Article Text

Download PDFPDF

Coronary and mortality risk of novel oral antithrombotic agents: a meta-analysis of large randomised trials
  1. Koon-Hou Mak
  1. Mak Heart Clinic, 6 Napier Road, #08-13, Singapore, Singapore, 258499
  1. Correspondence to Koon-Hou Mak; makheart{at}


Objective Oral direct thrombin and anti-Xa inhibitors have been shown to be efficacious in the prevention and treatment of venous thromboembolism, and prevention of embolic events in atrial fibrillation. Recent studies showed that dabigatran may be associated with increased rates of myocardial infarction (MI). Coronary risk for the other agents was unclear. The aim of the study is to determine the coronary risk among four novel antithrombotic agents.

Design Mixed treatment comparison meta-analysis.

Data sources and study selection Randomised controlled trials (RCTs) on ximelagatran, dabigatran, rivaroxaban and apixaban were obtained from PubMed search (February 2012) and major scientific meeting in 2011. The random-effects model was used to evaluate the effect of these agents on MI or acute coronary syndrome (MI/ACS), major bleeding complication and all-cause mortality.

Results From 28 RCTs (n=138 948), the risk for MI/ACS was higher for dabigatran (OR 1.30; 95% CI 1.04 to 1.63; p=0.021) but lower for rivaroxaban (OR 0.78; 95% CI 0.69 to 0.89; p<0.001). Ximelagatran showed a higher risk for MI/ACS, which was not statistically significant, while apixaban demonstrated a non-significant lower likelihood. Among the RCTs for MI/ACS among the four agents, only those pertaining to ximelagatran showed heterogeneity. Major bleeding complication rates varied considerably among different agents. Importantly, these agents were associated with a lower all-cause mortality, without heterogeneity among the studies.

Conclusions The risk for coronary events was significantly higher for dabigatran but not significantly higher for ximelagatran. Conversely, this risk was lower among anti-Xa inhibitors. All-cause mortality was lower among those receiving novel antithrombotic agents. This information may be useful in selecting agents for specific subsets of patients requiring anticoagulation.

  • Vascular Medicine

This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: and

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Article summary

Article focus

  • Novel oral anticoagulants have been efficacious in preventing thrombotic complications among patients with atrial fibrillation and venous thromboembolism.

  • There is concern regarding coronary risks.

  • The study aims to ascertain this risk among the novel antithrombotic agents.

Key messages

  • While oral direct thrombin inhibitors may be associated with a higher coronary risk, this risk may not be present for oral factor Xa inhibitors.

  • Individual coronary risk may influence the choice of oral anticoagulant.

Strengths and limitations of this study

  • This is a large meta-analysis, and the finding was consistent with previous meta-analysis evaluating the coronary risk of dabigatran.

  • Similar to any meta-analysis, the results are hypothesis generating for comparing direct groups of oral anticoagulants.


Several cardiovascular conditions are related to thromboembolism. In the past few decades, focus has been on the development of antiplatelet agents because of the perceived pre-eminent role played by thrombocytes in arterial, particularly coronary, thrombosis. For several years, advancement in anticoagulation has been limited to refining the heparin complex and parenteral direct thrombin inhibitors such as hirudin and bivalrudin. Indeed, warfarin was the sole oral anticoagulant for the past 60 years. Novel agents have been designed to act against factor Xa and thrombin recently. Their efficacy has been shown in preventing venous thromboembolism (VTE) among patients undergoing hip or knee surgery and embolic events among those with atrial fibrillation, and treating those with VTE or acute coronary syndromes (ACS). Amidst the enthusiasm of favourable results, higher rates of myocardial infarction (MI) among patients receiving dabigatran initially reported in the Randomised Evaluation of Long-term Anticoagulant Therapy (RE-LY) trial1 have generated concern regarding the overall effectiveness of this agent. Although subsequent re-analysis of the data following identification of another four clinical and 28 silent MI showed that the increase was not statistically significant (HR 1.28; 95% CI 0.98 to 1.67; p=0.07).2 However, a recent meta-analysis showed that the risk of coronary events was increased with the use of dabigatran, even after including the additional events.3

