Short dual antiplatelet therapy followed by P2Y₁₂ inhibitor monotherapy vs. prolonged dual antiplatelet therapy after percutaneous coronary intervention with second-generation drug-eluting stents: a systematic review and meta-analysis of randomized clinical trials

Abstract

Aims

After percutaneous coronary intervention (PCI) with second-generation drug-eluting stent (DES), whether short dual antiplatelet therapy (DAPT) followed by single antiplatelet therapy (SAPT) with a P2Y₁₂ receptor inhibitor confers benefits compared with prolonged DAPT is unclear.

Methods and results

Multiple electronic databases, including PubMed, Scopus, Web of Sciences, Ovid, and ScienceDirect, were searched to identify randomized clinical trials comparing ≤3 months of DAPT followed by P2Y₁₂ inhibitor SAPT vs. 12 months of DAPT after PCI with second-generation DES implantation. The primary and co-primary outcomes of interest were major bleeding and stent thrombosis 1 year after randomization. Summary hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated by fixed-effect and random-effects models. Multiple sensitivity analyses including random-effects models 95% CI adjustment were applied. A sensitivity analysis comparing trials using P2Y₁₂ inhibitor SAPT with those using aspirin SAPT was performed. A total of five randomized clinical trials (32 145 patients) were available. Major bleeding was significantly lower in the patients assigned to short DAPT followed by P2Y₁₂ inhibitor SAPT compared with those assigned to 12-month DAPT (random-effects model: HR 0.63, 95% 0.45–0.86). No significant differences between groups were observed in terms of stent thrombosis (random-effects model: HR 1.19, 95% CI 0.86–1.65) and the secondary endpoints of all-cause death (random-effects model: HR 0.85, 95% CI 0.70–1.03), myocardial infarction (random-effects model: HR 1.05, 95% CI 0.89–1.23), and stroke (random-effects model: HR 1.08, 95% CI 0.68–1.74). Sensitivity analyses showed overall consistent results. By comparing trials testing ≤3 months of DAPT followed by P2Y₁₂ inhibitor SAPT vs. 12 months of DAPT with trials testing ≤3 months of DAPT followed by aspirin SAPT vs. 12-month of DAPT, there was no treatment-by-subgroup interaction for each endpoint. By combining all these trials, regardless of the type of SAPT, short DAPT was associated with lower major bleeding (random-effects model: HR 0.63, 95% CI 0.48–0.83) and no differences in stent thrombosis, all-cause death, myocardial infarction, and stroke were observed between regimens.

Conclusion

After second-generation DES implantation, 1–3 months of DAPT followed by P2Y₁₂ inhibitor SAPT is associated with lower major bleeding and similar stent thrombosis, all-cause death, myocardial infarction, and stroke compared with prolonged DAPT. Whether P2Y₁₂ inhibitor SAPT is preferable to aspirin SAPT needs further investigation.

Antiplatelet therapy, Aspirin, Ticagrelor, Percutaneous coronary intervention, Drug-eluting stent, Meta-analysis

See page 320 for the editorial comment on this article (doi: 10.1093/eurheartj/ehaa741) 

Introduction

After percutaneous coronary intervention (PCI) with drug-eluting stent (DES) implantation, dual antiplatelet therapy (DAPT) with aspirin and a P2Y₁₂ receptor inhibitor is the standard treatment to prevent stent thrombosis.^{1  ,  2} Findings from early studies assessing different DAPT durations and first-generation DES experience have led for a long time to the indiscriminate recommendation of 12 months or more of DAPT after stenting with DES.^{3  ,  4} In recent years, several randomized clinical trials testing shorter DAPT durations have suggested a comparable anti-thrombotic efficacy, as a result of the availability of DES with higher biocompatibility, and a possible benefit, due to reduced major bleeding incidence.^3–7 In light of these results, the European Society of Cardiology (ESC) guidelines have endorsed 6 months of DAPT after DES implantation in stable coronary artery disease.²

Whether shorter DAPT regimens after PCI with DES confer sufficient protection against ischaemic events and enhance the benefits deriving from a reduction in major bleeding is still unclear due to mixed results across trials and statistical power precluding inference on crucial individual endpoints. Moreover, the use of more potent P2Y₁₂ inhibitors — ticagrelor and prasugrel — has shown to reduce the incidence of major cardiovascular endpoints at the cost of a higher risk of bleeding events.^{1  ,  2} Single antiplatelet therapy (SAPT) with a P2Y₁₂ inhibitor is an appealing strategy that may have superior anti-thrombotic efficacy compared with aspirin and limit the need for DAPT after DES implantation.^{8  ,  9}

During the last year, a number of randomized clinical trials reported results addressing these questions by comparing short — 1–3 months — with prolonged DAPT regimens and testing SAPT with a P2Y₁₂ inhibitor in place of aspirin.^10–15 The primary objective of this study is to systematically review and synthetize available data from randomized clinical trials on ≤3 months of DAPT followed by P2Y₁₂ inhibitor SAPT vs. prolonged DAPT in patients who underwent PCI with second-generation DES. The secondary objective of this meta-analysis is to compare trials testing short DAPT followed by P2Y₁₂ inhibitor SAPT with trials testing short DAPT followed by aspirin SAPT and at the same time, by comprehensive combination of data, to provide general results on short DAPT vs. prolonged DAPT regardless of the type of SAPT.

