Would the ISCHEMIA trial be quite the compass for conservative care it now is in stable coronary artery disease (CAD) if follow-up for the primary outcome had been a few years longer? That question may have helped inspire a new meta-analysis of 25 trials, with almost 20,000 patients, that pitted coronary revascularization plus medical therapy against meds alone in stable CAD populations.
ISCHEMIA showed no significant difference between the two strategies for its clinical primary endpoint after a median follow-up of about 3 years. The trial and these provocative findings have been influential and widely debated since their unveiling in 2019.
Conversely, in the new meta-analysis — with almost half the trials following patients for 5 to 15 years, and one much longer — the risk for cardiac death with invasive management fell 21% overall and was seen to decline 19% for every 4 years of follow-up vs a conservative strategy.
Benefits from invasive management were dependent on optimization of medical therapy as well as follow-up duration, and were partly driven by a 26% reduction in risk for spontaneous myocardial infarction (MI), reported Eliano P. Navarese, MD, PhD, Nicolaus Copernicus University, Bydgoszcz, Poland, this week during the virtual 2021 Congress of the European Association of Percutaneous Cardiovascular Interventions (EuroPCR).
The analysis suggests that, in patients with stable CAD and myocardial ischemia, “the benefits of revascularization and optimizing medical therapy are additive, and the combination of them can reduce cardiac mortality and spontaneous MI,” he said.
“Based on these findings, which are solid and consistent, I think it should be considered strongly, and it will be probably almost unethical not to offer revascularization to stable patients with coronary artery disease,” Navarese, who is also lead author on the study’s May 18 publication in the European Heart Journal, said in a panel discussion following his presentation.
ISCHEMIA had assigned 5179 stable patients with moderate-to-severe ischemia at stress testing to the invasive strategy or conservative care, that is, to optimal medical therapy with vs without cardiac catheterization and possible coronary revascularization. Investigators found no significant difference between the strategies for the primary endpoint, a composite of cardiovascular (CV) death, MI, hospitalization for heart failure or unstable angina, or resuscitated cardiac arrest.
As often noted, ISCHEMIA’s primary endpoint curves showing no significant difference between management strategies at a median of 3.2 years crossed after about 2 years on an apparent trajectory to show later superiority for the invasive strategy.
Although the trial was underpowered for conclusions about any individual clinical endpoints, Navarese observed, it produced a signal of benefit for the outcome of CV death at 5 years in patients assigned to the invasive strategy.
“Recommendations for medical therapy alone based on trials with limited follow-up have likely underestimated the benefits of revascularization plus medical therapy,” the meta-analysis publication states about managing patients with stable CAD, likely pointing a finger at ISCHEMIA.
“The weight of evidence suggests that revascularization in CCS [chronic coronary syndromes] should be considered a nonprognostic procedure,” except in patients with left-main coronary disease or severely impaired ventricular function, Christopher Cook, MD, National Heart and Lung Institute, Imperial College London, United Kingdom, said as an invited discussant after the Navarese presentation.
“But do the results of the current meta-analysis warrant a change?” The new analysis, he said, “is clearly methodologically robust. Secondly, the findings identify longer time horizons as being a potential reason for the lack of effect in the original study.” Finally, “its findings of reduced cardiac mortality and spontaneous MI are consistent with other datasets.”
Four Decades of Trials
Questioning the relevance of a meta-analysis of mostly older trials to current practice, Cook wondered about the role of “contemporary optimal medical therapy, particularly the impact of important drugs such as SGLT-2 inhibitors, which are underrepresented in the current trial data.”
Another discussant, Davide Capodanno, MD, PhD, University of Catania, Italy, pointed out that the trials contributing to the meta-analysis date from as early as 1979, with half conducted more than a decade ago. Also, the widely ranging studies all contributed to the meta-analysis to different degrees.
“Five studies carried the heavy weight as drivers of the full result” for the endpoint of cardiac mortality, making the data overall less broadly representative. The reduction in spontaneous MI was driven primarily by three studies, he said.
Navarese said such weighting didn’t seem to matter in successive analyses that each omitted the contribution of one trial, in succession. “No matter the weight of the study, the results were always the same, always strongly consistent.”
Also, he said, the chronological order of the trials, and therefore their variation in the CV therapy state of the art each represented, had no significant effect on the meta-analysis findings.
“Moreover, it has been argued that by current standards, older trials had a lower percentage of medication use — that the medical therapies were suboptimal in those trials,” Navarese said. But within each trial, the state of medical therapy was always the same in both treatment arms, “and the percent on medical therapy was generally comparable,” making it possible to assess the difference made by revascularization.
But the vintage of many trials in the meta-analysis is one of its major faults, William E. Boden, MD, Boston University School of Medicine and VA New England Healthcare System, told theheart.org | Medscape Cardiology.
“Frankly, it’s what I’d call a garbage meta-analysis,” he said. “They mix apples, oranges and pears” by including some “old and outmoded” studies from as early as the 1980s that were followed by seismic changes in the landscape of medical therapy for patients with CAD.
“I would submit that pooling data prior to 2000 is worthless, because you’re going to be pooling old data of inferior medical therapy,” said Boden, who chaired the COURAGE trial and was coprincipal investigator for ISCHEMIA.
“That’s the big criticism, the high degree of selection that was undertaken, sort of an indiscriminate pooling of very heterogeneous, disparate studies over a 30- to 40-year time horizon.”
Selection of Endpoints
The 25 trials in the meta-analysis encompassed 10,023 patients with stable CAD who had been randomly assigned to invasive intervention on top of medical therapy and 9783 who were managed with meds alone; their average follow-up was 5.7 years.
The relative risk (RR) for cardiac death in the invasive cohort compared with those managed conservatively was 0.79 (95% confidence interval [CI], 0.67 – 0.93; P < .01) and for spontaneous MI was 0.74 (95% CI, 0.64 – 0.86; P < .01). The study showed no significant differences in all-cause mortality risk or risk for stroke.
The reduction in risk for cardiac death for patients in the revascularization cohort was linearly associated with length of follow-up, Navarese reported, for an RR of 0.81 (95% CI, 0.69 – 0.96; P = .008) for every 4 additional follow-up years.
The selection of outcomes in the meta-analysis is another of its weaknesses, Boden said. “The entire analysis hinges on endpoints that were never primary endpoints in the trial to begin with.” Cardiac mortality is an example. In ISCHEMIA, “the reason we used all-cause mortality is because there’s no quibbling about it. Cardiac mortality, on the other hand, is very tricky because it usually has to be adjudicated.”
It’s unclear, he said, “how cardiac mortality was either defined in those studies, how it was measured, or whether or not it was adjudicated.”
In another example, the MI outcome in the meta-analysis is limited to only spontaneous MI. “In ISCHEMIA we had an overall MI endpoint with six subtypes,” Boden said, referring to the MI typing system in latest iterations of the Universal Definition of Myocardial Infarction. Limiting the analysis to spontaneous MI would, for example, exclude periprocedural MIs.
They “cherry picked” their endpoints, he said. “It’s manipulating the data to make the outcome that you want to achieve, achievable.”
2021 Congress of the European Association of Percutaneous Cardiovascular Interventions, Presented May 18.
Eur Heart J. Published online May 18, 2021. Full Text.
Navarese disclosed receiving research grants from Abbott and Amgen, and lecture fees or honoraria from Astra Zeneca, Bayer, Pfizer, and Sanofi-Regeneron. Boden disclosed receiving consultant fees or honoraria from AbbVie, Amgen, AstraZeneca, and Janssen, and research grants from Amarin.