March 16th, 2015

Better Than BEST? An Observational Comparison of PCI and CABG for Multivessel Coronary Disease

The CardioExchange Editors interview Sripal Bangalore, lead author of an observational study comparing CABG with everolimus-eluting stent PCI in patients with multivessel coronary disease. The study was presented at the 2015 ACC conference and is published in the New England Journal of Medicine. It was paired with the publication of the BEST randomized trial, which compared the same interventions.

CardioExchange Editors: Is there an advantage that this observational study is being paired with the BEST trial in the NEJM and at ACC 2015?

Bangalore: Randomized trials (such as BEST) are the gold standard to test a hypothesis, but there are limitations. BEST, a noninferiority trial, was originally powered for a sample size of >1800 but enrolled only just over 800 patients. It is therefore a very underpowered study. Moreover, the primary endpoint was a composite that included repeat revascularization. From a clinical trials perspective, composite endpoints are helpful to reduce sample size and make the trial more manageable; from a clinical practice perspective, composite endpoints are problematic. The value that patients place on each component of the composite may differ from what physicians value (e.g., patients may value prevention of death and stroke more than repeat revascularization). In addition, composite endpoints are routinely misinterpreted. The BEST trial showed that during longer follow-up, PCI was associated with a higher risk for the primary outcome, but this result was driven only by an increase in repeat revascularization, not death.

Our study has a few strengths. The sample size was 22-fold higher than that of BEST (>18,000 vs. 800), and we were in a better position to look at individual outcomes rather than a composite endpoint. Remarkably, both studies have somewhat concordant results for individual endpoints:

  • Death: no difference between PCI vs. CABG in either study
  • MI: a higher rate with PCI than with CABG in both studies. However, our study showed that the higher rate of MI with PCI did not hold for patients who were completely revascularized.
  • Repeat revascularization: a higher rate with PCI than with CABG in both studies
  • Stroke: BEST showed no difference between PCI and CABG for this endpoint, at odds with our study and with findings from most PCI vs. CABG randomized trials, where CABG has shown a higher risk for stroke. Is this because of BEST’s small sample size and extreme lack of power for a low-frequency event, among other points that the authors outline?
  • Our study also showed a higher risk for up-front death or stroke with CABG than with PCI, again concordant with prior studies. BEST did not report short-term outcomes.

CardioExchange Editors: What is the take-home point of your study for clinicians? How are the findings changing your practice?

Bangalore: The big take-home message from both studies is that, for mortality, the gap between PCI and CABG is narrowing: Both studies show no mortality difference between the two procedures. With the mortality benefit off the table, PCI and CABG are both reasonable options. Patients and physicians should weigh CABG’s up-front risk for death and stroke with PCI’s long-term risks for repeat revascularization and, perhaps, MI (among patients with incomplete revascularization).


How do the data from Dr. Bangalore’s study enhance your understanding of the findings from BEST?

To view all of our coverage from the ACC meeting, go to our ACC.15 Headquarters page.

One Response to “Better Than BEST? An Observational Comparison of PCI and CABG for Multivessel Coronary Disease”

  1. David Powell , MD, FACC says:

    The immediate “front-loaded” increased risk of CVA and death with CABG compared with PCI would likely dominate a cost benefit analysis with adjusted quality of life yrs, particularly using “present value” corrections. The mean ages were about 64; in the elderly, the upfront CVA risk with CABG may be accentuated. There is no way to compare an early significant CVA with a later revascularization. Hopefully, cardiovascular clinical research will steer away from misleading combined endpoints. (This present value concept is also essential when looking at carotid interventions vs. medical therapy.)