FREEDOM Follow-On Study: Revascularization of Multivessel CAD in Diabetics

Coronary artery bypass grafting (CABG) and percutaneous coronary intervention (PCI) are both complementary and competing revascularization strategies for coronary artery disease (CAD). In general, CABG provides a more durable revascularization (presuming arterial grafts are used) with less sensitivity to CAD complexity but at the cost of a much more invasive procedure with a markedly longer recovery. Conversely, PCI provides minimally invasive revascularization with a quick recovery period but is to some degree limited in very complex CAD and may provide a less durable result. Therefore, the optimal revascularization strategy for patients with stable angina despite adequate medical therapy, unstable angina or non-ST segment elevation myocardial infarction (STEMI) depends heavily on CAD complexity and comorbidities.

As CAD complexity increases and comorbidity burden decreases, CABG is generally preferred to PCI. Few would argue that an 85-year-old woman with single-vessel focal CAD of the dominant mid-RCA and class III angina despite medical therapy would be best treated with PCI. Conversely, a 58-year-old man with few comorbidities and severe three-vessel CAD including a critical terminal left main stenosis would be better served with CABG. But what about a patient with stable angina despite medical therapy and intermediate complexity two-vessel coronary artery disease involving the proximal LAD who is a reasonable candidate for CABG? According to current guidelines, the presence of left ventricular dysfunction or diabetes mellitus would favor CABG.

However, evidence for the superiority of CABG over PCI in patients with multivessel disease (MVD) and diabetes mellitus is limited. Among the 343 patients with MVD treated for diabetes mellitus in the Bypass Angioplasty and Revascularization Investigation (BARI), patients undergoing CABG had better survival at 5 and 10 years compared to those undergoing percutaneous transluminal angioplasty (PTCA) without stenting (80.6% vs. 65.5%, P = 0.003 at 5 years; 57.8% vs. 45.5%, P = 0.025 at 10 years).1,2 In the stent era, the CARDia (Coronary Artery Revascularization in Diabetes) randomized trial of 510 patients with diabetes and multivessel or complex single-vessel CAD showed no difference in the 1-year rate of the composite of death, myocardial infarction, and stroke for CABG (10.5%) compared to PCI with bare metal or first-generation drug-eluting stents (13.5%, P = 0.39).3 Conversely, a pre-specified subgroup analysis of the 452 patients with diabetes mellitus and three-vessel or left main CAD in the SYNTAX (Synergy between Percutaneous Coronary Intervention with Taxus and Cardiac Surgery) trial showed a higher rate of major adverse cardiac or cerebrovascular events in patients undergoing PCI with a first-generation drug-eluting stent (46.5%) compared to those undergoing CABG (29.0%, P < 0.001).4

In this context, the FREEDOM (Future Revascularization Evaluation in Patient with Diabetes Mellitus: Optimal Management of Multivessel Disease) trial randomized 1900 patients with MVD and diabetes mellitus to either PCI with mostly first-generation drug-eluting stents or CABG.5,6 Approximately one-third of patients had a recent acute coronary syndrome, and nearly two-thirds had intermediate or high complexity CAD as measured by the SYNTAX score (26.2 ± 8.4% for PCI, 26.1 ± 8.8% for CABG). At five years, the primary composite outcome of all-cause mortality, non-fatal myocardial infarction (MI), and non-fatal stroke occurred in 26.6% of patients randomized to PCI and 18.7% of those randomized to CABG (P = 0.005). Both all-cause mortality and non-fatal MI were reduced in the CABG group, though stroke was increased. Notably, there was an early hazard of CABG with regard to the primary outcome, and survival curves began to separate in favor of CABG at approximately 2 years.

Given the lag time for the benefit of CABG to become evident and the fact that the median follow-up in FREEDOM was only 3.8 years, longer-term follow up was strongly desired. Therefore, 25 high-enrolling centers (out of 140 original sites) with 943 patients (49.6%; 478 PCI, and 465 CABG) agreed to continue follow-up as part of the FREEDOM Follow-On study.7 Notably, the primary outcome was changed to all-cause mortality; while the investigators initially planned the same primary composite outcome as the original FREEDOM trial, several sites were unable to report outcomes other than mortality.

