Efficacy and safety of dual antiplatelet therapy de-escalation in East Asian individuals following percutaneous coronary intervention for acute coronary syndrome: a systematic review and meta-analysis

Abstract

Background

East Asian patients with acute coronary syndrome (ACS) are more prone to bleeding complications than Western populations under comparable antithrombotic regimens. This pattern presents a considerable obstacle to the routine use of intensive dual antiplatelet therapy (DAPT) undergoing percutaneous coronary intervention (PCI). Consequently, the effectiveness and safety of DAPT de-escalation strategies tailored to this particular group of patients remain inadequately defined.

Methods

An extensive search of both Chinese and English literature databases was performed up to June 2025 to locate cohort studies and randomized controlled trials assessing the effectiveness and safety of de-escalating DAPT in East Asian individuals diagnosed with ACS. The primary outcomes included major adverse cardiovascular events (MACE), defined as a composite of cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke, and bleeding outcomes, with bleeding classified according to Bleeding Academic Research Consortium (BARC) criteria. Meta-analytical methods were applied using either fixed-effects or random-effects models, depending on data suitability.

Results

A total of nine studies were included in the analysis, consisting of five RCTs and four cohort studies, encompassing 10,263 patients overall. Compared to standard DAPT, the de-escalation approach was significantly linked to lower risks of BARC type 1 bleeding (RR = 0.52; 95% CI: 0.32–0.86; p = 0.011), BARC type ≥ 2 bleeding (RR = 0.47; 95% CI: 0.37–0.60; p < 0.001), and BARC type ≥ 3 events (RR = 0.61; 95% CI: 0.40–0.94; p = 0.026). Nonetheless, the comparison between de-escalation and standard regimens revealed no statistically meaningful difference in MACE risk (RR = 0.91; 95% CI: 0.70–1.19; p = 0.497). This result remained stable across the separate elements comprising the MACE composite outcome.

Conclusion

Among East Asian patients with ACS, implementing a DAPT de-escalation strategy lowers bleeding risk without elevating ischemic events, suggesting a potentially favorable safety profile in East Asian patients. Further validation by RCTs is warranted.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-026-05527-z.

Introduction

Acute coronary syndrome (ACS) continues to rank among the leading global causes of mortality [1]. In recent years, percutaneous coronary intervention (PCI) has become a cornerstone therapy for ACS, improving clinical outcomes; however, in-stent thrombosis and recurrent ischemic events remain important challenges after PCI [2, 3]. Accordingly, dual antiplatelet therapy (DAPT)—comprising aspirin in combination with a P2Y12 receptor antagonist, has been widely adopted as the standard post-PCI regimen, effectively lowering the rate of cardiovascular complications [4]. Potent P2Y12 inhibitors (ticagrelor and prasugrel) provide faster and more consistent platelet inhibition than clopidogrel, but at the cost of increased bleeding risk [5, 6].

This ischemia–bleeding trade-off may be particularly pronounced in East Asian patients, who tend to have fewer ischemic events but a higher bleeding risk under standard-dose DAPT (the ‘East Asian paradox’) [7]. To mitigate the balance between enhanced antiplatelet efficacy and bleeding complications, growing interest has been directed toward DAPT de-escalation strategies. This strategy typically involves administering potent antiplatelet therapy during the acute, high-risk phase post-PCI in patients with ACS, followed by a stepwise reduction in the intensity or dosage of antiplatelet agents (e.g., de-escalating from ticagrelor to clopidogrel) once the patient’s clinical condition has stabilized, with the aim of mitigating long-term bleeding risk [8]. Multiple clinical studies in Western cohorts, including the TROPICAL-ACS trial, have confirmed both the effectiveness and safety of this strategy [9–11]. This discrepancy suggests that DAPT strategies and clinical guidelines developed based on Western populations may not be directly applicable to East Asian patients.

Although several systematic reviews and meta-analyses of DAPT de-escalation strategies have been published recently, the majority have predominantly included Western or mixed populations [12, 13]. Therefore, DAPT strategies and clinical guidelines developed largely from Western populations may not be directly applicable to East Asian patients. Specifically, East Asian studies often constitute only a limited proportion of pooled datasets, which constrains the precision of population-specific estimates. In addition, prior analyses frequently combine heterogeneous de-escalation approaches and timings, potentially masking effects that are unique to East Asian patients [14–16]. Accordingly, this study aims to perform a comprehensive systematic review and meta-analysis of clinical studies conducted in East Asia. The objective is to specifically evaluate the efficacy and safety of DAPT de-escalation in this population, thereby providing more tailored evidence to inform clinical decision-making. We hypothesized that, compared with standard potent DAPT, DAPT de-escalation in East Asian patients would reduce BARC-defined bleeding without increasing major adverse cardiovascular events (MACE).

Materials and methods

This systematic review and meta-analysis was designed and reported following the guidelines outlined in the PRISMA statement [17]. The protocol for this review was prospectively registered in advance on the PROSPERO under the identifier CRD420251078928.

