG versus angioplasty were similar on all clinical characteristics.
Compared with patients with CR intended if assigned to angioplasty, patients with IR intended had more triple vessel disease, clinically important lesions, total occlusions and class C lesions. They also had a lower ejection fraction, a higher wall motion score and a higher percentage of jeopardized myocardium. Among the 624 patients with IR intended, baseline angiographic characteristics were similar between patients randomized to CABG versus those randomized to angioplasty. However, despite the protection of randomization and similarity of clinical and angiographic characteristics, CABG patients had a higher proportion of clinically important lesions not intended compared with angioplasty patients.
Initial procedural outcome in patients who received their assigned treatment
Table 2presents the initial procedural outcome of the 608 patients with IR intended if assigned to angioplasty who received their assigned treatment, and the 579 angioplasty patients with CR intended. Coronary artery bypass graft surgery patients seemed to be nearly completely revascularized with an average of 2.9 grafts and 3.3 distal anastomoses per patient, although 13.2% had at least one diseased vessel after the procedure. By comparison, a substantial proportion of angioplasty patients with both CR and IR intended were incompletely revascularized at the initial angioplasty. This occurred because only about half of all lesions initially intended for angioplasty were both attempted and successfully dilated. However, there was a clear and significant separation in the extent of revascularization achieved (as determined from the mean percent of myocardium jeopardized after angioplasty) between angioplasty patients with CR and those with IR intended.
Despite the differences in the initial extent of revascularization achieved, the incidence of in-hospital death and death/Q-wave MI did not differ significantly between angioplasty patients with IR intended and both CABG patients with IR intended and angioplasty patients with CR intended. Similarly, among angioplasty patients only, the incidence of abrupt closure, emergent and nonemergent CABG and repeat angioplasty was similar between the CR and IR groups.
Freedom from untoward events—randomized comparison of CABG and angioplasty patients with IR intended if assigned to angioplasty
Table 3presents freedom from untoward event rates at five years for the randomized treatment comparison of patients with IR intended if assigned to angioplasty (CABG vs. angioplasty). There was a nonsignificant trend for CABG patients to have higher five-year survival and cardiac survival than angioplasty patients. However, almost all of the excess mortality occurred in the 142 patients with treated diabetes, with identical five-year cardiac survival (95.7%) seen between 466 nondiabetic CABG and angioplasty patients. Five-year freedom from MI was higher in all as well as nondiabetic CABG versus angioplasty patients. When the analysis was restricted to patients with class C or borderline lesions, the five-year results were similar to those of the entire cohort of patients with IR intended. Not surprisingly, the need for repeat revascularization was substantially greater in angioplasty versus CABG patients. In addition, CABG patients were more likely to be free of angina at five years.
As shown in Table 4, our study was underpowered to detect statistically significant mortality and especially cardiac mortality differences among CABG versus angioplasty patients with IR intended. On the other hand, the actual versus required risk ratios for MI and death/MI suggest that the study had adequate power to detect a clinically meaningful difference for these end points.
Freedom from untoward events at five years—nonrandomized comparison among angioplasty patients (CR vs. IR intended)
Table 5presents freedom from untoward event rates at 5 years for the nonrandomized comparison of angioplasty patients only (CR vs. IR intended). Despite an overall worse baseline clinical and angiographic profile in the IR group, freedom from death, cardiac death, MI and repeat revascularization (CABG/angioplasty) did not differ significantly between all as well as nondiabetic angioplasty patients in whom incomplete versus complete revascularization was intended. In addition, the prevalence of angina at five years did not differ between angioplasty patients with planned IR versus CR. Similarly, the prevalence and severity of angina were similar between angioplasty patients with planned IR versus CR throughout five-year follow-up (Fig. 2). Finally, there was a nonsignificant trend for more CABG during follow-up in angioplasty patients with planned IR versus those with planned CR.
Given the sample size of 579 patients in the reference group (angioplasty patients with CR intended), the study has sufficient power to detect moderate differences in clinical outcome among angioplasty patients with CR versus IR intended. In particular, the sample size was sufficiently large to detect a moderate excess of CABG at five years in the group with IR intended (Table 6).
In patients with multivessel coronary artery disease who are referred for revascularization and who are suitable for both angioplasty and CABG, the choice of procedure often presents a clinical dilemma. Recent randomized clinical trials have consistently shown that, overall, midterm survival (between one and five years) and survival without myocardial infarction do not differ significantly between the two treatment strategies, but that an initial angioplasty strategy is usually associated with a greater incidence of repeat revascularization and more residual angina during follow-up. Some trials have included only patients who could be completely revascularized by angioplasty (28,29), whereas others enrolled patients in whom incomplete revascularization was expected (30–32). In BARI, incomplete revascularization was permitted although the protocol required relief of the major ischemic areas. The BARI design was unique in that the operators had to declare a priori (e.g., before randomization) their intention to treat or not to treat by angioplasty each significant lesion in a given patient and also to rate each lesion (whether intended or not) according to its clinical importance. Thus, the BARI design allowed for an unbiased prospective evaluation of the outcome of angioplasty patients with intended incomplete revascularization versus randomly matched CABG patients, and a nonrandomized comparison of outcomes between angioplasty patients with intended incomplete versus complete revascularization.