Methods

This meta-analysis was conducted in keeping with the recommendations of PRISMA (Supplementary material online, Table S1) and Cochrane Collaboration.^{16  ,  17} The protocol was registered with PROSPERO (CRD42020182109).

Eligibility criteria

Studies were included in the meta-analysis when all the following pre-specified eligibility criteria were satisfied: (i) patients who underwent PCI; (ii) use of second-generation DES; (iii) random allocation of antiplatelet therapy regimens; (iv) comparison of different durations of DAPT; (v) ≤3 months of DAPT in the short DAPT arm and 12-month DAPT in the prolonged DAPT arm; and (vi) P2Y₁₂ inhibitor SAPT after DAPT interruption in the short DAPT arm.

Studies meeting all the eligibility criteria with the exception of the last one were included for the secondary analysis comparing the effects from trials employing P2Y₁₂ SAPT with those from trials employing aspirin SAPT. The same set of trials was also used to pool all available evidence on ≤3 months vs. 12 months of DAPT.

Search, endpoints, and qualitative assessment

Search strategy, study selection, data extraction, endpoints definition, and qualitative assessment methodology are extensively illustrated in the Supplementary material online, Appendix.

Statistical analysis

According to the original reports, categorical variables were reported as counts and proportions, while continuous variables as weighted means and weighted standard deviations, when overall cohort values were not available. According to original time-to-event analyses, outcomes between antithrombotic regimens were measured by using hazard ratios (HRs) and 95% confidence intervals (CIs). We estimated log HRs with corresponding standard errors for each trial and combined values by using fixed- and random-effects models with inverse variance weighting.¹⁸ When HRs were not reported, log HRs and standard errors were back-calculated from log-rank test P-values.¹⁹ Although one of the included trials was designed to assess superiority of short DAPT followed by ticagrelor-based SAPT at 2 years, in this meta-analysis only results at the landmark time point of 1 year were used to ensure uniform follow-up.^{10  ,  15} Confidence intervals of random-effects models were validated by adjustment according to the Hartung–Knapp method.²⁰ Prediction intervals for the treatment effect of a new study were estimated.²¹ Results were displayed by using forest plots illustrating the relative contribution to the summary estimate of individual trials. Heterogeneity between trials was measured according to τ  ² statistic and significance formally assessed by Q test.²² Heterogeneity was graded according to the I  ² statistic: values <25%, between 25% and 50%, and ≥50% indicated low, moderate, and high heterogeneity, respectively.²² The Baujat plot, which is a scatter plot showing on the y-axis the contribution of each trial to the Q statistic and on the x-axis the standardized difference of the overall treatment effect with and without each trial, was used to explore outcomes with high heterogeneity.²³ Sensitivity analyses by leave-one-out and cumulative inclusion of trials in a chronologic order were used; cumulative analyses were complemented with O’Brien-Fleming sequential monitoring boundaries when high heterogeneity was detected.¹⁷ Additional subgroup analyses with treatment-by-subgroup interaction testing were conducted with respect to aspirin interruption time and type of P2Y₁₂ inhibitor used in the P2Y₁₂ inhibitor SAPT group. Finally, an exploratory analysis was used to compare and combine trials included in this meta-analysis with trials of patients who underwent PCI with second-generation DES randomly assigned to short DAPT (≤3 months) followed by aspirin vs. prolonged (12 months) DAPT. Statistical analysis was conducted by using R 4.0.0 and STATA 14.

Results

Search strategy results and study selection process are illustrated in Supplementary material online, Figure S1 and Table S2. A total of five eligible randomized clinical trials were identified.^10–15 The full dataset of the GLOBAL LEADERS trial was used for main analyses and most of sensitivity analyses, as reported in the tables and figures illustrating the results.¹⁰ Data from the GLASSY trial, a pre-specified ancillary sub-study of the GLOBAL LEADERS trial including patients from 20 top-enrolling participating sites who underwent independent clinical events committee adjudication of investigator reported or eventually unreported events, were used to validate primary results across specific secondary analyses (Supplementary material online, Appendix).¹⁵

The main characteristics of trials are summarized in Table 1 and Supplementary material online, Table S3 and Figure S2. A total of 32 145 patients were available for the primary analysis, of whom 16 057 were assigned to short DAPT followed by P2Y₁₂ inhibitor SAPT and 16 088 to prolonged DAPT. In two trials aspirin was interrupted 1 month after PCI, while in the other three trials aspirin was interrupted 3 months after PCI.^10–15 The main characteristics of patients are summarized in Table 2 and Supplementary material online, Table S4. Mean age at PCI ranged from 61 years in the TICO trial to 69 years in the STOPDAPT-2 trial, prevalence of female gender ranged from 20.5% in the TICO trial to 26.6% in the SMART-CHOICE trial, and prevalence of diabetes ranged from 25.3% in the GLOBAL LEADERS trial to 38.5% in the STOPDAPT-2 trial.^10–15 From approximately 40% to approximately 65% of patients enrolled in the GLOBAL LEADERS, SMART-CHOICE, STOPDAPT-2, and TWILIGHT trials and all patients (100%) included in the TICO trial were admitted for acute coronary syndrome.^10–15 In the STOPDAPT-2 trial, the number of lesions per patient was low and the use of endovascular imaging very high compared with the other trials.¹² In the TICO trial, radial access was less frequent compared with the other trials in which proportions ranged from 72.8% to 83.0%.¹⁴

