Oral anticoagulation with either vitamin K antagonists (VKAs) or direct oral anticoagulants
(DOACs) is very efficient in preventing atrial fibrillation (AF)-related thromboembolism.
Inevitably, anticoagulation increases the risk of bleeding, which may sometimes be
catastrophic. HAS-BLED (Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile International normalized ratio [INR], Elderly [> 65], Drugs/alcohol concomitantly) score[1]
[2] is widely used in clinical practice to stratify bleeding risk in patients with AF
receiving anticoagulation.
The risk of bleeding is even higher in AF patients who undergo percutaneous coronary
intervention (PCI), as these patients require dual antiplatelet therapy in addition
to anticoagulation.[3] The WOEST trial first confirmed the safety of dual antithrombotic therapy (DAT)
with VKA and clopidogrel after PCI compared with triple antithrombotic therapy with
VKA, aspirin, and clopidogrel, in a mixed population requiring anticoagulation.[4] In addition, large randomized trials[5]
[6]
[7]
[8] and meta-analyses[9] have demonstrated that DAT with a DOAC and a P2Y12 inhibitor, mostly clopidogrel,
is associated with a lower bleeding risk compared with a VKA-based triple antithrombotic
therapy in AF patients undergoing PCI. However, it is currently unclear whether the
baseline bleeding risk as estimated by HAS-BLED score is a treatment-effect modifier.
We investigated whether the reduction of the risk of bleeding with a DOAC-based DAT
compared with a VKA-based regimen in AF patients undergoing PCI depends on the baseline
bleeding risk as assessed with HAS-BLED score. For this purpose, we meta-analyzed
all four major randomized trials comparing antithrombotic therapy strategies in AF
patients undergoing PCI (PIONEER-AF PCI, REDUAL-PCI, AUGUSTUS, and ENTRUST-AF PCI).[5]
[6]
[7]
[8] The primary outcome of interest was the trial-defined primary bleeding event. Despite
some differences in the definition of bleeding in the above studies, the primary bleeding
event was a combination of major bleeding and clinically relevant nonmajor bleeding.
The trial-defined primary ischemic event was a secondary outcome in our meta-analysis.
For each trial, we recorded a risk estimate for the primary bleeding event and the
primary ischemic event in the DOAC-based versus the VKA-based subgroups, both in the
low bleeding risk (HAS-BLED ≤ 2) and in the high bleeding risk subgroups (HAS-BLED ≥ 3).
The risk estimates are reported as hazard ratio (AUGUSTUS) or dichotomous frequency
data (PIONEER-AF PCI, REDUAL PCI, and ENTRUST-AF PCI). We treated hazard ratio as
relative risk (RR). The random effects model was used to obtain the pooled RR. Risk
estimates between subgroups (HAS-BLED ≤ 2 and ≥ 3) were compared with a test of interaction.
All analyses were performed with Comprehensive Meta-Analysis Version 2 (Biostat, Englewood,
New Jersey, United States).
A total of 10,701 and 9,288 patients have been meta-analyzed for the primary and the
secondary outcome, respectively (as ENTRUST-AF PCI does not provide major ischemic
outcomes according to HAS-BLED score). The clinical data of the patients included
in the analysis as well as the absolute rates of the trial-defined primary bleeding
events stratified according to HAS-BLED score are shown in [Table 1]. In subjects with HAS-BLED ≤ 2, the cumulative RR for the trial-defined primary
bleeding event in the DOAC-based versus the VKA-based antithrombotic regimen was 0.57
(95% confidence intervals [CIs] 0.47 to 0.69), corresponding to a risk reduction of
43% with the DOAC-based therapy. In subjects with HAS-BLED ≥ 3, the respective cumulative
RR was 0.69 (95% CIs 0.61 to 0.78), corresponding to a reduction of 31% for the primary
bleeding event with the DOAC-based therapy ([Fig. 1A]). There was no significant difference of the RR estimates between the subgroups
of HAS BLED ≤ 2 and HAS-BLED ≥ 3 (Q value 3.03, p = 0.082). No significant heterogeneity across the four trials regarding the primary
bleeding event was observed (I
2 22.75%, p = 0.248). Sensitivity analysis after excluding ENTRUST AF - PCI trial confirmed the
findings, with a cumulative RRs of 0.57 (95% CIs 0.46 to 0.69) in the HAS BLED ≤ 2
subgroup and 0.66 (95% CIs 0.58 to 0.75) in the HAS-BLED ≥ 3 subgroup. There was no
significant change of the risk for the trial-defined primary ischemic event in the
DOAC-based versus the VKA-based regimen ([Fig. 1B]); similarly, no difference of the RR estimates between the two subgroups of HAS
BLED was observed (Q value 0.89, p = 0.35). We observed significant heterogeneity across the three trials regarding
the primary ischemic event (I
2 59.17%, p = 0.032). Bias analysis with funnel plots and the fail-safe N test showed absence
of significant publication bias as far as the primary bleeding event is concerned.