Patients in the FREEDOM Follow-On study had a median follow up of 7.5 years (mean 6.7 ± 3.1 years) and similar baseline characteristics to those who were not included. When all patients and data from the FREEDOM trial and the Follow-On study were included, all-cause mortality was significantly reduced in patients undergoing CABG compared to PCI (18.3% vs. 24.3%, P = 0.01). When the analysis was restricted to only patients in the Follow-On study, the same trend was present, though that difference did not quite reach statistical significance (18.7% vs. 23.7% for CABG vs. PCI, P = 0.076). Furthermore, a supplemental analysis allowing for a non-constant hazard suggested the absence of a treatment effect during the first two years but a mortality benefit of CABG thereafter, consistent with findings from the original FREEDOM trial. Interestingly, of the 17 centers (415 patients) reporting outcomes other than mortality, myocardial infarction (4.7% vs. 4.0% for PCI vs. CABG) and stroke (2.3% vs. 1.5% for PCI vs. CABG) were numerically similar. Finally, subgroup analyses suggested that patients younger than the median age at randomization (63.3 years) were more likely to benefit from CABG (P value for interaction = 0.001). The authors conclude that patients with diabetes and MVD have reduced long-term all-cause mortality with CABG compared to PCI with drug-eluting stents.

What are we to take away from these long-term results from the largest randomized trial to date of CABG vs PCI in diabetic patients with MVD? First, it's important to remember the population studied: diabetic patients with three-vessel CAD (82%) and a preserved LVEF (2% had LVEF <40%) but without significant left main disease who are judged to be equally eligible for PCI or CABG. In this population, the FREEDOM and Follow-On data are convincing that CABG offers these patients improved survival compared with PCI, though the mechanism remains uncertain. While tempting, results should not be extrapolated to dissimilar patients, particularly those with a reduced LVEF.

All-cause mortality is certainly an important outcome, but missing from FREEDOM Follow-On is any information about the mechanism of death. The original FREEDOM trial showed an excess of myocardial infarction in PCI patients (13.9% vs. 6.0%, P <0.001) with a trend towards an excess of cardiovascular death (10.9% vs. 6.8%, P = 0.12).6 While it is easy to presume these trends continued for the next 5 years of follow-up and lead to an excess of all-cause mortality, occasionally there are unexpected mortality events (i.e., the excess cancer-related death in the Dual Antiplatelet Therapy [DAPT] study) which may be relevant. Without information regarding the cause or mechanism of death, we are left wondering what may be done to improve the excess mortality in diabetics with three-vessel CAD undergoing PCI rather than CABG.

Additionally, death is not the only outcome that matters – stroke can be a debilitating complication of PCI or CABG, and in the FREEDOM trial patients undergoing CABG had an excess of strokes at 5 years (2.4% vs. 5.2%, P = 0.03).6 Little can be learned from the 17 centers reporting this outcome in the Follow-On study which represent only 415 patients and 8 total strokes – we do not know how representative those patients are of the entire cohort. It is likely that the stroke hazard of CABG is limited to the early post-operative period (due to cardiopulmonary bypass, embolism during surgery, etc.), but it's possible that PCI could have a late hazard related to an increase in subsequent revascularization and its resultant stroke risk. Regardless, stroke is an important outcome that should be discussed with all patients deciding between revascularization options.

It is also important to note that the 25 sites included in the Follow-On study account for only 18% of FREEDOM study sites but 50% of FREEDOM patients. We can also surmise that seven sites in the Follow-on Study contributed 528 patients to the analysis, or more than 25% of the total FREEDOM study participants (because 17 sites representing 415 patients reported the stroke outcome). These are likely overall high-volume sites, and it may be that compared to low-volume centers such sites have better outcomes after CABG.8 Differential outcomes at high-volume sites comprising a significant proportion of the entire study cohort could bias the overall study results in favor of CABG.

Furthermore, age is and should be an important factor in choosing between revascularization strategies. In addition to being a major component of cardiac surgery risk scores, age is related to the potential benefit of CABG over PCI insofar as it relates to life expectancy. FREEDOM showed that the benefit of CABG did not become apparent until the 2-year mark, and the Follow-On study subgroup analysis suggested that younger patients are more likely to benefit from CABG over PCI. CABG should only be offered to patients in this population with a life expectancy over two years, and patients with a relatively limited life expectancy that exceeds two years should have an individualized discussion with their physicians.