Table 1

Design of randomized clinical trials included in the meta-analysis

	GLOBAL LEADERS		SMART CHOICE		STOPDAPT-2		TWILIGHT		TICO
	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT
Patients	15 968		2993		3009		7119		3056
	7980	7988	1495	1498	1500	1509	3555	3564	1527	1529
Randomization time	In-hospital		0–3 months		In-hospital		3 months, event-free		In-hospital
Before DAPT discontinuation in the SAPT arm	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT
	Ticagrelor plus aspirin	Clopidogrel (stable CAD) or ticagrelor (unstable CAD) plus aspirin	Clopidogrel, prasugrel, or ticagrelor plus aspirin		Clopidogrel or prasugrel plus aspirin		Ticagrelor plus aspirin		Ticagrelor plus aspirin
Aspirin interruption time	1 month		3 months		1 month		3 months		3 months
Treatment allocated by randomization	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT	P2Y12-I SAPT	DAPT
	Ticagrelor	Clopidogrel (stable CAD) or ticagrelor (unstable CAD) plus aspirin	Clopidogrel, prasugrel, or ticagrelor	Clopidogrel, prasugrel, or ticagrelor plus aspirin	Clopidogrel	Clopidogrel plus aspirin	Ticagrelor plus placebo	Ticagrelor plus aspirin	Ticagrelor	Ticagrelor plus aspirin
Centres	130		33		90		187		38
Region	Austria, Belgium, Bulgaria, Denmark, France, Germany, Hungary, Italy, Netherlands, Poland, Portugal, Spain, Switzerland, United Kingdom, Australia, Brazil, Canada, Singapore		South Korea		Japan		United States, Canada, Austria, Germany, Israel, Italy, Poland, Spain, United Kingdom, China, India		South Korea
Enrolment	July 2013 to November 2015		March 2014 to July 2017		December 2015 to December 2017		July 2015 to December 2017		August 2015 to October 2018
Primary analysis time after PCI	24 monthsa		12 months		12 months		15 months		12 months
Primary endpoint	All-cause death or new Q-wave myocardial infarction		All-cause death, myocardial infarction, or stroke		Cardiovascular death, myocardial infarction, definite stent thrombosis, ischaemic or haemorrhagic stroke, or TIMI major or minor bleeding		BARC 2, 3, or 5		All-cause death, myocardial infarction, stent thrombosis, stroke, target vessel revascularization, or TIMI major bleeding
Hypothesis	Superiority		Non-inferiority		Non-inferiority		Superiority		Non-inferiority
Masking	Open-label		Open-label		Open-label		Double-blind		Open-label
CEC	Nob		Yes		Yes		Yes		Yes
DSMB	Yes		Yes		Yes		Yes		Yes
Registrationc	NCT01813435		NCT02079194		NCT02619760		NCT02270242		NCT02494895

BARC, Bleeding Academic Research Consortium; CEC, Clinical Events Committee; DAPT, dual antiplatelet therapy; DSMB, Data and Safety Monitoring Board; P2Y₁₂-I, P2Y₁₂ inhibitor; PCI, percutaneous coronary intervention; SAPT, single antiplatelet therapy; TIMI, thrombolysis in myocardial infarction.

a

Only outcomes at 12-month follow-up of the GLOBAL LEADERS trial were pooled in this meta-analysis.

b

This important limitation of the GLOBAL LEADERS trials was mitigated by the GLASSY ancillary sub-study. Sensitivity analyses by including only the GLASSY dataset are reported in the Supplementary material online, Appendix.

c

clinicaltrials.gov.

