One of the most pressing aspects of the management of patients with stable ischemic
heart disease (SIHD) is weighing the risk/benefit trade‐off for coronary revascularization.
This shared decision with the patient has been framed by the clinical intent of revascularization
(survival benefit versus symptom management), cardiovascular risk modifiers (eg, ejection
fraction and diabetes mellitus status), and the severity of coronary atherosclerosis.
Landmark randomized trials have enhanced our understanding of the role for coronary
revascularization in patients with SIHD,
1
,
2
,
3
,
4
,
5
,
6
,
7
but questions remain. Now‐historical trials of surgical coronary revascularization
demonstrated a survival benefit compared with medical therapy in patients with 3‐vessel
coronary artery disease (CAD) or significant left main CAD. However, contemporary
randomized trials of principally percutaneous revascularization have not shown a difference
in survival compared with optimal medical therapy, even among patients with moderate
to severe ischemia on functional imaging, including patients with 3‐vessel disease
involving the left anterior descending artery. Nonetheless, patients with left main
CAD were excluded from these trials, and there remains a possibility of selection
pressures having limited enrollment of patients with complex 3‐vessel CAD.
7
Therefore, for some clinicians, there is lingering uncertainty whether coronary revascularization
might reduce the incidence of myocardial infarction (MI) or death in patients with
severe CAD.
In this issue of the Journal of the American Heart Association (JAHA),
8
Bainey and colleagues address this question in a nonrandomized, observational analysis
from the APPROACH (Alberta Provincial Project for Outcomes Assessment in Coronary
Heart Disease) registry database, which includes data from 2002 to 2016 from the 3
cardiac catheterization laboratories in Alberta, Canada. Paired with additional public
data sources to capture long‐term patient outcomes, the authors examined clinical
outcomes, comparing initial management strategies in 9016 patients with suspected
SIHD referred for coronary angiography and found to have high‐risk coronary anatomical
features.
What Did the Investigators Observe?
High‐risk coronary anatomical features were defined as angiographically defined stenoses
≥70% in all 3 epicardial arteries or left main disease ≥50%. Patients were categorized
as having undergone revascularization if percutaneous coronary intervention or coronary
artery bypass graft surgery was performed within 3 months of the index coronary angiogram.
The primary end point of interest was the composite of death or MI. The median follow‐up
time was 6.2 years. The association between treatment strategy and outcomes was assessed
using a Cox proportional hazards model with adjustment using inverse probability of
treatment weighting (IPTW) to help account for baseline differences between the groups.
In total, 61% of patients underwent coronary revascularization within 3 months of
the index coronary angiogram, 2175 by percutaneous coronary intervention and 3312
by coronary artery bypass graft surgery. The principal observation from this analysis
was that patients who underwent revascularization had lower rates of the composite
end point (2.7% versus 6.8%; hazard ratio [HR], 0.62; 95% CI, 0.58–0.66), with consistent
associations for percutaneous coronary intervention (HR, 0.64; 95% CI, 0.59–0.70)
and coronary artery bypass graft surgery (HR, 0.61; 95% CI, 0.57–0.66). Patients who
underwent revascularization also had lower rates of all‐cause death and cardiovascular
mortality individually. There was a statistically significant interaction between
these observations and coronary anatomical subgroup, with the strongest associations
among patients with severe (≥70%) left main disease (HR, 0.29; 95% CI, 0.19–0.45)
or 3‐vessel disease with a proximal left anterior descending artery stenosis ≥95%
(HR, 0.42; 95% CI, 0.33–0.54).
Cardiovascular medication prescription information was available for a subset of 4974
patients at 6 months.
Although patients who were revascularized had higher rates of use of lipid‐lowering
agents, β‐blockers, and P2Y12 inhibitors, adjustment for these differences did not
significantly alter the major findings.
What Have We Learned?
Bainey et al have addressed a timely topic central to the management of patients with
SIHD. The population studied was at high risk for events and inclusive of some subgroups
not well represented in contemporary randomized trials. The investigators have carefully
performed analyses aiming to mitigate the unreconcilable confounding attributable
to selection bias that is inherent in this observational analysis. Their observations
raise thought‐provoking hypotheses for these anatomical subgroups.