The group with intended incomplete angioplasty revascularization amounted to one third of all BARI patients. Clearly, this group had more clinical risk factors, more severe coronary artery disease, (particularly triple vessel disease with chronic total occlusions or other class C lesions) and more significant left ventricular dysfunction than the group in whom complete revascularization was intended. In addition, nearly half of these patients had lesions considered clinically important which were not intended for angioplasty. We postulated that in patients with multivessel disease in whom incomplete angioplasty revascularization was intended, an initial strategy of CABG might result in a better five-year survival than an initial strategy of angioplasty. Even among patients with intended complete angioplasty revascularization, not all intended lesions are actually attempted, and not all attempted lesions are successfully dilated. Therefore, what is achieved at angioplasty is usually less than what was intended initially, and this is true equally for patients with intended complete or incomplete revascularization. The aim of this study was to examine the influence of an intended incomplete angioplasty revascularization strategy on long-term outcome, not taking into account the degree of revascularization actually achieved as reported in several recent studies (6–18).
Freedom from untoward events (death, MI, angina) in relation to intended completeness of revascularization (if randomized to angioplasty)
After accounting for the excess mortality previously noted in randomized angioplasty patients with diabetes, five-year total mortality and cardiac-only mortality did not differ significantly between angioplasty, and CABG patients with intended incomplete angioplasty revascularization. This comparison held true for the subset of 311 patients with clinically important lesions not intended for dilation. These findings suggest that in nondiabetic patients with clinical and angiographic characteristics similar to those seen in BARI patients, an initial strategy of intended incomplete angioplasty revascularization probably does not increase the risk of five-year mortality compared with an initial strategy of CABG. These results are in agreement with previous reports which suggest that midterm survival is similar with and without complete revascularization after angioplasty (8–10).
Five-year freedom from MI was significantly less in nondiabetic angioplasty patients with intended incomplete revascularization than in similar randomly assigned CABG patients (85.2% vs. 92.4%). Although our study had adequate power to detect a clinically meaningful difference between treatment groups for freedom from MI and freedom from death/MI, the interpretation of these findings remains unclear given that the five-year freedom from MI was 85.1% and 84.7% in the groups of nondiabetic patients with complete revascularization intended (angioplasty and CABG, respectively). Because CABG and angioplasty patients with complete revascularization intended had, on average, a more favorable baseline profile than patients with intended incomplete revascularization, it would seem that the rate of freedom from MI of 92.4% at 5 years in nondiabetic CABG patients with intended incomplete revascularization if assigned to angioplasty is unexpectedly high. In the BARI trial, non–Q-wave MI was not diagnosed within 96 h of either angioplasty or CABG because of the difficulty of interpreting the clinical significance of cardiac enzyme elevations after bypass surgery. Whether this might have influenced long-term infarction rates is unknown, but appears unlikely. Moreover, the finding of identical rates of late cardiac deaths but higher rates of nonfatal MI after angioplasty than CABG in nondiabetic patients, if confirmed, would be hard to explain. The finding that freedom from MI at five years is greater in CABG patients with intended incomplete than in those with intended complete revascularization if assigned to angioplasty was also unexpected. A possible explanation, however, could be that surgical bypass of lesions considered to be of borderline clinical importance often leads to occlusion of both the graft and the native vessel with a poorer outcome than if the vessels had not been attempted. Finally, previous studies have shown that midterm Q wave MI rates are similar in patients with multivessel disease in whom complete or incomplete revascularization is achieved after angioplasty (8–10,18). Thus, whether intended incomplete angioplasty revascularization adversely influences the long-term risk of MI remains unclear at this time.
The prevalence of angina during follow-up was significantly higher in angioplasty patients with intended incomplete revascularization compared with randomly assigned CABG patients. However, the excess angina prevalent in angioplasty patients appeared unrelated to whether or not complete revascularization was intended. Thus, the specific angiographic characteristics and clinical judgment which determine whether an operator intends to attempt complete revascularization do not seem to influence the occurrence of angina after angioplasty revascularization.