Table 2

Main characteristics of patients across included trials

	GLOBAL LEADERS	SMART CHOICE	STOPDAPT-2	TWILIGHT	TICO
Age	64.6 ± 10.3	64.5 ± 10.7	68.6 ± 10.7	65.1 ± 10.3	61 ± 11
Female	3714 (23.3)	795 (26.6)	672 (22.3)	1698 (23.9)	628 (20.5)
BMI	28.2 ± 4.6	24.6 ± 3.2	24.3 ± 3.5	28.6 ± 5.6	24.9 ± 3.3
Region
Europe	15 325 (96.0)	0	0	2509 (35.2)	0
North America	170 (1.1)	0	0	2972 (41.7)	0
South America	248 (1.6)	0	0	0	0
Asia	142 (0.9)	2993 (100)	3009 (100)	1638 (23.0)	3056 (100)
Australia	83 (0.5)	0	0	0
Diabetes	4038 (25.3)	1122 (37.5)	1159 (38.5)	2620 (36.8)	835 (27.3)
Hypertension	11 715 (73.6)	1840 (61.5)	2221 (73.8)	5154 (72.4)	1541 (50.4)
Dyslipidaemia	10 768 (69.6)	1352 (45.2)	2244 (74.6)	4303 (60.4)	1846 (60.4)
Current smoker	4169 (26.1)	791 (26.4)	710 (23.6)	1548 (21.8)	1142 (37.4)
Prior myocardial infarction	3710 (23.3)	127 (4.2)	406 (13.5)	2040 (28.7)	113 (3.7)
Prior PCI	5221 (32.7)	349 (11.7)	1032 (34.3)	2998 (42.1)	262 (8.6)
Prior CABG	943 (5.9)		59 (2.0)	710 (10.0)	18 (0.6)
LVEF	―	59.9 ± 10.8	59.8 ± 10.4	―	―
Prior stroke	421 (2.6)	201 (6.7)	186 (6.2)	―	126 (4.1)
Peripheral vascular disease	1005 (6.4)	―	196 (6.5)	489 (6.9)	―
Chronic kidney diseasea	2171 (13.7)	97 (3.2)	166 (5.5)	1145 (16.8)	620 (20.3)
Clinical presentation
Stable CAD	8481 (53.1)	1250 (41.8)	1861 (61.8)	2503 (35.2)	0
Unstable angina	2022 (12.7)	958 (32.0)	407 (13.5)	2494 (35.0)	926 (30.3)
NSTEMI	3373 (21.1)	469 (15.7)	180 (6.0)	2120 (29.8)	1027 (33.6)
STEMI	2092 (13.1)	314 (10.5)	561 (18.6)	0	1103 (36.1)
Multivessel CAD	―	1483 (49.5)	―	4466 (62.7)	1703 (55.7)
Multivessel PCI	2346 (14.7)	705 (23.6)	216 (7.2)	―	520 (17.0)
Lesions treated	20841	3734	3187	9287	3779
Lesions site	Per-lesion	Per-lesion	Per-patient	Per-lesion	Per-lesion
LM	387 (1.9)	58 (1.6)	80 (2.7)	353 (5.0)	94 (2.5)
LAD	8666 (41.6)	1853 (49.6)	1682 (55.9)	4003 (56.2)	1821 (48.2)
LCx	5077 (24.4)	775 (20.8)	573 (19.0)	2297 (32.3)	711 (18.8)
RCA	6490 (31.1)	1048 (28.1)	846 (28.1)	2500 (35.1)	1153 (30.5)
Graft	221 (1.1)	0	6 (0.2)	134 (1.9)	0
Radial access	11761 (74.0)	2182 (72.9)	2496 (83.0)	5186 (72.8)	1698 (55.6)
Drug-eluting stent type	Per-lesion	Per-patient	Per-patient	Per-patient	Per-patient
Everolimus-eluting CoCr DP	―	1051 (35.1)	3009 (100)	second-generation   6958 (97.7)b	0
Everolimus-eluting PtCr BP	―	967 (32.3)	0		0
Sirolimus-eluting CoCr BP	―	972 (32.5)	0		3056 (100)
Zotarolimus-eluting CoCr DP	―	1 (<0.1)	0		0
Biolimus-eluting SS	19415 (94.6)c	0	0		0
Other	―	1 (<0.1)	0		0
Stent length	Per-lesion	Per-patient	Per-patient	Per-patient	Per-lesion
	24.8 ± 14.0	37.9 ± 22.7	30.4 ± 16.7	39.9 ± 24.3	28.1 ± 12.8
P2Y12-I at discharge
Clopidogrel	―	2312 (77.2)	1852 (61.6)	0	0
Prasugrel	―	129 (4.3)	1151 (38.3)	0	0
Ticagrelor	―	552 (18.4)	0	7119 (100)	3056 (100)
Ticlopidine	―	0	4 (0.1)	0	0

Data are number (proportion) or mean ± standard deviation.

BP, Bioresorbable-polymer; CABG, Coronary artery bypass grafting; CAD, Coronary artery disease; CoCr, Cobalt-chromium; DP, Durable-polymer; LAD, Left anterior descending; LCx, Left circumflex; LM, Left main; LVEF, Left ventricular ejection fraction; NSTEMI, Non-ST-segment elevation myocardial infarction; P2Y₁₂-I, P2Y₁₂ inhibitor; PCI, Percutaneous coronary intervention; PtCr, Platinum-chromium; RCA, Right coronary artery; SS, Stainless steel; STEMI, ST-segment-elevation myocardial infarction.

a

GLOBAL LEADERS, SMART-CHOICE, TWILIGHT, and TICO trials = Estimated glomerular filtration <60 mL/min/1.73 m²; STOPDAPT-2 trial = Estimated glomerular filtration <30 mL/min/1.73 m².

b

Proportions among second-generation stents implanted were not disclosed.

c

The type of other stents implanted was not disclosed.

