PCI versus CABG in Diabetics

Abstract & Commentary

By Andrew J. Boyle, MBBS, PhD, Assistant Professor of Medicine, Interventional Cardiology, University of California, San Francisco. Dr. Boyle reports no financial relationships relevant to this field of study.

Source: Kapur, A et al. Randomized comparison of percutaneous coronary intervention with coronary artery bypass grafting in diabetics. 1-year results of the CARDia (Coronary Artery Revascularization in Diabetes) trial. J Am Coll Cardiol. 2010;55:432-440.

Coronary artery bypass surgery (CABG) has long been the preferred revascularization strategy for diabetic patients, but recent advances in technology have resulted in improved outcomes following percutaneous coronary intervention (PCI). Some studies, such as the recent SYNTAX study, have suggested that PCI is now approaching equipoise with CABG in medium-term clinical outcomes. However, diabetes confers a higher risk of peri-procedural complication and a worse long-term prognosis. Thus, Kapur et al performed a prospective, randomized, controlled trial comparing CABG and PCI in diabetic patients. They enrolled 510 diabetic patients in 24 centers throughout the United Kingdom and Ireland, who had either multi-vessel coronary artery disease or complex single-vessel disease. Each case was reviewed by an interventional cardiologist and a cardiac surgeon, who agreed that the case was suitable for either treatment. Exclusion criteria included age older than 80 years, left main disease, prior revascularization, cardiogenic shock, recent ST-elevation myocardial infarction (MI), ejection fraction < 20%, and contra-indications to anti-platelet therapy. Operators in both treatment arms were encouraged to perform complete revascularization whenever possible. In the CABG arm, contemporary techniques, such as off-pump surgery and arterial revascularization, were recommended. In the PCI arm, abciximab was routinely administered and staged procedures were allowed, if specified before the end of the index procedure. Bare-metal stents (BMS) were used until drug-eluting stents (DES) became available and, after that, time drug-eluting stents (all Cypher) were used. The protocol required 1-3 months of clopidogrel for BMS and 12 months for DES. The primary endpoint was a composite of death, MI, and stroke after one year. Analyses were performed using the intention-to-treat principle, and the study was designed as a non-inferiority study.

The baseline demographics were the same in both groups, with approximately 64% male, 38% requiring insulin, 60% had triple-vessel disease, 5% with renal impairment, and 18% with moderate or severe LV impairment. Of the 254 patients randomized to CABG, 230 (91%) actually underwent CABG, 14 (6%) crossed over to PCI, nine withdrew, and one patient died. Of those randomized to PCI, 99% actually underwent PCI, two withdrew, and one patient crossed over to CABG. The average time from randomization to revascularization was 64 days in the CABG group and 38 days in the PCI group. In the CABG group, 60% had triple-vessel disease, of whom 90% had complete revascularization. On average, 2.9 grafts were used per patient, 94% had left internal thoracic artery grafts, 17% had at least two arterial grafts, and 31% underwent off-pump surgery. In the PCI group, 65% had triple-vessel disease, of whom 89% had complete revascularization and only 2.6% had staged procedures. On average, 3.6 stents were implanted per patient, 31% received BMS, and 69% received DES.

After 12 months, statin and aspirin use was approximately 85% in each group, but more CABG patients were taking insulin (41% vs. 30%; p = 0.014) and more PCI patients were taking clopidogrel (54% vs. 10%; p < 0.001). Both treatments improved angina, but CABG resulted in more patients being completely free of angina (89% vs. 71%, p = 0.001). The primary endpoint of death, MI, or stroke occurred in 10.5% of patients treated with CABG and 13.0% of those who had PCI (Hazard ratio 1.25, 95% CI: 0.75-2.09; p = 0.39). This did not reach the authors' predefined criteria for non-inferiority and, thus, the study did not demonstrate that PCI is non-inferior to CABG. Mortality was the same in each group at 12 months (3.2%). Repeat revascularization rates were 2.0% for CABG and 11.8% for PCI (p < 0.001). The occurrence of major bleeding was higher with CABG (6.1% vs. 1.2%, p = 0.009). Peri-procedural MI rate was the same, but PCI patients had more late MIs (5.5% vs. 1.2%, p = 0.016). Because the stents used changed during the trial from BMS to DES, the authors performed a post-hoc analysis of CABG vs. BMS and CABG vs. DES. The primary endpoint occurred in 5.7% vs. 15.9% in CABG vs. BMS (p = 0.076) and 12.4% vs. 11.6% in CABG vs. DES (p = NS). The authors conclude that these results did not show that PCI is non-inferior to CABG, but that PCI is feasible in diabetics. They also state that longer follow-up is needed.


The CARDia study enrolled high-risk patients: diabetics with multi-vessel disease or complex single-vessel disease. Several aspects of this study are noteworthy. Firstly, they had a low 12-month mortality in both treatment arms, despite the high-risk profile of these patients. Secondly, the PCI group achieved complete revascularization in the vast majority of patients in a single procedure, with a very low repeat revascularization rate of 11.8%. This compares favorably with other trials, particularly since all patients were diabetic and, therefore, predisposed to in-stent restenosis, and since nearly a third of patients received bare-metal stents. Thirdly, there were several limitations with the design and execution of the study that are acknowledged by the authors. There was no core lab to adjudicate MI; they relied on the physicians' reporting of events. The study was underpowered to reach their primary outcome. Finally, changing the type of stent during the trial was clinically appropriate, as outcomes with DES are superior to BMS in diabetics; however, this makes the comparison between treatment groups even more difficult to interpret. The authors have demonstrated that CABG remains an excellent treatment option in diabetics, with less need for revascularization, fewer subsequent non-fatal MIs, and better relief from angina than PCI. PCI is a feasible option in this patient population; it improved angina and had lower bleeding rates than CABG with similar 12-month mortality. However, the curves are starting to separate in favor of CABG late in the first year. Longer-term outcomes are eagerly anticipated.