Data sources

A thorough literature search was conducted in PubMed, Embase, and the Cochrane databases, along with Chinese databases including CNKI, Weipu, and Wanfang, to identify eligible RCTs and cohort studies published up to June 2025. The search approach incorporated both MeSH terms and free-text keywords. Terms used included ‘acute coronary syndrome’, ‘East Asia’, ‘percutaneous coronary intervention’, and ‘de-escalation antiplatelet therapy’. Two reviewers (Z.C. and J.C.) independently screened the retrieved records, and discrepancies were resolved through consultation with a third investigator (Y.W.). A full account of the search algorithms used for each source is provided in Supplementary Table 1.

Study selection

Eligible studies were selected according to the PICOS framework, with the following inclusion criteria: Population: Adult individuals of East Asian ethnicity who underwent PCI for ACS. Intervention: DAPT de-escalation (potency- or dose-based, operationally defined in Sect. 2.3). Comparison: Continuing treatment with a standard-dose DAPT regimen incorporating a potent P2Y12 inhibitor. Outcomes: The study reported outcomes concerning either MACE or bleeding events, with bleeding categorized according to Bleeding Academic Research Consortium (BARC) criteria. Exclusion criteria included: [1] studies involving inappropriate populations [2]. Studies evaluating non-relevant intervention strategies, such as aspirin-free/P2Y12 inhibitor monotherapy or “escalation” therapy (e.g., initiating treatment with clopidogrel and switching to a potent inhibitor) [3]. Ineligible publication types, including case reports, reviews, editorials, letters, guidelines, conference abstracts, other meta-analyses, or animal experiments. Based on the predefined criteria, two reviewers (Z.C. and J.C.) independently screened the titles and abstracts of all identified records and proceeded to full-text assessment when necessary. Any disagreements were resolved through discussion with a third investigator (Y.W.).

Definition and classification of de-escalation strategies

In this review, DAPT de-escalation was defined as a planned reduction in P2Y12-inhibitor intensity while maintaining aspirin-based DAPT after PCI for ACS; strategies that discontinued aspirin (aspirin-free/P2Y12 monotherapy) or involved “escalation” were excluded (see eligibility criteria). We classified de-escalation into two operational categories: [1] potency de-escalation, switching from a potent P2Y12 inhibitor (ticagrelor/prasugrel) to clopidogrel; and [2] dose de-escalation, reducing the dose of a potent P2Y12 inhibitor from the standard to a lower dose. Because both approaches are clinically used to reduce overall P2Y12 inhibitory intensity in East Asian patients (who are prone to bleeding on standard potent DAPT), we pooled them in the primary analysis to answer the overarching question of any DAPT de-escalation vs. continued standard potent DAPT.

Data extraction

Investigators (Z.C. and J.C.) individually extracted all relevant data from the eligible studies. The collected data encompassed study-level details (first author, publication year, geographic region, study type, number of participants, and follow-up duration) as well as baseline patient information (age, gender, clinical status such as ACS or AMI, and the de-escalation strategy). Any discrepancies during data extraction were resolved through discussion with a third investigator (Y.W.) until consensus was reached. When a study reported the same outcome at multiple time points (e.g., in-hospital vs. post-discharge, 6-month vs. 12-month), we included one effect size per study per outcome in the primary meta-analysis to avoid double-counting participants. We preferentially extracted estimates at ~ 12 months after index PCI (or the time point closest to 12 months) to maximize comparability across studies; if 12-month data were unavailable, we used the estimate from the longest reported follow-up. If both in-hospital and post-discharge outcomes were reported, we prioritized post-discharge/long-term outcomes.

For multi-arm RCTs, to prevent double-counting from shared controls, we prespecified that each study contributes one effect size per outcome in the overall meta-analysis. When trials included multiple de-escalation-related arms (e.g., guided and unguided), arms were pooled within the study by strategy category: de-escalation arms were combined into a single de-escalation group and continued potent P2Y12 inhibitor arms into a single control group, generating one comparison for the overall model. When guided vs. unguided effects were evaluated, the corresponding comparisons were included only in their prespecified subgroup analyses.

Study outcomes

The main outcomes included MACE, characterized as a composite endpoint comprising cardiovascular mortality, non-fatal MI, and non-fatal stroke, as well as bleeding events classified in accordance with BARC criteria. Bleeding outcomes were stratified into Minor Bleeding (BARC type 1), Clinically Relevant Bleeding (BARC type ≥ 2), and Major Bleeding (BARC type ≥ 3). Secondary endpoints encompassed the individual elements of MACE—namely cardiovascular mortality, all-cause mortality, non-fatal MI, non-fatal stroke, and unplanned revascularization—as well as bleeding events not defined by BARC criteria.

Risk of bias

The included RCTs were assessed for risk of bias using the Cochrane RoB 2.0 tool [18]. Each RCT was subsequently rated as having “low risk,” “some concerns,” or “high risk” of bias. The quality of cohort studies was appraised using the Newcastle-Ottawa Scale (NOS) [19]. The NOS evaluates studies using eight criteria spanning three domains—selection, comparability, and outcome—with total scores ranging from 0 to 9; a score of 7 or higher was considered indicative of high quality. All risk of bias assessments were independently performed by two investigators (Z.C. and J.C.), with any disagreements resolved through discussion with a third reviewer (Y.W.) until consensus was reached.