Finally, the FREEDOM trial began enrolling patients in 2005, two years after FDA approval of the first drug-eluting stents. Stent technology has evolved rapidly since that time, and current second-generation drug-eluting stents have thinner struts, more biocompatible polymers, and elute different antiproliferative agents as compared to first-generation drug-eluting stents. Only 6% of patients in FREEDOM received second-generation drug-eluting stents and the contemporary guideline-recommended one year of dual antiplatelet therapy was only completed in 78.1% of patients.6,7 Additionally, we are not given any information about the completeness of revascularization in patients undergoing PCI. Since CABG is generally able to achieve a more complete revascularization, this missing information may shed light on the mechanism of the mortality benefit of CABG.

In conclusion, the FREEDOM Follow-On Study provides the best evidence we have to guide revascularization of diabetic patients with multivessel CAD and a preserved LVEF, and convincingly demonstrates the all-cause mortality benefit of CABG over PCI in this setting. However, questions remain regarding the mechanism of the mortality benefit, the risks of outcomes other than mortality (including stroke), and the impact of the latest iteration of coronary stents. This information should be used to guide the multidisciplinary heart team in formulating recommendations for each individual patient which take into consideration patient preferences, comorbidities, surgical risk and CAD complexity.


  1. Bypass Angioplasty Revascularization Investigation Investigators. Comparison of coronary bypass surgery with angioplasty in patients with multivessel disease. N Engl J Med 1996;335:217-25.
  2. BARI Investigators. The final 10-year follow-up results from the BARI randomized trial. J Am Coll Cardiol 2007;49:1600-6.
  3. Kapur A, Hall RJ, Malik IS, et al. Randomized comparison of percutaneous coronary intervention with coronary artery bypass grafting in diabetic patients. 1-year results of the CARDia (Coronary Artery Revascularization in Diabetes) trial. J Am Coll Cardiol 2010;55:432-40.
  4. Kappetein AP, Head SJ, Morice MC, et al. Treatment of complex coronary artery disease in patients with diabetes: 5-year results comparing outcomes of bypass surgery and percutaneous coronary intervention in the SYNTAX trial. Eur J Cardiothorac Surg 2013;43:1006-13.
  5. Farkouh ME, Dangas G, Leon MB, et al. Design of the Future REvascularization Evaluation in patients with Diabetes mellitus: Optimal management of Multivessel disease (FREEDOM) Trial. Am Heart J 2008;155:215-23.
  6. Farkouh ME, Domanski M, Sleeper LA, et al. Strategies for multivessel revascularization in patients with diabetes. N Engl J Med 2012;367:2375-84.
  7. Farkouh ME, Domanski M, Dangas G, et al. Long-term survival following multivessel revascularization in patients with diabetes: the FREEDOM Follow-On study. J Am Coll Cardiol 2019;73:629-38.
  8. Kim LK, Looser P, Swaminathan RV, et al. Outcomes in patients undergoing coronary artery bypass grafting surgery in the United States based on hospital volume, 2007 to 2011. J Thorac Cardiovasc Surg 2016;151:1686-92.

Clinical Topics: Acute Coronary Syndromes, Cardiac Surgery, Heart Failure and Cardiomyopathies, Invasive Cardiovascular Angiography and Intervention, Stable Ischemic Heart Disease, Atherosclerotic Disease (CAD/PAD), Aortic Surgery, Cardiac Surgery and Heart Failure, Cardiac Surgery and SIHD, Interventions and ACS, Interventions and Coronary Artery Disease, Chronic Angina

Keywords: Diabetes Mellitus, Acute Coronary Syndrome, Angina, Stable, Angina, Unstable, Cardia, Angioplasty, Cardiac Surgical Procedures, Comorbidity, Cardiopulmonary Bypass, Constriction, Pathologic, Coronary Artery Bypass, Coronary Artery Disease, Drug-Eluting Stents, Embolism, Follow-Up Studies, Follow-Up Studies, Life Expectancy, Myocardial Infarction, Myocardial Revascularization, Neoplasms, Patient Preference, Percutaneous Coronary Intervention, Polymers, Metals, Random Allocation, Stroke, Taxus, Research Personnel, Ventricular Dysfunction, Left

< Back to Listings