Dr. Best received her medical degree from the Medical
College of Pennsylvania and is finishing her cardiovascular and
interventional training at the Mayo Clinic.
Dr. Berger is a Professor of Medicine and an
Interventionalist at the Mayo Clinic.
Randomized clinical trials per-formed in the 1970s and early
1980s established coronary artery bypass grafting (CABG) as the
primary therapy for multivessel coronary artery disease. Although
some of the studies revealed disparate results, differences in the
baseline patient characteristics contributed to these discrepancies
(Table 1 and figure 1).
Of the three major studies, the Coronary Artery Surgery Study
(CASS) had the lowest risk population, with more than one-fourth
having single-vessel disease.
Since patients with single-vessel disease have a higher mortality
with surgery compared with medical therapy, it is not surprising
that there was no difference demonstrated in overall survival.
The Veterans Administration Coronary Artery Bypass Surgery Study
(VA) had a higher risk population because of the inclusion
requirement of an abnormal resting electrocardiogram suggestive of
In this study, CABG increased survival at 7 years, but this
advantage was lost at 11 years. When the highest-risk patients,
such as those with left-main disease or three-vessel disease with a
reduced left ventricular ejection fraction (EF), were analyzed, a
distinct survival advantage with CABG was seen and this was
maintained throughout the follow-up period.
The third major trial, the European Coronary Surgery Study
(ECSS), required at least two-vessel coronary disease for study
This study also demonstrated a significant survival advantage with
CABG at 5 years, and although the benefit diminished over the
12-year follow-up, it remained clinically and statistically
significant. Patient subgroups that derived the greatest benefit
from CABG were older patients with abnormal electrocardiograms and
positive treadmill exercise tests, those patients with peripheral
arterial disease, and those with a proximal left anterior
descending (LAD) coronary artery lesion.
In a meta-analysis of these three studies and four smaller
studies comparing CABG to medical therapy by the Coronary Artery
Bypass Graft Surgery Trialists Collaboration, CABG reduced symptoms
in patients with multivessel disease compared with medical therapy,
and increased survival in patients with three-vessel disease, or
two-vessel disease involving the proximal LAD (figure 1).
These studies provided the basis for the belief that
single-vessel disease is most appropriately treated medically or
with percutaneous coronary intervention (PCI) but generally not
with CABG. These studies also supported the concept that the higher
a patient's risk for future cardiovascular events, the greater the
likelihood of benefit from CABG.
PTCA versus CABG surgery
More than a decade after the trials comparing CABG and medical
therapy were performed, randomized clinical trials were performed
comparing percutaneous transluminal coronary angioplasty (PTCA) and
CABG. The Bypass Angioplasty Revascularization Investigation (BARI)
is the largest study comparing CABG (n = 914) with PTCA (n = 915)
for the treatment of multivessel coronary disease. Three-vessel
disease was present in 41% of patients. There were more in-hospital
myocardial infarctions (MI) in the CABG group, but the PTCA group
had greater need for emergent CABG and repeat PTCA. Survival at 5
years was not different between the revascularization groups (CABG
89.3%, PTCA 86.3%,
= 0.19). However, there was an increased need for repeat
revascularization with PTCA within first 5 years (54% vs. 8%). By 7
years, there was a survival benefit with CABG (84.4% vs. 80.9%,
= 0.043), which could be explained by the treated diabetes group
where this difference was marked (CABG 76.4%, PTCA 55.7%,
= 0.0011). There was no difference in survival between the two
groups in patients without diabetes (CABG 86.4%, PTCA 86.8%,
Furthermore, one of the most important lessons from BARI, which has
been borne out in all randomized trials comparing PCI and CABG, is
that <50% of patients with multivessel disease are appropriate
for PCI, whereas >90% of patients are appropriate for CABG.
Therefore, the debate is really about the most appropriate form of
revascularization for the minority of patients amenable to both
procedures. For most patients with multivessel disease who require
revascularization, CABG is the only appropriate therapy.
Eight smaller studies that preceded BARI were analyzed together
in a meta-analysis.
This analysis included 3371 patients with a mean follow-up of 2.7
years (figure 2). No survival difference was found between CABG and
PTCA. However, 17.8% of the PTCA group required CABG during the
year after enrollment. Thus, data from both the BARI study and a
meta-analysis of 8 trials suggest that PTCA does not adversely
affect survival, except perhaps in diabetics. Nonetheless, the need
for repeat procedures in the PTCA group, including subsequent CABG,
Coronary stenting compared with PTCA
Within the past decade, stenting has become a routine part of
PCI. Stenting increases procedural success rates, decreases major
complications, (including the need for emergency CABG), and
increases the postprocedural coronary luminal diameter, which is
maintained at 6 months.
One of the main setbacks to PTCA is restenosis; stenting decreases
the frequency of restenosis by approximately 50%.
