By Michael H. Crawford, MD, Editor

Professor of Medicine, Chief of Clinical Cardiology, University of California, San Francisco

Dr. Crawford reports no financial relationships relevant to this field of study.

SYNOPSIS: A propensity score-matching analysis of all patients undergoing coronary angiography in Alberta, Canada identified a subgroup with diabetes, multivessel coronary artery disease, and left ventricular ejection fraction < 50% who were undergoing revascularization and could be separated into a group undergoing percutaneous coronary intervention (PCI) and another coronary artery bypass grafting (CABG). At five years follow-up, the CABG group experienced significantly fewer major cardiac or cerebral vascular events compared to PCI and a low risk of stroke that was similar to that observed with PCI.

SOURCES: Nagendran J, Bozso SJ, Norris CM, et al. Coronary artery bypass surgery improves outcomes in patients with diabetes and left ventricular dysfunction. J Am Coll Cardiol 2018;71:819-827.

Velazquez EJ, Petrie MC. CABG or PCI for diabetic patients with left ventricular dysfunction. J Am Coll Cardiol 2018;71:828-831.

There are no randomized trials or sufficiently powered subgroup analyses of trials that have compared coronary artery bypass grafting (CABG) to percutaneous coronary intervention (PCI) in diabetic patients with systolic left ventricular (LV) dysfunction. Thus, investigators from Canada performed a propensity score-matching analysis of the database of all patients undergoing cardiac catheterization in the province of Alberta between 2004 and 2016. Of the more than 110,000 patients identified, non-diabetics, single-vessel disease patients, those with LV ejection fraction (EF) > 50%, and those not undergoing revascularization were excluded. This left almost 3,000 patients, of whom 1,738 could be propensity matched for important clinical variables, with 869 undergoing PCI and 869 undergoing CABG. The primary outcome was major adverse coronary and cerebral events (MACCE). Secondary outcomes included mortality, stroke, myocardial infarction (MI), or repeat revascularization.

PCI resulted in higher MACCE at five years compared to CABG in those with EF 35-49% (51% vs. 28%; P < 0.001) and EF < 35% (61% vs. 29%; P < 0.001). Mortality followed the same pattern (26% vs. 16%; P < 0.001) in those with higher EFs (35% vs. 19%; P < 0.002) and in those with lower EFs. Stroke rates were all < 5% and did not differ between the two treatment groups. Repeat revascularization was consistently higher after PCI, but MI was only higher with PCI in the EF < 35% group. A sensitivity analysis of time during the trial showed similar results over the years of the study. The authors concluded that in patients with multivessel coronary artery disease, diabetes, and reduced systolic LV performance, CABG resulted in significantly fewer MACCE than PCI over long-term follow-up, without a higher risk of stroke.


Previous trials have shown that CABG is preferable to PCI in patients with multivessel disease and diabetes, but these trials provided insufficient data about those with reduced LV function to assume that the same holds for them. There are reasons to believe that the results could be different. Reduced LVEF could increase surgical mortality more than PCI mortality, neutralizing any advantage to surgery. Conversely, CABG may lead to better improvement in LVEF due to more complete revascularization than PCI, leading to better long-term outcomes. Thus, a study such as this one is important for clinical decision-making.

This study also was more real world in that about two-thirds of the patients studied exhibited an acute coronary syndrome on their index admission. One important finding was that there was no difference in stroke rates between the two treatment groups (3% for both). This is in contrast to other studies such as FREEDOM, which showed stroke rates at five years of 5.2% for CABG and 2.4% for PCI. Perhaps the low LVEFs in this study leveled the playing field in comparison to other studies. The editorial accompanying this article noted the tremendous public health implications of the study. Picking CABG instead of PCI in these patients effected a 16% reduction in mortality over five years. Compare that to the 2.8% reduction in mortality observed through randomization to sacubitril/valsartan vs. enalapril to put things in perspective.

Of course, this isn’t a randomized trial, so it will not produce the same effect that a double-blinded, randomized, drug comparison study would. Propensity matching can’t eliminate all sources of bias. Another criticism is the 12-year span of the study, since significant changes in revascularization techniques may have occurred. However, by 2004, drug-eluting stents were in widespread use in Canada, so only refinements in the further generations of these stents would be operant. Also, a time-based sensitivity analysis didn’t show any difference in the results. It would have been interesting to know how the 3,000 patients who received medical therapy only performed. In addition, the operative or procedural mortality was not given; rather, these data were included in the overall five-year results. Finally, the authors provided very little other clinical data, such as brain natriuretic peptide levels, mitral regurgitation, LV hypertrophy, or the concomitant medical therapy the patients received. Despite these flaws, the clinical message is clear and consistent with previous studies: CABG should be the first-line treatment for patients with symptomatic multivessel coronary artery disease, diabetes, and reduced systolic LV function — unless predicted operative mortality is excessive.