The absolute event rates reported herein are notable. The aggregate rate of death
or MI was ≈4% at 1 year, reinforcing the high‐risk nature of this anatomical cohort.
These rates are even higher for patients with severe (≥70%) left main disease, who
had a 1‐year rate of death or MI as high as 11.8% among those managed medically (3.0%
among those who received revascularization). Although the absolute rates in individual
treatment cohorts to some degree reflect underlying comorbidities and confounding,
they nonetheless highlight the concerningly high event rates of these high‐risk subgroups.
This analysis from the APPROACH registry is to be interpreted in conjunction with
the ISCHEMIA (International study of comparative health effectiveness with medical
and invasive approaches) trial,
1
which enrolled patients with SIHD and moderate or severe ischemia on stress testing.
Coronary computed tomography angiography was performed in most patients to exclude
left main or nonobstructive CAD. Qualifying patients were randomized to a strategy
of intended revascularization, which was to occur within 30 days, versus a strategy
of optical medical management. Of the 5179 patients enrolled, 79% in the revascularization
strategy group and 21% in the optimal medical therapy group underwent revascularization
during the trial. The rate of the composite end point of death from cardiovascular
causes, MI, or hospitalization for unstable angina, heart failure, or resuscitated
cardiac arrest did not differ between trial arms at 5 years, although patients assigned
to an invasive strategy had a greater improvement in symptoms. Patients randomized
to an invasive strategy had an early imbalance in procedural MI followed by lower
rates of spontaneous MI compared with medically managed patients.
How does the cohort studied by Bainey and colleagues compare with the ISCHEMIA trial
population? First, the ISCHEMIA trial excluded patients with left main CAD. In the
ISCHEMIA trial, 45% of patients had 3‐vessel disease that was defined by ≥50% stenosis.
By these criteria, Bainey's analysis cohort from the APPROACH registry had greater
severity of coronary atherosclerosis, with all patients having angiographically significant
left main (≥50%) or 3‐vessel disease at a more stringent threshold of severity (≥70%).
The results of functional testing for ischemia were not reported in the APPROACH registry.
Angiographic assessment of lesion severity in the intermediate range correlates poorly
with functional significance,
9
,
10
and although we know that irrespective of angiographically defined coronary anatomical
features 45% of patients in the ISCHEMIA trial had severe ischemia on stress testing,
we do not know the cumulative physiological impact of the coronary disease in the
APPROACH registry. Nevertheless, by angiographic criteria, the patients included in
the report by Bainey et al reflect a higher‐risk coronary anatomical cohort than those
enrolled in the ISCHEMIA trial.
In the ISCHEMIA trial, although there was no statistically significant interaction
between treatment strategy and the extent of coronary disease, defined at the relatively
inclusive threshold of 50% stenosis, there were trends toward treatment benefit with
revascularization among patients with severe ischemia on stress testing, 3‐vessel
disease, or proximal left anterior descending artery disease.
1
For example, in patients with 1‐vessel disease, the difference in event rate with
an initial invasive management versus conservative strategy was 1.4% (95% CI, −4.1%
to 7.1%), showing a trend toward a higher event rate with invasive therapy, versus
−3.2% (95% CI, −9.5% to 3.2%) in those with 3‐vessel disease, favoring the direction
toward revascularization. As such, for some experts, uncertainty remains about the
potential benefits of initial revascularization in the population with left main disease
or severe 3‐vessel disease, as defined by Bainey and colleagues, a cohort who had
a higher absolute mortality rate at 1 year in the APPROACH registry (3%) than that
seen in the ISCHEMIA trial (1%).
Considering patients with left main CAD, the data from randomized trials supporting
revascularization compared with medical management for patients with SIHD are dated
and pertain largely to coronary artery bypass graft surgery as the mode of revascularization.