Major bleeding

As illustrated in Figure 1, patients assigned to P2Y₁₂ inhibitor SAPT had significantly lower major bleeding compared with those assigned to prolonged DAPT by both fixed-effect (HR 0.69, 95% CI 0.58–0.82, P < 0.0001) and random-effects models (HR 0.63, 95% CI 0.45–0.86, P = 0.004). Regardless of the model used, the GLOBAL LEADERS trial provided the highest relative contribution to summary estimate (fixed-effect: 46.9%; random-effects: 29.3%), followed by the TICO (fixed-effect: 24.9%; random-effects: 25.2%) and TWILIGHT trials (fixed-effect: 18.3%; random-effects: 22.8%). After conservative adjustment of the 95% CI obtained by random-effects model, summary estimate became borderline non-significant (HR 0.63, 95% CI 0.39–1.01).

Figure 1

Major bleeding. CI confidence interval; DAPT, dual antiplatelet therapy; HR, hazard ratio; P2Y₁₂-I SAPT, P2Y₁₂ inhibitor single antiplatelet therapy; W_Fixed, relative weight according to fixed-effect model; W_Random, relative weight according to random-effects model. Events were adjudicated according to the Bleeding Academic Research Consortium (BARC) type 3 or 5 definitions.

Heterogeneity was high (I  ² = 61.8%) and prediction interval (0.23–1.72) described the uncertainty in the point estimate collocation of a future trial. As illustrated by Baujat plot (Figure 2, left), the two trials providing the larger contribution to global heterogeneity were the GLOBAL LEADERS and STOPDAPT-2. The latter had the lowest relative weight and highest magnitude of corresponding effect, while the GLOBAL LEADERS trial had the largest relative weight and neutral corresponding effect. The relative distribution of heterogeneity was quantified by omitting each trial at a time. (Figure 2, right upper). By using random-effects models with conservative 95% CI adjustment, given the strong differences in relative weight, the omission of the GLOBAL LEADERS trial led to a significant reduction of global heterogeneity (I  ² = 36.0%) and a significant effect despite conservative 95% CI adjustment (HR 0.55, 95% CI 0.31–0.97), while the omission of the STOPDAPT-2 trial led only to a modest reduction of global heterogeneity (I  ² = 49.3%) and still non-significant summary estimate. Of note, the omission of one of the other trials at a time in any case did not significantly reduce the magnitude of heterogeneity (I  ² > 50%, τ  ² > 0.04) and continued to produce non-significant summary estimates. By adding trials in a chronologic order, cumulative meta-analysis showed that the TWILIGHT trial allowed to reach both the conventional thresholds (z = 1.96) and the α-spending function boundary (z = 2.14) of statistical significance (Figure 2, right lower).

Figure 2

Assessment of heterogeneity in major bleeding. CI, confidence interval; HR, hazard ratio; P = P-value after Hartung–Knapp correction; P  _Q, Q test P-value; P2Y₁₂-I SAPT, P2Y₁₂ inhibitor single antiplatelet therapy. Left: Baujat plot. The GLOBAL LEADERS trial provided the highest contribution to summary estimate and significantly influenced global heterogeneity. The STOPDAPT-2 trial effect introduced significant heterogeneity but the relative impact on summary estimate was low. Right, upper: The leave-one-out sensitivity analysis by random-effects models with 95% CI adjustment showed that by omitting the GLOBAL LEADERS trial heterogeneity became significantly lower (I  ² = 36.0%) and summary estimate became statistically significant (HR 0.55, 95% CI 0.31–0.97, P = 0.044), while by omitting the STOPDAPT-2 trial heterogeneity was mildly reduced and summary estimate remained not significant. By omitting each one of the other trials at a time, in each case heterogeneity remained high (I  ² > 50%, τ  ²>0.04) and the corresponding summary estimate was not statistically significant. Estimates were obtained by random-effects model with 95% CI adjustment according to the Hartung–Knapp method.²⁰  Right, lower: Cumulative analysis by adding each trial in chronological order. Trial-level estimates were pooled by random-effects model. Evidence added by the TWILIGHT trial led cumulative evidence to cross both the conventional threshold of statistical significance (z = 1.96) and the α-spending monitoring boundary accounting for multiple statistical testing (z = 2.14).

Stent thrombosis

As described in Figure 3, stent thrombosis was not significantly different between antithrombotic strategies (random-effects model: HR 1.19, 95% CI 0.86–1.65). The incidences of stent thrombosis were not significantly different between P2Y₁₂ inhibitor SAPT and prolonged DAPT arms in each trial ranging from 0.2% vs. 0.1% in the SMART-CHOICE trial to 0.7% vs. 0.5% in the GLOBAL LEADERS trial and 0.4% vs. 0.6% in the TWILIGHT trial. The GLOBAL LEADERS trial had the highest relative weight (64.7%) followed by the TWILIGHT trial (22.8%). Heterogeneity was not detectable (I  ² = 0%). Individual trials had no significant influence on summary estimate (Supplementary material online, Table S5).

Figure 3

Stent thrombosis. CI, confidence interval; DAPT, dual antiplatelet therapy; HR, hazard ratio; P2Y₁₂-I SAPT, P2Y₁₂ inhibitor single antiplatelet therapy; W_Fixed, relative weight according to fixed-effect model; W_Random, relative weight according to random-effects model.

Secondary endpoints

Secondary endpoints are illustrated in Figure 4.