To date, there have been four novel oral anticoagulants that have been evaluated in large clinical trials for thromboembolic conditions: ximelagatran (Exanta; Astra-Zeneca, London, UK), dabigatran (Pradaxa; Boehringer Ingelheim, Ingelheim, Germany), rivaroxaban (Xarelto; Bayer, Leverkusen, Germany) and apixaban (Eliquis; Bristol-Myers Squibb, New York City, New York, USA). The first two agents are direct thrombin inhibitors while the latter two act against factor Xa. With different mechanism of action, the aim of the study is to review the risk of acute coronary events among these agents. Taken together, they may provide a better ascertainment of the coronary risk between direct thrombin inhibitors and anti-Xa agents.


Using the PubMed, a search was conducted on 18 February 2012 with the terms, ximelagatran or dabigatran or rivaroxaban or apixaban, and was limited to clinical trials. Additional records were identified from abstracts presented at major scientific meetings in 2011: namely, the 60th Annual Scientific Session of the American College of Cardiology (, the XXIII Congress of the International Society of Thrombosis and Haemostasis ( and the American Heart Association Scientific Session 2011 ( Only studies with at least 1000 subjects were included. Manuscripts that did not report on the occurrence of acute coronary events or all-cause mortality were excluded. Various doses of the same study drug were grouped together as treatment arm as the numbers of patients and events in each of the doses were small, especially in phase II studies. The primary outcome was acute coronary events comprising either MI or ACS (unstable angina, MI or cardiac death), based on individual reports. All-cause mortality and major bleeding complication rates were secondary outcome measures. However, the definition of major bleeding complication varied among the studies.

Study quality was assessed by the Jadad scale,4 which scored up to 2 points for randomisation, 2 points for blinding and 1 point for description of withdrawals and drop-outs. Points may be deducted for inappropriateness in randomisation or blinding. A score of 3 or more points suggest the trial was of high quality. Meta-analysis was performed using Comprehensive Meta-analysis V.2 (Biostat, Inc, Engelwood, New Jersey, USA). The associations between risk of each of the outcomes in the control groups (baseline risk): acute coronary events, major bleeding complications and all-cause mortality, with the corresponding OR of the use of each of the antithrombotic agents for each of the indication of use, namely VTE prophylaxis, treatment of thromboembolism, prevention of thromboembolism among those with non-valvular atrial fibrillation and ACSs, were evaluated with a linear fixed-effects meta-regression model. For studies using dissimilar agents in the control group, the random-effects model was applied instead. In the overall results, the random-effects model was used. Heterogeneity was quantified with I2 statistics.5 Publication bias was determined by Funnel plot and Egger regression test.6


A total of 274 abstracts were identified and reviewed. Of these, 42 full-text articles were appraised, and eventually, 28 randomised control trials (RCTs) were selected (figure 1), consisting of 138 948 participants. The numbers of trials evaluating ximelagatran, dabigatran, rivaroxaban and apixaban were six, nine, seven and seven, respectively, and were sponsored by their respective pharmaceutical companies. They were conducted in the setting VTE prevention among patients undergoing hip or knee surgery (13 studies), treatment of individuals with VTE (5 studies), prevention of embolic events in patients with atrial fibrillation (6 studies) and treatment of subjects with ACSs (4 studies). Study participants were followed from about a week to 2 years. The characteristics of the trials are provided in table 1.

Table 1

Characteristics of randomised controlled trials evaluating novel antithrombotic agents in various medical conditions

Figure 1

PRISMA (Preferred reporting items for systematic reviews and meta-analyses) flow diagram of study selection.

