The identification of the optimal periprocedural antithrombotic therapy for percutaneous coronary interventions is still a matter of debate . On the one hand percutaneous coronary intervention per se induces thrombin generation which makes necessary administration of adjunctive antithrombotic therapy , on the other hand bleeding complications can significantly influence acute- and long-term survival and morbidity . The ideal antithrombotic medication should have rapid onset of action, predictable dose-dependent effect, quick reversibility, and wide therapeutic window. So far, no medication fulfills all these requirements. Medications with stronger antithrombotic power are unavoidably associated with higher bleeding risk, thus making the quest for the optimal treatment extremely challenging .

The network meta-analysis of 41 randomized clinical trials conducted by Lipinski and colleagues ( Fig. 1 ) presented in this issue of the journal clearly depicts the amount of uncertainty surrounding the optimal periprocedural antithrombotic therapy in percutaneous coronary interventions and attempts to draw general conclusions about this highly debated topic. According to the results of the meta-analysis, overall heparin plus a glycoprotein IIb/IIIa inhibitor showed the highest effectiveness in terms of major adverse cardiovascular events, all-cause mortality and myocardial infarction, with mixed results according to the type of glycoprotein IIb/IIIa inhibitor, while bivalirudin was associated with the lowest risk of major bleeding, minor bleeding and need for red blood cells transfusion .

Network of the comparisons performed in the study of Lipinski et al. .

These findings broadly confirm the strengths and weaknesses of each medication. Unfractionated heparin has traditionally been used for anticoagulation in percutaneous coronary intervention. However, it has a number of limitations such as inter- and intra-individual variations, non-linear dose-response, and inability to interact with clot-bound thrombin . The increasing evidence supporting the primary role of platelets in coronary clot development and the need for stronger adjunctive pharmacologic therapy in complex percutaneous coronary interventions or acute coronary syndromes led to the development of the glycoprotein IIIb/IIa inhibitors . Although the results of several early randomized trials proved the efficacy of this class of medications , introduction of new and more potent P2Y12-receptor inhibitors, owing to significant advances in devices biocompatibility and appropriate discrimination across different clinical subsets more recent trials failed to lend support to the systematic use of glycoprotein IIb/IIIa inhibitors mainly due to the concerns related to an excess in bleeding . Bivalirudin has been shown to overcome some limitations of available medications. Periprocedural treatment with this direct thrombin inhibitor has been associated with similar effectiveness and lower incidence of bleeding events as compared to heparin plus provisional or systematic use of glycoprotein IIb/IIIa inhibitors . However, the administration of bivalirudin remains limited in current clinical practice mainly due to higher costs and repeated observations of an increased risk of acute stent thrombosis.

Bleeding complications is one of the critical endpoints to determine the safety of an antithrombotic strategy. In addition, a reduction in major bleeding events after percutaneous coronary intervention has been associated with improved survival and represents an important step in improving outcomes by balancing the safety and efficacy of the therapy .

Stratifying patients according to bleeding and ischemic risk before coronary angiography and percutaneous coronary intervention and individualizing periprocedural anticoagulant and antiplatelet therapy accordingly might be essential in increasing the safety and efficacy of the procedure . The use of glycoprotein IIb/IIIa inhibitors should be preferred in ST-segment elevation myocardial infarction, thrombus-rich lesions, inadequate pretreatment with oral antiplatelet drugs and low bleeding risk patients. Indeed, the relatively young patient without significant comorbidities or conditions associated with a predictable increase in the risk of bleeding events, presenting at angiography with clearly unstable and complex lesion, relevant thrombotic burden, or evidence of distal embolic material can benefit from more aggressive antithrombotic therapy with heparin plus glycoprotein IIb/IIIa inhibitors. On the contrary, bivalirudin can substantially improve the outcome of patients who require femoral access and present comorbidities and conditions which may impose an excessive risk of bleeding. A more frequent selection of radial vascular access seems to be able to reduce the need for glycoprotein IIb/IIIa inhibitors and the incidence of bleeding complications after percutaneous coronary intervention respectively .

The study by Lipinski et al. is a Bayesian arm-based trial-level network meta-analysis which provides an additional interpretation of available evidence. In recent years the use of meta-analysis in medical research has exponentially grown because such systematic reviews have proven to be useful in detecting discrepancies between treatments and trials, identifying gaps in the existing evidence, delineating general clinical recommendations and guiding future investigations . Multiple-treatment meta-analysis or more generally network meta-analysis is the extension of standard pair-wise comparison of treatments to a network of different competitor strategies . The methodology presents highly attractive advantages compared to pair-wise meta-analysis such as the simultaneous pooling of the evidence related to multiple interventions, the combination of direct and indirect components of the evidence in a joint estimate, and the comparison of treatments without a direct connection on the basis of the existing trials . Indeed, in the simplest case of a network including three different terms (i.e. A, B, and C), each pair of treatments (e.g. A vs. B), can be compared either directly or indirectly via a common reference (e.g. C); in complex evidence networks, indirect evidence derives from many possible routes and not necessarily by a single common reference . However, network meta-analysis is based on some intrinsic assumptions, such as transitivity and consistency, which always require to be verified for each specific dataset . Although Lipinski and colleagues have applied an inconsistency model without detecting significant variations from consistency model, it might not be able to identify local inconsistency. Node-split and loops decomposition are extremely useful tools to verify the presence of local inconsistency . Additionally, network meta-analysis can be influenced by several non-uniform trial-level factors, includable within the broad concept of heterogeneity, which need always to be accounted for by subgroup analyses and, when feasible, by meta-regression . Regardless of the type of the meta-analysis (frequentist or Bayesian) and the source of the data (trial-level or patient-level), a meta-analysis requires multiple meaningful subgroup analyses with the dual objective to test main analysis results and to explore potential source of heterogeneity across dissimilar subsets . Another limitation of the data analyzed is that radial vascular access is less frequently represented as in the daily practice in different regions outside US. This might have impacted on the risk balance between ischemic complications and bleeding for the different antithrombotic therapies.

Notwithstanding the limitations mentioned above inherent to all network meta-analyses as well as to the lack of individual patient data, the study by Lipinski et al. is one of the most comprehensive comparative assessments of different adjunctive antithrombotic treatments during percutaneous coronary intervention. It clearly defines the 2 borders between which we have to move during each procedure: the combination of heparin and glycoprotein IIb/IIIa inhibitors provides the best protection against ischemic complications and bivalirudin the best bleeding avoidance strategy. Further specific patient risk-based studies are needed to safely guide our movement toward one border or the other.