Freedom from repeat revascularization
Despite similarity in the need for repeat revascularization (CABG or angioplasty), a trend for more follow-up CABG was seen in angioplasty patients with incomplete compared with angioplasty patients with complete revascularization intended. Our sample size was sufficiently large to detect a moderate excess of CABG at five years in the group with IR intended. Given these results and the results of previous studies (11–18), a strategy of intended incomplete angioplasty revascularization seems to increase the likelihood of a patient requiring subsequent CABG (e.g., to alleviate symptoms of ischemia). However, the apparent excess CABG seen in angioplasty patients with intended incomplete revascularization did not adjust for adverse clinical and angiographic characteristics which were more common in these patients. The relatively modest difference in CABG in our study in comparison with previous reports (14,15)may reflect the a priori declaration of intended completeness of revascularization, as opposed to studies of achieved completeness of revascularization where, by definition, all patients with failed angioplasty, are considered incompletely revascularized and probably more likely to undergo CABG.
Patients were enrolled in BARI between August 1988 and August 1991. Because of the uncertainty then about the outcome of procedures involving new technologies such as stents, atherectomy and laser angioplasty, these new devices were prohibited as part of the initial angioplasty strategy. Thus, our data are particularly relevant to the large number of patients with multivessel coronary disease still managed exclusively with balloon angioplasty. Ongoing studies will hopefully determine how the addition of stents and other devices to conventional angioplasty influence the results of interventional cardiology as compared with current surgical techniques. In this study, we assessed intended incomplete anatomic revascularization via angioplasty. It has been suggested that incomplete anatomic but complete functional revascularization may result in outcomes similar to those of complete anatomic revascularization (33). However, when the analysis of freedom from death and freedom from MI was restricted to patients with intended IR due to class C or to borderline lesions in this study, outcomes were similar to those of the entire cohort of patients with IR intended.
Five-year total and cardiac survival did not differ between nondiabetic angioplasty patients with intended incomplete revascularization and randomly assigned CABG patients. Although freedom from MI at five years was significantly higher in nondiabetic CABG versus angioplasty patients with intended incomplete revascularization (92.4% vs. 85.2%), whether this finding is genuine remains unclear in light of the five-year MI-free rate of approximately 85% observed in patients with complete revascularization intended. Whereas intended incomplete revascularization seems to increase the risk of subsequent CABG in patients undergoing initial angioplasty, this strategy appears to be unrelated to the five-year risk of repeat angioplasty, any revascularization (CABG/angioplasty) or angina.
Taken together, our results suggest that experienced angioplasty operators are able to determine which clinically important lesions should and should not be intended and attempted in patients with multivessel disease, without compromising long-term patient survival.
☆ BARI is supported by grants (HL38493, HL38504, HL38509, HL38512, HL38514-6, HL38518, HL38524-5, HL38529, HL38532, HL38556, HL38610, HL38642 and HL42145) from the National, Heart, Lung and Blood Institute, Bethesda, Maryland.
- Bypass Angioplasty Revascularization Investigation
- coronary artery bypass graft surgery
- complete revascularization
- incomplete revascularization
- myocardial infarction
- Received July 9, 1998.
- Revision received December 3, 1998.
- Accepted January 21, 1999.
- American College of Cardiology
- Schaff H.V,
- Gersh B.J,
- Pluth J.R,
- et al.
- Bell M.R,
- Gersh B.J,
- Schaff H.V,
- et al.
- Bell M.R,
- Bailey K.R,
- Reeder G.S,
- Lapeyre A.C,
- Holmes D.R
- Reeder G.S,
- Holmes D.R,
- Detre K.M,
- Costigan T,
- Kelsey S.F
- Vandormael M.G,
- Chaitman B.R,
- Ischinger T,
- et al.
- Mabin T.A,
- Holmes D.R,
- Smith H.C,
- et al.
- Cowley M.J,
- Vandormael M,
- Topol E.J,
- et al.
- investigators of the NHLBI PTCA Registry,
- Bourassa M.G,
- Yeh W,
- Holubkov R,
- Sopko G,
- Detre K.M
- Bypass Angioplasty Revascularization Investigation (BARI) Investigators
- Alderman E.L,
- Stadius M
- Ryan T.J,
- Faxon D.P,
- Gunnar R.M,
- et al.
- Sheehan F.H,
- Braunwald E,
- Canner P,
- et al.
- Chaitman B.R,
- Rosen A.D,
- Williams D.O,
- et al.
- Botas J,
- Stadius M.L,
- Bourassa M.G,
- et al.
- ERACI Group,
- Rodriguez A,
- Boullon F,
- Perez-Balino N,
- Paviotti C,
- Sosa Liprandi M.I,
- Palacios I.F
- German Angioplasty Bypass Surgery Investigation (GABI),
- Hamm C.W,
- Reimers J,
- Ischinger T,
- Rupprecht H.J,
- Berger J,
- Bleifeld W
- Multivessel Angioplasty Prognosis Study (MAPS) Group,
- Cowley M.J,
- Vandermael M,
- Topol E.J,
- et al.