Impact on MI/ACS

Of the four drugs, the risk for MI/ACS was higher for dabigatran (OR 1.30; 95% CI 1.04 to 1.63; p=0.021) but lower for rivaroxaban (OR 0.78; 95% CI 0.69 to 0.89; p<0.001) (figure 2). The other oral direct thrombin inhibitor, ximelagatran, showed a higher risk for MI/ACS, which was not statistically significant, and apixaban, a factor Xa inhibitor, demonstrated a non-statistically significant lower likelihood. Unlike trials involving dabigatran, rivaroxaban and apixaban, there was marked heterogeneity for studies evaluating ximelagatran (I2=79.69; p=0.007).

Figure 2

Risk of coronary events.

Major bleeding complications

Overall, the risk of major bleeding complications was comparable between oral direct thrombin inhibitors and warfarin (figure 3). When the trial on patients with ACS was excluded, dabigatran was associated with a reduced risk for major bleeding complications (OR 0.89; 95% CI 8.80 to 0.999; p=0.049). But there was still considerable heterogeneity among the studies (I2=67.29; p=0.003). Conversely, the risk for major bleeding complication was 15% higher for rivaroxaban. Again, there was marked heterogeneity because of dissimilar trial design, the heightened risk for major bleeding complication was attenuated after excluding the study on ACS (OR 1.03; 95% CI 0.90 to 1.19; p=0.638). Test for heterogeneity became non-significant (I2=3.32; p=0.395). Overall, apixaban was associated with a non-statistically significant lower likelihood for major bleeding, with marked heterogeneity among trials. When the studies on ACS were excluded, the risk for major bleeding of significantly lower for apixaban (OR 0.69; 95% CI 0.61 to 0.79; p<0.001) and without significant heterogeneity (I2=1.84). Conversely, major bleeding complications occurred more frequently among patients receiving apixaban in ACS trials (OR 2.61; 95% CI 1.52 to 4.72; p<0.001) without significant heterogeneity (I2<0.001).

Figure 3

Risk for major bleeding complications.

All-cause mortality

Aside from ximelagatran, the use of dabigatran, rivaroxaban and apixaban was associated with the reduction in all-cause mortality (figure 4). Importantly, there was no significant heterogeneity among the trials.

Figure 4

Risk for all-cause mortality.

Funnel plot with Engger regression test did not show evidence for publication bias for the various outcomes (figure 5 showing data only for MI/ACS). Meta-regression analysis did not show any relationship between each antithrombotic agent and individual outcome measures, except for ximelagatran with MI/ACS (p=0.007), rivaroxaban with major bleeding complication (p<0.001) and apixaban with major bleeding complication (p=0.004).

Figure 5

Fixed-effects funnel plot with Engger regression test for the evaluation of publication bias for coronary events.


This meta-analysis showed that dabigatran was associated with increased risk for acute coronary events. Conversely, the greater likelihood for coronary events for the other oral direct thrombin inhibitor, ximelagatran, was not statistically significant. The excess risk associated with dabigatran was comparable to the findings of the earlier meta-analysis.3 Conversely, the risk for MI/ACS was lower for rivaroxaban, and a non-statistically significant reduction was observed for apixaban. Therefore, it appeared that the coronary risk differed between oral direct thrombin inhibitors and anti-Xa agents. Although the variation in the use of antiplatelet agents could have accounted for some of these differences, it was interesting to note that dabigatran was associated with a higher and rivaroxaban was associated with a lower risk for MI/ACS in clinical studies of ACS patients. Majority of them would have been treated with at least one antiplatelet agent. Therefore, based on these findings, those with heightened coronary risk, the use of anti-Xa agents may be preferable to direct thrombin inhibitors.