Figure 4

Secondary endpoints. CI, confidence interval; DAPT, dual antiplatelet therapy; HR, hazard ratio; P2Y₁₂-I SAPT, P2Y₁₂ inhibitor single antiplatelet therapy; W_Fixed, relative weight according to fixed-effect model; W_Random, relative weight according to random-effects model.

All-cause death (random-effects model: HR 0.85, 95% CI 0.70–1.03) and myocardial infarction (random-effects model: 1.05, 95% CI 0.89–1.23) did not significantly differ between antithrombotic treatments. Heterogeneity was not detectable for either of the two endpoints (I  ² = 0). Stroke did not differ between antithrombotic treatments, regardless of the model used (random-effects model: HR 1.08, 95% CI 0.68–1.74), with a moderate degree of heterogeneity (I  ² = 47.8%).

Sensitivity analyses

Individual trial influence on secondary endpoints was not significant (Supplementary material online, Table S5).

The comparison between trials with aspirin interruption at 1 month with those with aspirin interruption at 3 months (Figure 5) and between trials using SAPT with thienopyridine with those using SAPT with ticagrelor (Supplementary material online, Figure S3) did not reveal inconsistency.

Figure 5

Sensitivity analyses according to aspirin interruption time. CI, confidence interval; DAPT, dual antiplatelet therapy; HR, hazard ratio; P2Y₁₂-I SAPT, P2Y₁₂ inhibitor single antiplatelet therapy; P = Wald test; P  _int, treatment-by-subgroup interaction; P  _Q,  Q test; W_Random, relative weight according to random-effects model. Trials included in the primary analysis were pooled. Major bleeding events were adjudicated according to the Bleeding Academic Research Consortium (BARC) type 3 or 5 definitions.

Take home figure

Short vs. prolonged dual antiplatelet therapy with stratification according to the type of single antiplatelet therapy used in the short dual antiplatelet therapy group. CI, confidence interval; DAPT, dual antiplatelet therapy; HR, hazard ratio; P2Y₁₂-I SAPT, P2Y₁₂ inhibitor single antiplatelet therapy; P, Wald test; P  _int, treatment-by-subgroup interaction; P  _Q, Q test; W_Random, relative weight according to random-effects model. Trials included in the P2Y₁₂ inhibitor group (GLOBAL LEADERS, SMART-CHOICE, STOPDAPT-2, TWILIGHT, and TICO) randomized patients to 1-3 months of DAPT followed by P2Y₁₂ inhibitor monotherapy vs. 12 months of DAPT.^{10  ,  11  ,  12  ,  13  ,  14} Trials included in the aspirin group (RESET, OPTIMIZE, REDUCE) randomized patients to 3 months of DAPT followed by aspirin monotherapy vs. 12 months of DAPT.^24–26

Replacement of the GLOBAL LEADERS trial with the GLASSY sub-study

As illustrated in Supplementary material online, Figures S4–S6, after replacement of the GLOBAL LEADERS trial with the GLASSY sub-study, main analysis results remained consistent. Leave-one-out analyses with the GLASSY sub-study did not change main conclusions (Supplementary material online, Table S6).

P2Y₁₂ inhibitor vs. aspirin single antiplatelet therapy and short vs. prolonged dual antiplatelet therapy

Three trials of patients randomly assigned to short DAPT (3 months) followed by aspirin with prolonged DAPT for 12 months after PCI with second-generation DES were identified (Supplementary material online, Figure S1 and Tables S7–S9).^24–26 Only 12-month outcomes were pooled.^24–26 The comparison of trials employing SAPT with a P2Y₁₂ inhibitor with those employing traditional SAPT with aspirin did not show significant treatment-by-subgroup interaction and summary estimates between subgroups showed consistent magnitude and direction for all the endpoints assessed (Take home figure). By combining trials testing P2Y₁₂ inhibitor SAPT with those using conventional aspirin-based SAPT, short DAPT reduced major bleeding (HR 0.63, 95% CI 0.48–0.83) without increasing stent thrombosis (HR 1.21, 95% CI 0.91–1.62), all-cause death (HR 0.90, 95% CI 0.73–1.11), myocardial infarction (HR 1.06, 95% 0.92–1.23), and stroke (HR 1.04, 95% CI 0.75–1.43).

Qualitative assessment

Individual trial risk of bias due to design and methodology was overall low (Supplementary material online, Figure S7). The TWILIGHT trial showed the highest quality mainly due to placebo control.

Discussion

The main results of this meta-analysis can be summarized as follows:

1. One to 3 months of DAPT followed by P2Y₁₂ inhibitor SAPT reduces major bleeding compared with prolonged DAPT;

2. This advantage does not occur at the cost of an increase in stent thrombosis and major endpoints including all-cause death, myocardial infarction, and stroke;

3. High heterogeneity exists among trials with respect to the endpoint of major bleeding and conservative summary estimate 95% CI adjustment reveals that more data are needed;

4. Whether short-term DAPT followed by P2Y₁₂ inhibitor SAPT is associated with different clinical outcomes compared with short-term DAPT followed by aspirin SAPT requires further evaluation.