These advantages of stenting over PTCA make it necessary to
readdress the role of stenting in the treatment of multivessel
coronary artery disease.
Newer developments in CABG surgery
Significant advances have also occurred for surgical
revascularization; perhaps the most significant has been the use of
the internal mammary artery (IMA) graft. IMA grafts increase
10-year survival in patients after CABG compared with saphenous
vein grafts alone (86.6 vs. 75.9%,
The use of an IMA graft to the LAD reduced the risk of late MI by
approximately 40% and the need for repeat cardiac surgery by 50%.
Particular benefit from IMA grafts was seen in diabetic patients in
the BARI study, where 81% received at least one IMA graft.
A survival benefit in diabetics undergoing CABG in BARI was
apparent only in patients who received an IMA graft (7-year
survival: 83.2%). There was no difference in survival between PCI
and CABG among diabetic patients if a saphenous vein graft was
placed to the LAD (54.5% vs. 55.5%).
Despite the benefits of surgical treatment for multivessel
coronary disease, there are many disadvantages. These include: a
prolonged recovery time; cost; the need for repeat procedures,
which increases dramatically in the 5 to 10 years after the
procedure; the risk of stroke; and even more commonly, a decline in
cognitive function. Decline in cognitive function is significant
after CABG and is seen in 53% of patients at discharge, 24% at 6
months, and 42% at 5 years.
Of note, however, is that patients at the highest risk for
cognitive abnormalities after CABG also tend to be the worst PCI
candidates, such as those patients with peripheral and cerebral
vascular disease, and the elderly. Newer surgical techniques, such
as off-pump bypass grafting and hybrid revascularization with
video-thoroscopyassisted minimally invasive direct coronary artery
bypass and PCI, may decrease these complications.
Large prospective studies will be needed to truly define what role,
if any, these newer surgical techniques may have.
Randomized trials of stenting versus CABG
As the technology for PCI has advanced and stenting has become
used routinely for percutaneous coronary revascularization, the
question of the best therapy for multivessel coronary artery
disease needs to be revisited. Several randomized studies designed
to evaluate modern treatment options for multivessel coronary
disease have been performed, and initial results have been reported
this past year. These studies contain conflicting findings, and a
careful evaluation of the study populations, trial designs, and
results is necessary to understand these differences (Table 2).
The Argentine randomized study (ERACI II) compared coronary
stenting with CABG in the treatment of 450 patients with
symptomatic multiple-vessel disease.
This study included patients with left main coronary artery disease
if the left main stenosis was appropriate for stenting (n = 21).
Seventeen percent of patients in ERACI II had diabetes. Abciximab
was administered to 28% of the PCI group. The Gianturco Roubin II
stent (Cook, Bloomington, IN) was the primary stent used. Among the
450 patients in the study, major adverse cardiovascular events
(MACE), including death, Q-wave MI, the need for repeat
revascularization procedures, or stroke occurred in 3.6% of
patients in the PCI group versus 12.3% of patients in the CABG
= 0.002). Thirty-day mortality was also lower with PCI (0.9% vs.
<0.013), although the mortality in the CABG group was higher
than might be expected. With a mean follow-up of 18.5 months, the
PCI group had greater estimated survival at 900 days compared with
CABG (96.9% vs. 92.5%,
<0.017). However, the PCI group had a greater frequency of
repeat revascularization (16.8% vs. 4.8%,
<0.001) and were less likely to be free from angina (84% vs.
= 0.01). Thus, in ERACI II, PCI for multivessel or left main
coronary artery disease was associated with a lower mortality and
lower frequency of other adverse events at the expense of a greater
need for repeat revascularization procedures and less complete
relief of angina.
The Arterial Revascularization Therapies Study (ARTS) was a
multinational study of stenting compared with CABG.
The highest-risk patients were excluded from this study, and
approximately two-thirds of the patients had two-vessel disease. At
1 year, there was no difference between the groups in the MACE rate
(defined as death, MI, stroke, or reversible ischemic neurologic
deficits) or in frequency of mortality alone (CABG 2.8%, PCI 2.5%).
The PCI group had an increased repeat revascularization rate (16.8%
vs. 3.5%) at 1 year. PCI patients also had an increased requirement
for medical therapy and were less likely to be free from angina.
Thus, this study suggests that stenting for multivessel disease
patients is as efficacious as CABG at 1 year in terms of major
adverse events, but with greater incidence of repeat
Dr. Rodney Stables presented preliminary data from the Stent or
Surgery Trial (SOS)
at the 2001 American College of Cardiology scientific sessions.
Patients with multivessel coronary disease were randomized to
either stenting or CABG. This study demonstrated that the primary
end point of repeat revascularization was significantly higher for
stenting (20.3%) than for CABG (5.8%). Mortality at 1 year was
lower in CABG patients (0.5% vs. 2.5%,
= 0.05), and the difference at 2 years was even greater (1.2% vs.