11
,
12
Yet, given guideline recommendations for revascularization in these patients
13
along with trends in real‐world practice reflecting widespread use of revascularization
compared with medical therapy in patients with left main CAD,
14
it is not likely that there will soon be a randomized trial of revascularization versus
medical management in this anatomical subgroup. Moreover, although the ISCHEMIA trial
did increase the overall degree of severity of CAD studied compared with the COURAGE
(Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation) trial,
6
the especially high‐risk patients with severe 3‐vessel disease studied by Bainey et
al may not have been included. Additional analyses based on more detailed anatomical
subsets in the ISCHEMIA trial as well as long‐term follow‐up may be forthcoming and
useful to addressing this question.
Limitations of This Analysis
Interpretation of the data from Bainey's analysis of the APPROACH registry must be
made in the context of their limitations. Most critically, this analysis was not based
on a randomized comparison and the selection by treating physicians and patients of
one treatment strategy or another is inherently and deeply confounded. Patients who
were managed conservatively were older and had higher rates of comorbidities. Moreover,
evidence‐based secondary preventive therapies were used more commonly in the cohort
who underwent revascularization. Furthermore, many of the patients in the study population
had an indication for revascularization according to the European Society of Cardiology
guidelines relevant during the study period, which give a class 1A recommendation
for revascularization in the setting of left main CAD >50% or proximal left anterior
descending artery disease >50%.
13
Therefore, specific clinical considerations presumably led to the deferral of their
revascularization.
It is worthwhile to recognize the strengths and limitations of an IPTW analysis. Weighting
patients inversely to their likelihood of receiving a certain treatment (IPTW), as
was done in the current report, has the potential to lessen treatment‐selection bias
confounding, but also creates challenges to interpretation. Patients' baseline variables
are used to create scores that quantify the likelihood, or propensity, for a given
patient to receive one treatment or the other when analyzing nonrandomized observational
data. For example, a young patient with severe left main disease and few comorbidities
would have a high propensity for undergoing revascularization, whereas an elderly
patient with severe kidney disease would be more likely to be managed medically. In
an IPTW approach, patients who have a high propensity for one treatment strategy but
in fact were treated with the opposite strategy are weighted more heavily than are
patients who received the most likely treatment given their clinical profiles. Some
argue that this approach reduces confounding with the idea that if a patient was expected
to receive treatment A but actually received treatment B, it is almost as if that
patient had been randomized to treatment B.
15
Conversely, others argue that this approach overweights the unusual cases with the
most confounding. Take, for example, a 55‐year‐old otherwise procedurally low‐risk
man with severe left main disease who does not undergo revascularization: it is almost
certainly not random that the patient or treating heart team decided to defer revascularization.
His eventual outcome may reflect an important clinical factor not captured in the
database and propensity score, such as a new cancer diagnosis. Although IPTW attempts
to account for confounding, it is not a substitute for evidence based on randomized
clinical trials for making causal inferences.
Summary
In summary, Bainey and colleagues have performed well‐designed analyses from a robust
data set pertaining to an important and understudied subset of patients with SIHD.
They have reinforced the high‐risk nature of patients with left main or severe 3‐vessel
CAD and, in a propensity‐weighted, nonrandomized observational analysis, have found
a significant association between revascularization and lower rates of adverse outcomes
in these patients. Revascularization for left main CAD, specifically in patients with
SIHD, has not been investigated in a contemporary randomized trial, and there remains
debate about the representation of severe multivessel disease. Bainey et al have drawn
attention to an incompletely resolved issue in the management of SIHD.
Disclosures
Dr Bergmark reports grant support: Pfizer, AstraZeneca, and Abbott Vascular; and consulting
fees: Philips, Abbott Vascular, Servier, Daiichi‐Sankyo, Janssen, and Quark. Dr Morrow
reports consulting fees: AstraZeneca, Bayer Pharma, InCarda, Merck, Novartis, and
Roche Diagnostics. Drs Bergmark and Morrow are members of the TIMI (Thrombolysis in
Myocardial Infarction) Study Group, which has received institutional grant support
through the Brigham and Women's Hospital from: Abbott, Amgen, Aralez, AstraZeneca,
Bayer HealthCare Pharmaceuticals, Inc, Daiichi‐Sankyo, Eisai, GlaxoSmithKline, Intarcia,
Janssen, MedImmune, Merck, Novartis, Pfizer, Poxel, Quark Pharmaceuticals, Roche,
Takeda, The Medicines Company, and Zora Biosciences.