This meta-analysis shows that short DAPT followed by P2Y₁₂ inhibitor SAPT is associated with a 31–37% relative risk reduction in major bleeding compared with prolonged DAPT. Reduced long-term major bleeding after PCI in patients assigned to P2Y₁₂ inhibitor SAPT remained consistent across sensitivity analyses, while the application of a conservative model accounting for the high heterogeneity observed resulted in a borderline decrease (P = 0.052). In our meta-analysis, we focused on major bleeding since the prognostic impact of minor events is less pronounced^27–30 In prior studies, major bleeding has resulted to have strong association with survival, but in our meta-analysis we did not detect a significant reduction in all-cause death despite the considerable reduction in major bleeding. Although the reasons for this are unclear and likely include the influence of competing risk due to non-bleeding events, it might also be possible that patient selection across trials generated a relatively resilient population requiring more statistical power to detect a concomitant significant reduction in all-cause death.

Of note, regardless of some differences in ischaemic risk profile, clinical presentation, and coronary artery disease pattern across trials, there was no significant difference in stent thrombosis and myocardial infarction at 12–15 months between antithrombotic strategies, without significant between-trial heterogeneity and with consistent summary estimates across sensitivity analyses. Coupled with the significant reduction in major bleeding, these results may support a wider application in clinical practice of 1–3 months of DAPT followed by P2Y₁₂ inhibitor SAPT after uncomplicated PCI with DES and uneventful progress within the short DAPT period. However, some caution is required when considering the generalizability of the results to the acute myocardial infarction setting. Indeed, estimates according to stable and unstable coronary artery disease were not uniformly disclosed across original reports and when available related only to the primary composite endpoint of ischaemic and/or bleeding events.^10–15 On the other hand, trials including high proportions or only patients with acute coronary syndrome did not show signals of harm from short DAPT and, with the exception of the TWILIGHT trial, all trials included patients admitted for ST-segment elevation myocardial infarction with rates up to ∼36% (TICO trial).^10–15 Individual patient data analysis might permit further exploration of risks and benefits not only in the subgroups of patients with acute coronary syndrome but also in those with stable coronary artery disease. Indeed, it would be interesting to test the significant reduction in major bleeding observed after short DAPT followed by P2Y₁₂ inhibitor SAPT at the landmark time point of 6 months in patients with low bleeding risk profile and non-complex, stable coronary artery disease under the consideration that current European guidelines recommend 6 months of DAPT in this subset.²

One of the most important findings of our meta-analysis is the high heterogeneity observed in terms of major bleeding. Although it might be supposed that differences in trial design, individual bleeding risk propensity of patients (age, gender, biometrics, ethnicity, etc.), and P2Y₁₂ inhibitor SAPT composition (clopidogrel, prasugrel, or ticagrelor) are sufficient to explain the significant between-trial heterogeneity, by qualitative review of available data, we found no evident explanation and multiple measured and unmeasured conditions have likely interplayed. Indeed, in trials presenting comparable proportions of known bleeding risk conditions contrasting results were observed. Conversely, by comparing trials with pronounced differences in bleeding risk conditions distribution, the risk reduction in major bleeding associated with short DAPT was consistent. Interestingly, we found that the GLOBAL LEADERS trial — largest sample size, mostly European patients — showed a neutral effect between treatment strategies with respect to major bleeding. In contrast, in the TWILIGHT trial — second largest sample size, multiethnic study population — patients assigned to ticagrelor monotherapy exhibited a significant 51% relative risk reduction. The heterogeneity in major bleeding observed across trials presents a further argument for an individual patient data analysis of available evidence to define key clinical, angiographic, and procedural characteristics enabling reproducible results in daily practice. In addition, a large-scale trial designed to prove with a reasonably small margin the non-inferiority of 1–3 months of DAPT to 12 months of DAPT in patients with non- and ST-segment acute myocardial infarction with respect to individual major ischemic endpoints is still lacking. 

Our exploratory analysis comparing trials testing P2Y₁₂ inhibitor SAPT vs. trials using conventional aspirin-based SAPT showed no treatment-by-subgroup interaction for all the endpoints explored, with substantial consistency in magnitude and direction of point estimates. By combining trials testing P2Y₁₂ inhibitor SAPT with those using conventional aspirin-based SAPT with the objective of performing a meta-analysis on short vs. prolonged DAPT duration, 1–3 months of DAPT followed by any-type SAPT reduced major bleeding without increasing stent thrombosis, all-cause death, myocardial infarction, and stroke at 12–15 months. In light of these findings, new analyses should provide clarification on whether the reduction in major bleeding observed in patients assigned to P2Y₁₂ inhibitor SAPT mainly depends on the short DAPT duration or the type of SAPT. However, studies comparing aspirin or clopidogrel with ticagrelor monotherapy in other clinical settings did not find significant differences between treatments in terms of major bleeding and a recent meta-analysis of trials comparing aspirin with P2Y₁₂ inhibitor SAPT for secondary prevention in patients with established atherosclerosis showed a quite comparable risk of major bleeding between groups (OR 0.90, 95% CI 0.74–1.10).^{31  ,  32} Against this background, the aspect that may change the equipoise in SAPT selection might be the known drug resistance, intolerance (6–20%), and hypersensitivity (0.6–2.4%) of aspirin and the presence of non-responders to clopidogrel.^{8  ,  33  ,  34} SAPT with a newer P2Y₁₂ inhibitor indeed might improve predictability in individual response and be associated with a favourable profile of tolerance.