= 0.007). Of note, the chance occurrence of 8 cancer deaths in the
PCI group and only 1 in the CABG group largely accounted for the
difference in mortality. Furthermore, the CABG group had remarkably
low 1-year mortality compared with previous randomized studies and
registry reports, further increasing the difference between these
Perhaps the most remarkable finding in the ARTS and ERACI II
studies is the reduction in the repeat revascularization rates with
stenting compared with prior PTCA studies (Table 3). Despite being
able to treat more complex lesions and have more complete
revascularization, the frequency of repeat revascularization was
less than half of that seen in the earlier trials. This is due
primarily to two factors. The first is that stents reduce the
frequency of restenosis by approximately one-half. However, the
other major reason for the lower frequency of repeat
revascularization among PCI patients in the stent versus CABG
trials than the PTCA versus CABG trials is that repeat angiography
and follow-up functional tests were not performed routinely as part
of the study protocol. It is important to recognize that the
routine performance of follow-up angiography and functional tests
in asymptomatic patients increases the frequency of repeat
revascularization procedures, and it does so more among patients
initially treated with PCI than among patients initially treated
Also at this year's American College of Cardiology meeting,
preliminary data from the Medical, Angioplasty, and Surgery Study
were presented by Dr. Whady Hueb. In this study, 611 patients were
randomized to either medical therapy (n = 203), PCI (n = 205), or
CABG (n = 203). Preliminary data indicate that the CABG patients
had fewer MIs (CABG 1%, medical therapy 2%, PCI 8;
= 0.0015) and greater event-free survival (CABG 98%, medical
therapy 94%, PCI 78%). The results of MASS II differ from those of
the larger studies described above for reasons that aren't entirely
clear. Important data about MASS II have not been reported,
including the PCI success rate, number of vessels treated per
patient, and the number of vessels treated by each method (PTCA vs.
stents, etc.). However, the strength of this study is the inclusion
of an arm in which medical therapy was administered. Given the
improvements in medical therapy since the early
CABG-versus-medical-therapy studies previously described, it
remains unclear whether, or to what extent, revascularization
provides a survival advantage over current medical therapy.
The most recent study of coronary disease is unique in that it
only included groups of patients that have been excluded from all
the prior randomized trials comparing PCI and CABG. In the Angina
With Extremely Serious Operative Mortality Evaluation (AWESOME)
study, a multicenter randomized clinical trial performed
exclusively at Veteran's Administration hospitals, PCI was compared
with CABG in "high-risk" patients known to have an increased risk
Such patients included those with medically refractory unstable
angina and at least one of the following characteristics: age
>70 years, prior cardiac surgery, a left ventricular ejection
fraction <35%, an MI ¾ 7 days, or the need for an intra-aortic
balloon pump. Preliminary data presented at the 2001 Society of
Cardiac Angiography and Interventions meeting by Dr. Douglass
Morrison demonstrated that there is no difference in 3-year
survival between the two groups (CABG
79%, PCI 80%,
>0.46 ). The AWESOME data indicate that even among these
high-risk patients, PCI and CABG appear to be equivalent in terms
of survival and survival free of MI, though as in all the studies,
repeat procedures were required more frequently among patients
initially undergoing PCI.
Special subgroups are frequently identified within clinical
studies that are at high risk of cardiac mortality and at high risk
from revascularization. Since sufficiently large studies comparing
stenting with CABG for multivessel disease have not been performed
in these high-risk groups to be certain about the most appropriate
therapy, a discussion of treatment options in these groups is
Diabetes is associated with diffuse, severe, and distal coronary
disease with more complex coronary lesions and is associated with
rapid progression of atherosclerosis. As a result, the mortality
rate from coronary disease is four times higher in the diabetic
Diabetics also have increased morbidity and mortality with coronary
revascularization. Despite high initial success rates with PTCA in
both diabetics and nondiabetics, the restenosis rate is nearly
twice as high in diabetics.
Restenosis is lower in diabetics who undergo stenting rather than
PTCA, but there is still a 10% greater risk of in-stent restenosis
in diabetics than in nondiabetics.
CABG is also associated with a higher complication rate in
diabetics, including a two-fold higher in-hospital mortality.
The results of BARI initiated great concern about PCI in
diabetic patients with multivessel coronary disease.
Among the 353 diabetic patients, 7-year survival was much greater
with CABG (76.4% vs. 55.7%,
= 0.0011). Similarly, the Emory Angioplasty versus Surgery trial
showed a greater survival in diabetics undergoing CABG versus PCI.