Table 1
Basic clinical characteristics and primary bleeding outcome of the included populations
|
Pioneer-AF PCI
|
REDUAL-PCI
|
AUGUSTUS
|
ENTRUST-AF PCI
|
|
Age (y)
|
70.1
|
70.2
|
70.7
|
69.5
|
|
ACS (%)
|
49.9
|
49.7
|
37.3
|
52
|
|
PCI (%)
|
100
|
100
|
76.1
|
100
|
|
Primary bleeding event (VKA, HAS-BLED ≤2) (%)
|
19.4
|
22.5
|
11.1% (odds ratio DOAC vs. VKA 0.59)
|
17.6
|
|
Primary bleeding event (VKA, HAS-BLED ≥3) (%)
|
25.9
|
28.1
|
13.3% (odds ratio DOAC vs. VKA 0.72)
|
21.9
|
|
Primary bleeding event (DOAC, HAS-BLED ≤2) (%)
|
13.0
|
13.3
|
11.1% (odds ratio DOAC vs. VKA 0.59)
|
10.9
|
|
Primary bleeding event (DOAC, HAS-BLED ≥3) (%)
|
16.7
|
19.9
|
13.3% (odds ratio DOAC vs. VKA 0.72)
|
20.1
|
Abbreviations: ACS, acute coronary syndrome; DOAC, direct oral anticoagulant; ISTH,
International Society on Thrombosis and Haemostasis; PCI, percutaneous coronary intervention;
TIMI, Thrombolysis In Myocardial Infarction; VKA, vitamin K antagonist.
Note: Primary bleeding event was TIMI major and TIMI minor and bleeding requiring
medical attention (PIONEER-AF PCI) or ISTH major and clinically relevant nonmajor
bleeding (AUGUSTUS, REDUAL PCI, ENTRUST AF).
Fig. 1 Cumulative relative risk and 95% confidence interval for the primary bleeding event
(A) and the primary ischemic event (B) in direct oral anticoagulant (DOAC)-based dual antithrombotic therapy compared with
vitamin K antagonist (VKA)-based triple antithrombotic therapy, stratified according
to the bleeding risk (HAS-BLED score).
Decision regarding the optimal antithrombotic regimen in AF patients taking anticoagulants
who need additional antiplatelet therapy after PCI comprises a challenge and is mainly
driven by estimating the balance between bleeding and ischemic risk.[3] Our meta-analysis shows that a DOAC-based DAT is associated with a lower risk of
bleeding not only in subjects with a high bleeding risk (HAS-BLED ≥ 3), but also in
patients with a lower bleeding risk (HAS-BLED ≤ 2). Although a few years ago guidelines
suggested triple therapy up to 6 months post-PCI in low-bleeding risk acute coronary
syndrome patients with AF,[10] our analysis clearly shows that DAT is beneficial without compromising efficacy
even in this subgroup. This is reflected in the recent 2020 European Society of Cardiology
Guidelines for the diagnosis and management of AF, which recommend that triple therapy
should not be taken for more than 1 month after PCI in acute coronary syndrome patients
with AF, independent of the bleeding risk.[11]
Apart from inherent limitations of the type of analysis performed, it is notable that
data obtained from the AUGUSTUS trial refer to a DOAC-based versus a VKA-based strategy
with both strategies including DAT and triple therapy subgroups. Moreover, 23.9% of
the AUGUSTUS patients included in our analysis underwent no PCI at baseline.[7] Meta-analysis for individual components of the primary bleeding or ischemic events
(i.e., intracranial hemorrhage or stent thrombosis, etc.) has not been performed as
none of the trials provide such data stratified according to HAS-BLED score.
With the foregoing limitations, we conclude that in AF patients undergoing PCI, a
DAT with DOAC plus P2Y12 inhibitor is safer compared with a VKA-based triple antithrombotic
therapy irrespectively of the baseline bleeding risk. It should be kept in mind that
bleeding risk is dynamic and may change over time, and that the change of HAS-BLED
score during follow-up may be more predictive for major bleeding than baseline HAS-BLED
score.[12] A high bleeding risk (HAS-BLED ≥ 3) should activate clinicians to address modifiable
bleeding risk factors and review the patients earlier and more frequently.[11]