While both ximelagatran and low-molecular-weight heparins were able to reduce platelet activation, thrombin generation35 and endogenous thrombin potential,36 the time reduction for endogenous thrombin potential was greater for dalteparin compared with ximelagatran.36 Conversely, rivaroxaban was superior to dalteparin in preventing thrombin generation following hip and knee replacement surgery37 and reduces tissue factor induced platelet aggregation.38 In vitro studies indicated that direct thrombin inhibitors were associated with paradoxical coagulation compared with factor Xa inhibitors,39 which was likely mediated by preventing thrombin-induced activation of Protein C. This is a natural anticoagulant and part of the negative feedback system after thrombin generation. Furthermore, inflammatory markers were increased with long-term use of direct oral thrombin inhibitors.40 Urinary 11-dehydrothromboxane β2 was elevated for those receiving dabigatran compared with warfarin among 502 patients with atrial fibrillation and not treated with aspirin.41 But the preliminary results from a substudy of the RE-LY trial did not show this relationship.42 Nonetheless, taken together, the differences in thrombotic, inflammatory and platelet pathways could have accounted for some of the differences in coronary events. Furthermore, there was discordance in the main findings of SPORTIF III25 and SPORTIF V.26 Although both studies were similar in design, there were important dissimilarities. SPORTIF III25 was conducted in Europe, Asia plus Australasia and SPORTIF V26 was performed in North America. The design of the latter study26 was double-blind but SPORTIF III was an open-label trial.25 Of note, the primary endpoint, consisting of stroke and systemic embolism, was 2.3% per year for the warfarin group and 1.6% per year for the ximelagatran group in SPORTIF III.25 Conversely, it was 1.2% per year for the warfarin group and 1.6% per year for the ximelagatran group in SPORTIF V.26 There were also differences in the occurrence of major bleeding complications (figure 3A). The authors attributed the differences to better dose regulation, control of hypertension or hyperlipidaemia, other differences in patient characteristics or management or chance.26

Evaluation for a summarised risk for major bleeding complications among these studies has been challenging because of the marked variation in study protocol and endpoint definition (table 2). Although there was little difference in major bleeding complications for the four agents when compared with control, the rates were higher for rivaroxaban32 and apixaban33 ,34 in ACS patients, and influenced this outcome. Likely, several of these patients were receiving antiplatelet therapy, and probably treated with these two agents. Indeed, major bleeding complication rates have been noted to increase by 40–70% among those receiving aspirin plus clopidogrel in the RE-LY trial.43 Majority of these ACS patients were receiving dual antiplatelet agents. Not surprisingly, when these trials were excluded from analysis, evidence for heterogeneity was lost. Therefore, extreme caution has to be exercised when considering combining antiplatelet and antithrombotic agents because of the high bleeding risk.

Table 2

Definition of major bleeding complication and use of antiplatelet agents

Despite differences in the risks for MI/ACS and major bleeding complications, all-cause mortality was lower among those treated with dabigatran, rivaroxabn and apixaban compared with control. Better survival was also observed among patients with ACS treated with oral anticoagulation. All-cause and vascular mortalities were significantly lower among those receiving moderate intensity of warfarin plus aspirin compared to aspirin alone.44 Part of the reason for lower mortality for patients treated with novel antithrombotic agents may be related to the lower rates of haemorrhagic stroke for those with atrial fibrillation.27 ,28 ,30 If this finding is real then it may supersede the shortfalls of these agents such as lack of antidote for reversal of effects and assay to determine its therapeutic efficacy.

There are several limitations in the study. Importantly, there were differences in study population, protocol and procedures. Duration of follow-up varied considerably across trials. Individual patient outcome information was also not available. Definitions of outcome measures varied considerably in the studies and there were subjective elements in adjudication, especially for bleeding complications. Furthermore, not all the outcomes were reported in every trial. Silent MI may be actively sought out for in some studies, especially after revascularisation procedures, with routine electrocardiography or cardiac enzyme assays. But this approach may not be adopted in other trials. While this difference could have accounted for variation observed among studies, it was less likely to impact on the results within a study. Another limitation was that there was only one author in the study; there may be potential bias in study appraisal and selection stages. However, this concern is mitigated somewhat by relatively small total number of trials and fairly well-defined outcome measures. Nonetheless, these findings were instructive in providing insight on the relative occurrence adverse cardiovascular events impacting on the choice of these agents in specific patient subsets requiring anticoagulation. As with any results from meta-analyses, a firm conclusion can only be drawn from well-conducted, adequately powered randomised trials.



  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement There are no additional data available.