Limitations

The results of this meta-analysis have to be interpreted based on the following limitations. First, included trials showed some differences with respect to early phase design, timing of randomization, timing of aspirin interruption, and blinding in the administration of treatments. However, sensitivity analyses significantly mitigated concerns related to individual trial characteristics. The different timing of aspirin interruption in the SAPT arm across trials had limited impact on results as showed in the sensitivity analysis comparing outcomes between trials with SAPT initiation at 1 month with those at 3 months, while leave-one-out analysis showed that design of the TWILIGHT trial — only uneventful patients after 3 months of DAPT — did not significantly influence conclusions. Second, reporting of outcomes showed some differences. In particular, in the GLOBAL LEADERS trial, only definite stent thrombosis was available and in the TWILIGHT trial stroke was reported individually only as ischaemic.^{10  ,  13} However, leave-one-out analysis showed the limited individual influence of these studies for the endpoints of stent thrombosis and stroke, respectively. In addition, in the GLASSY sub-study definite or probable stent thrombosis events were adjudicated and the sensitivity analysis by replacement of the GLOBAL LEADERS trial data showed overall consistent findings.¹⁵ With respect to the analysis on DAPT followed by aspirin- vs. P2Y₁₂-based monotherapy, in the REDUCE trial major bleeding at 12 months was not reported.²⁶ Third, the use of different P2Y₁₂ inhibitors might have introduced heterogeneity. However, we performed a sensitivity analysis according to P2Y₁₂ inhibitor SAPT composition—thienopyridine or ticagrelor — and no significant interaction was observed for all the endpoints. Fourth, individual trials were powered only for composite endpoints. The substantial inconsistency in composite endpoints composition, definition, and reporting across trials did not allow pooling data with respect to these outcomes. However, one of the main reasons for performing a meta-analysis is the need for statistical power enhancement. The large number of patients included enabled the comparison between antithrombotic strategies in terms of individual endpoints. Fifth, although our findings can be applied to a large proportion of patients undergoing PCI, eligibility criteria used across trials led to exclusion or underrepresentation of specific, high ischaemic risk subsets. It is therefore likely that our results are not applicable to all patients undergoing PCI and prolonged regimens could be more protective in complex clinical and anatomic settings. On the other hand, eligibility criteria across trials have also limited the inclusion of high bleeding risk patients and benefits of short DAPT may be even more pronounced in this subset. A tailored approach in high ischaemic and/or bleeding risk patients should be preferred. Finally, about one-third of patients included across trials were enrolled in Asiatic centres. Whether this subgroup of patients have influenced the benefit observed in terms of major bleeding needs to be explored by further analysis.

Conclusion

In patients who underwent PCI with second-generation DES implantation, 1–3 months of DAPT followed by monotherapy with a P2Y₁₂ inhibitor reduces major bleeding at long-term follow-up without increasing stent thrombosis. Secondary analyses suggest no differences in all-cause death, myocardial infarction, and stroke between regimens. By also including trials using short DAPT followed by aspirin monotherapy, regardless of the type of SAPT used, short-term DAPT is associated with lower major bleeding without increasing any of major cardiovascular endpoints. Whether P2Y₁₂ inhibitor SAPT is preferable to aspirin SAPT needs further investigation.

Supplementary material

Supplementary material is available at European Heart Journal online.

Conflict of interest: G.G. reports personal fees from Daiichi Sankyo. R.A.B. reports research funding to the institution of prior employment from Celonova Biosciences; D.C. reports speaker or consulting fees from Amgen, Bayer, AstraZeneca, Boehringer, Sanofi, Boston Scientific, Abbott. M.V. reports personal fees from Astra Zeneca, Terumo, Alvimedica/CID, Abbott Vascular, Daiichi Sankyo, Bayer, CoreFLOW, Idorsia, Universität Basel—Klinische Forschung, Bristol Myers Squib SA, Medscape, Vesalio, and grants from Terumo; R.M. reports receiving consulting fees from Abbott Vascular Laboratories, Boston Scientific, Medscape/WebMD, Siemens Medical Solutions, Phillips/Volcano/Spectranetics, Roviant Sciences, Sanofi Italy, Bracco Group, Janssen, and AstraZeneca, grant support, paid to her institution, from Bayer, CSL Behring, DSI, Medtronic, Novartis Pharmaceuticals, OrbusNeich, Osprey Medical, PLC/RenalGuard, and Abbott Vascular, grant support and advisory board fees, paid to her institution, from BMS, fees for serving on a data and safety monitoring board from Watermark Research Funding, advisory fees and lecture fees from Medintelligence (Janssen), and lecture fees from Bayer; G.T. is proctor for Boston Scientific and received lecture fees from Edwards Lifesciences and Medtronic. The remaining authors have no disclosures.