WASHINGTON – For patients with diabetes, there are trade-offs for selecting a percutaneous intervention (PCI) over coronary artery bypass grafting (CABG) for left main artery disease when either can be considered, according to a hypothesis-generating pooled analysis.
The pooled data from four trials indicate that either method of revascularization is “reasonable,” but risk of myocardial infarction and revascularization is higher and risk of stroke is lower in patients with diabetes following PCI relative to CABG, Prakriti Gaba, MD, said in presenting the analysis at the Cardiovascular Research Technologies conference, sponsored by MedStar Heart & Vascular Institute.
Ted Bosworth/MDedge News
Dr. Prakriti Gaba
Despite decades of advances in both PCI and CABG, the findings are remarkably similar to those of Emory Angioplasty Versus Surgery Trial (EAST), the first major study to compare PCI to CABG, which were published almost 30 years ago. In the new analysis, like in EAST, PCI and CABG were comparable for a primary composite endpoint overall, but patients with diabetes were the exception. In those, outcomes were modestly better after CABG, said Dr. Gaba, a cardiology fellow at Brigham and Women’s Hospital, Harvard Medical School, both in Boston.
“More and more I am hearing from practitioners that diabetes does not matter, but what I get from your data is that diabetes still matters,” said Spencer B. King, MD, a pioneer of PCI affiliated with Emory University, Atlanta.
Dr. King, the first author of the 1994 paper and a panelist in the late-breaking trial session where the new data were presented, pointed out that a relatively limited proportion of patients with diabetes are equally suitable for PCI and CABG because of other considerations. However, he said an updated look once again suggesting that PCI and CABG are not equivalent for left main lesions in patients with diabetes “is helpful to see.”
CABG traditionally preferred for left main revascularization
The issue was revisited because CABG has been preferred traditionally for left main disease, but there was increasing evidence that PCI is associated with similar survival, according to Dr. Gaba. These new data support that contention, even if it shows that outcomes are not the same in those with diabetes relative to those without.
In this pooled analysis, data were drawn from four trials. Each compared PCI with drug-eluting stents with CABG in patients that were considered suitable for either. From the four trials, the numbers in this analysis included 705 patients from SYNTAX, 600 patients from PRECOMBAT, 1,184 patients from NOBLE, and 1,905 patients from EXCEL.
The focus was on the 1,104 patients with diabetes relative to the 3,289 without. The primary endpoint was all-cause death at 5 years. The multiple secondary endpoints included cardiovascular (CV) death, MI, stroke, and revascularization.
Overall, the 5-year mortality, independent of revascularization procedure, was 14.8% for those with diabetes and 9.3% for those without (P < .001). For this endpoint, the rates were numerically lower but not statistically different for CABG whether patients had diabetes (14.1% vs. 15.3%) or no diabetes (8.9% vs. 9.7%).
However, the rate of spontaneous MI was twice as great with PCI than with CABG for those with diabetes (8.9% vs. 4.4%), which doubled the hazard ratio within significant confidence intervals (HR, 2.01; 95% CI, 1.21-3.35). The rates of revascularization were also about twice as great with PCI than with CABG (24.5% vs. 12.4%), again producing a twofold increase in risk (HR, 2.12; 95% CI, 1.56-2.87).
For stroke in patients with diabetes, there was no difference in events at 5 years for PCI relative to CABG (2.1% in both groups). However, in those without diabetes, a trend approaching significance favored CABG over PCI (1.2% vs. 2.1%; HR, 0.177; 95% CI, 0.99-1.77). This difference was concentrated in the first year, when stroke rates among those treated with CABG were more than double the rates among those treated with PCI. Over time, this difference dissipated so that the difference was reduced to a trend at the end of follow-up.