Analysis of the New Northern New England registry similarly found a
mortality advantage to CABG in diabetic patients who met BARI
inclusion criteria, with a hazard ratio (HR) for PTCA of 1.49 (95%
confidence interval (CI): 1.02 to 2.17), which increased with
three-vessel disease and diabetes (HR 2.02; 95% CI: 1.04 to 3.91).
The greater survival after CABG in diabetics may be derived from
the high mortality associated with MI in this population. The
relative risk of death after an MI in a diabetic with a prior CABG
is 0.09 (95% CI: 0.03 to 0.29) compared with a diabetic with no
Despite these studies suggesting that CABG is superior to PCI in
the treatment of diabetics with multivessel disease, there are
still some diabetic patients with multivessel disease in whom PCI
may be appropriate. In the BARI registry, which included 2010
patients who fulfilled inclusion criteria but were not randomized
(due to physician or patient preference), twice as many patients
underwent PCI than underwent CABG. Nevertheless, in the registry,
there was no difference in mortality at 7 years between patients
who underwent CABG and PCI, even within the diabetic subgroup (CABG
14.2%, PTCA 13.9%,
However, patients in the registry who underwent CABG were more
likely to have three-vessel disease and had more complex lesions
with more class C lesions, longer lesions, and more proximal LAD
lesions. Thus, those diabetics with the greatest number of lesions
were preferentially referred for CABG; the remaining two-thirds of
the diabetic patients with less severe multivessel disease
underwent PCI. Furthermore, stents and glycoprotein (GP) IIb/IIIa
inhibitors may be more beneficial in the diabetic population;
however, these agents were not utilized in BARI.
Patients with renal failure are at extremely high risk for
cardiovascular events. Such patients are also at a greatly
increased risk from coronary revascularization. Chronic renal
failure (CRF) is one of the most potent risk factors for restenosis
Stenting decreases the need for repeat revascularization by more
than 50% among CRF patients, but the restenosis rate is still
almost twice that of patients without CRF.
Moreover, procedural success and in-hospital mortality after PCI is
However, CABG also has a greater morbidity and mortality in the CRF
population, with in-hospital mortality reported as high as 20% in
There have been no randomized studies comparing PCI and CABG in CRF
patients, and, in fact, such patients have been systematically
excluded from all major randomized studies. Only one prospective
trial in diabetic patients with coronary disease has been
performed, in which an invasive approach (PTCA or CABG) was
compared with medical therapy given to patients on dialysis.
Although the outcome of this trial, performed in 1980s, indicated
that an invasive strategy was far superior to that of patients
receiving medical management, neither treatment arm is truly
reflective of today's practices. Thus, there is no good evidence to
guide the clinician in the therapy of multivessel disease in a
patient with renal failure . Since CRF patients typically have
diffuse, severe coronary disease, one might speculate that CABG
might be of greatest benefit in these patients. However, operative
mortality is exceptionally high in such patients. The most
appropriate therapy for patients with CRF remains unknown.
Beyond the trials
Although randomized controlled trials provide the strongest
evidence for a difference in treatment outcomes in patients with
multivessel disease, these studies have persistent limitations.
Study populations in clinical trials generally represent a select
minority of the population of patients being studied because of
restrictive inclusion/exclusion criteria, because centers
participating in the studies are often different from the majority
of hospitals in important ways, and because patients who accept
randomization often differ from the population of patients at
large. Therefore, further information can be obtained from registry
information that may be more reflective of clinical practice.
In the BARI registry, twice as many patients underwent PCI than
underwent CABG, indicating patient or physician preference for this
less invasive approach. Despite this, there was no difference in
mortality between the two treatment arms. Even the mortality
difference based on diabetic status was eliminated.
Thus, the registry suggests that by encompassing all of the
clinical variables present in each individual case, such as extent
of coronary disease, feasibility of revascularization with each
technique, and overall health of the patient, etc., PCI is an
acceptable alternative to CABG. The importance of these more subtle
factors that are incorporated into clinical decision-making
regarding revascularization techniques is also emphasized by the
EAST registry, in which 3-year survival was greater in the registry
patients (96.4%) than in randomized patients (93.4%,
Thus, PCI or CABG are both acceptable methods of revascularization
for patients with multivessel coronary disease. Consideration of
individual patient factors may make the choice of revascularization
more appropriate for a given patient.
Advances in PCI, cardiac surgery, and medical therapy have
improved the prognosis of patients with multivessel coronary artery
disease. Large-scale clinical studies are not available to
delineate the most appropriate therapy for all clinical decisions.
Revascularization of patients with multivessel disease not only
reduces symptoms, but also prolongs survival in certain patient
subsets. Larger studies with longer follow-up are needed to exclude
small differences in outcome between these different
revascularization strategies, as well as medical therapy.
Currently, the existing data suggest equivalent outcomes in
patients with multivessel disease treated with PCI or CABG if they
are amenable to both procedures.