In this issue of Cardiovascular Revascularization Medicine , Lipinski and colleagues report the results of a meta-analysis comparing heparin versus bivalirudin in patients undergoing percutaneous coronary intervention (PCI). Among 30,446 patients in 14 randomized controlled trials, there was no difference in all-cause mortality or myocardial infarction (MI) at 30 days. There was, however, a significantly higher rate of definite stent thrombosis with bivalirudin versus heparin (relative risk [RR] 1.61, 95% confidence interval [CI] 1.18–2.20; p = 0.003). Although there was significantly less non-coronary artery bypass grafting major bleeding with bivalirudin versus heparin (RR 0.59, 95% CI 0.51–0.70; p < 0.001), this effect was attenuated when glycoprotein IIb/IIIa (GPI) use was provisional in patients receiving heparin (RR 0.67, 95% CI 0.48–0.94; p = 0.02). Perhaps most importantly, meta-regression analysis indicated that as the difference in GPI use between heparin and bivalirudin increased, so did the risk of major bleeding.

This meta-analysis may at first appear to further confound the choice of anticoagulation for PCI. In reality, however, there are a few key variables that inform a nuanced interpretation of landmark clinical trials comparing heparin versus bivalirudin, both past and present: (1) mandatory versus provisional GPI use; (2) heparin dose versus duration of bivalirudin; (3) population studied (elective versus acute coronary syndromes); (4) P2Y12 inhibition; and (5) radial versus femoral access. In regards to the first variable, it is important to recall that until recently, most randomized trials compared bivalirudin with provisional GPI to heparin with mandatory GPI. This is true of the Harmonizing Outcomes with Revascularization and Stents in Acute Myocardial Infarction (HORIZONS-AMI) and Intracoronary Stenting and Antithrombotic Regimen: Rapid Early Action for Coronary Treatment 4 (ISAR-REACT 4) trials, which both showed an increase in major bleeding with heparin plus GPI when compared to bivalirudin alone. Similarly, the European Ambulance Acute Coronary Syndrome (EUROMAX) trial compared heparin (69% GPI) to bivalirudin (11% GPI) in patients with ST segment elevation MI (STEMI), again showing an increase in bleeding with heparin . The strategy of routine GPI use upon a background of P2Y12 inhibition and heparin, however, does not appear to show a reduction in infarct size or ischemic endpoints in patients undergoing PCI for STEMI . Yet it is well-established that GPI use increases bleeding across the spectrum of patients presenting for PCI . The first question to ask of any trial comparing heparin to bivalirudin is thus: Was GPI use mandatory or provisional? This may seem to be an obvious statement, but if a GPI is given with anything, there will be more bleeding.

The second key variable when comparing trials of heparin versus bivalirudin is also intuitive: heparin dose (when given without a GPI). Despite the emergence of sophisticated procedural and pharmacological interventions, the dosing of heparin for PCI remains startlingly empiric. Nevertheless, the trend over time has been toward decreasing doses of weight-based heparin. Patients undergoing elective PCI and receiving 100 units/kg, when compared to those receiving 140 units/kg, appear to have less major bleeding without an increase in ischemic events . This trend has continued toward even lower heparin doses, with 70 units/kg in STEMI patients showing rates of bleeding similar to bivalirudin in the How Effective Are Antithrombotic Therapies in Primary Percutaneous Coronary Intervention (HEAT-PPCI) trial . It would appear reasonable to conclude that lower heparin doses decrease bleeding, and at least HEAT-PPCI suggests that there is no price to pay in ischemic complications. The recently presented Bivalirudin in Acute Myocardial Infarction versus Glycoprotein IIb/IIIa and Heparin: A Randomized Controlled Trial (BRIGHT) adds a further layer of complexity. This study, performed exclusively in China, randomized acute MI patients to heparin, heparin and tirofiban, or bivalirudin. The dose in the heparin arm was 100 units/kg, but it added the wrinkle of a prolonged infusion of bivalirudin post-PCI, with a mean duration of 4 hours. This appeared to eliminate the excess of stent thrombosis previously seen with bivalirudin, in spite of the fact that overall rates were quite low (< 1%). Although there was no difference in ischemic endpoints, there was more BARC major bleeding observed with heparin vs. bivalirudin at 30 days. This study perhaps raises more questions than it answers. Is heparin 100 units/kg too much? Were stent thrombosis rates so low because of the Asian population? And does it really make sense to give an intravenous anticoagulant for 4 hours after PCI to decrease bleeding?

The third key variable in comparisons of heparin versus bivalirudin is the population studied and, more specifically, the bleeding risk of the population. The recently presented Novel Approaches in Limiting or Preventing Events III (NAPLES III) trial is instructive in that it compared bivalirudin versus heparin 70 units/kg in patients at high risk of bleeding and undergoing elective PCI with femoral access. Despite this high-risk population, there was no difference in major bleeding by REPLACE-2 criteria between heparin and bivalirudin (2.6% vs. 3.3%, p = 0.54) . This suggests that although periprocedural bleeding is not uncommon in patients undergoing femoral access, there is unlikely to be a major benefit of bivalirudin over heparin in the elective setting. It is also worth noting that in the present meta-analysis, 6 of the 14 trials included patients that were primarily undergoing elective PCI.

The fourth key variable is the P2Y12 inhibitor used in trials of heparin versus bivalirudin. This is a very important question that is likely to remain unanswered for the near future, as prasugrel and ticagrelor have just recently emerged as alternatives to clopidogrel during the trials in the present meta-analysis. It is interesting to note that in the HEAT-PPCI trial, 89% of patients received prasugrel or ticagrelor. One would therefore expect overall lower rates of stent thrombosis, given that both drugs significantly decrease stent thrombosis in acute coronary syndromes (ACS) when compared to clopidogrel. Despite this more potent P2Y12 inhibition, rates of definite or probable stent thrombosis at 1 month were still higher with bivalirudin versus heparin (3.4% vs. 0.9%, p = 0.001). Although overall rates were lower, EUROMAX also showed a higher rate of acute stent thrombosis with bivalirudin versus heparin (1.1% vs. 0.2%, p = 0.007) despite the use of the newer P2Y12 inhibitors in just under 50% of patients. These results are reflected in the present meta-analysis, where the risk of definite stent thrombosis at 30 days among patients undergoing primary PCI was markedly increased (RR 2.15, 95% CI 1.15–4.03; p = 0.02). It would therefore appear that the more potent P2Y12 inhibitors do not necessarily decrease or abolish the excess of acute stent thrombosis in patients receiving bivalirudin.

The fifth key variable is the predominant route of access used in trials of heparin versus bivalirudin. This is perhaps the most contentious issue in recent ACS trials, for the simple fact that radial access is not the principal route used in the United States. It is also, not surprisingly, quite variable among the recent trials comparing heparin to bivalirudin: primarily radial access in HEAT-PPCI, primarily femoral in ISAR-REACT 4, and somewhere in the middle for EUROMAX. It has been postulated that the lack of a difference in bleeding outcomes between heparin and bivalirudin in HEAT-PPCI may in part be due to an overall decrease in access-related bleeding, given the 81% radial access. This assumes that bivalirudin decreases primarily access-related bleeding, with less impact upon non-access related bleeding. It is important to remember, however, that more than 60% of Thrombolysis in Myocardial Infarction major and minor bleeding events in the original bivalirudin trials (REPLACE 2, ACUITY, and HORIZONS-AMI) were non-access related . In other words, the majority of bleeding events are not access-related. Whether bivalirudin decreases non-access related bleeding remains an open question; BRIGHT suggests that it does, whereas HEAT-PPCI suggests that it does not. It is therefore uncertain if there is more benefit of bivalirudin in patients with ACS undergoing femoral access for PCI, when compared to heparin alone; in essence, such a trial would look like HEAT-PPCI but with femoral access. Although a radial proponent might say that if you care that much about bleeding, you should probably use radial access.

Despite this abundance of variables, this publication is an important addition to the contemporary evidence base that calls into question the benefit of routine bivalirudin for PCI, and in particular PCI for ACS. There has been a preponderance of rather public hand-wringing regarding this controversy of bivalirudin versus heparin, in part because recent trials like HEAT-PPCI seem to contradict the paradigm established by trials like HORIZONS-AMI that bivalirudin decreases bleeding. The present meta-analysis, however, draws attention to what has been the elephant in the room: GPI use increases bleeding. Until recently, it was easy to forget that in HORIZONS-AMI the use of GPI was mandatory with heparin but provisional with bivalirudin. It was similarly easy to discount the increase in acute stent thrombosis observed with bivalirudin, in part because there was no significant increase in stent thrombosis observed at 30 days. Yet when one realizes that the cost of heparin is minuscule compared to bivalirudin, and that heparin is at least as efficacious and safe, these facts will become more and more difficult to ignore.