Statistical analysis

All statistical analyses were conducted using the ‘meta’ command in Stata version 15.0. For dichotomous outcomes, risk ratios (RRs) and corresponding 95% confidence intervals (CIs) were calculated. Heterogeneity across studies was assessed using the I² statistic. A fixed-effect model was used when heterogeneity was low (I² < 50%), whereas a random-effects model was applied in cases of substantial heterogeneity (I² ≥ 50%). Given the anticipated clinical heterogeneity across trials (ACS spectrum, timing of de-escalation, drug/dose regimens, adherence, and concomitant medications), we additionally performed random-effects sensitivity analyses for all key outcomes regardless of I² to assess robustness. A P-value below 0.05 was considered statistically significant. For outcomes with high heterogeneity, subgroup and meta-regression analyses were performed to investigate possible sources. A leave-one-out sensitivity analysis was carried out to assess the influence of each individual study on the pooled estimates and to evaluate the robustness of the overall results. We prespecified subgroup analyses by de-escalation strategy (potency vs. dose) and by approach (guided vs. unguided where applicable), and assessed subgroup differences using tests for interaction.

Since fewer than 10 studies were ultimately included, formal statistical evaluations of publication bias were not conducted, given the limited power of these methods under such conditions. The certainty of evidence for each outcome was assessed using the GRADE approach and classified as ‘high’, ‘moderate’, ‘low’, or ‘very low’. This evaluation was independently performed by two reviewers (Z.C. and J.C.), with any discrepancies settled through discussion with a third investigator (Y.W.).

Results

Study selection

A total of 985 records were initially retrieved through database searches. After removing 305 duplicates, 680 unique records were screened by title and abstract, leading to the exclusion of 576 studies. The remaining 104 articles were reviewed in full to determine eligibility. Among them, 95 were excluded based on the following criteria: 59 studies investigated antiplatelet monotherapy, 28 did not evaluate DAPT de-escalation, 4 focused on triple antiplatelet therapy, 3 examined an aspirin-free de-escalation strategy [20–22], and one study involved a mixed cohort of patients with ACS and those with stable coronary artery disease [23]. Ultimately, a total of 9 studies, comprising five RCTs [24–28] and four NRCTs [29–32], met the inclusion criteria. An overall summary of the study selection process is illustrated in the PRISMA flow diagram (Fig. 1).

Fig. 1.

PRISMA flowchart of study selection

Study characteristics

This meta-analysis included nine studies in total, consisting of five RCTs and four cohort studies, encompassing 10,263 participants. Overall, the included population had a weighted mean age of 61.5 ± 11.6 years, and 78.4% were men. Bleeding risk stratification was variably reported across studies (e.g., ARC–HBR major criteria in one cohort and baseline PARIS bleeding score in another). The publication years of the included studies ranged from 2018 to 2025. Among the included studies, male representation varied from 63.3% to 89.7%, while the average age of participants ranged between 58 and 68 years. Regarding the intervention strategy, seven studies investigated potency de-escalation, while two studies examined dose de-escalation. Among the RCTs included, two studies (Gong et al. and Zhao et al.) employed multi-arm designs, evaluating both guided and unguided strategies of DAPT de-escalation in comparison to standard therapy, with each intervention arm matched to a separate control. Table 1 provides a summary of the key characteristics of all included studies. These strategies were pooled as overall DAPT de-escalation in the primary analysis, with strategy-stratified subgroup analyses conducted to explore potential effect modification.

Table 1.

Basic characteristics of included studies

Study	Country	Sample Size		Treatment duration	Age(years old)		Male n(%)		Participants	De-escalation Strategy	Outcomes	Study design
		T	C		T	C	T	C
Gong-2021 (Unguide)	China	99	101	1 year	60.6 ± 9.4	60.4 ± 7.8	70(70.7)	72(71.3)	ACS	Potency	MACE + BARC + Non-BARC Bleeding	RCT
Gong-2021 (Guide)	China	98	103	1 year	62.8 ± 7.4	60.4 ± 9.9	66(67.3)	68(66.0)	ACS	Potency	MACE + BARC + Non-BARC Bleeding	RCT
Kim-2020	Korean	1170	1168	1 year	58.7 ± 9.0	58.9 ± 9.1	1050 (89.7)	1037 (88.8)	ACS	Dose	MACE + BARC	RCT
Kim-2021	Korean	1349	1348	1 year	60.1 ± 11.3	59.9 ± 11.4	1132(83.9)	1111 (82.4)	AMI	Potency	MACE + BARC	RCT
Wang-2022	China	90	90	1 year	62.1 ± 5.2	62.4 ± 5.4	57(63.3)	61(67.8)	ACS	Potency	MACE + BARC + Non-BARC Bleeding	RCT
Zhao-2023 (Unguide)	China	60	59	1 year	63.8 ± 8.7	64.8 ± 10.3	51(85.2)	52(88.1)	ACS	Potency	MACE + BARC	RCT
Zhao-2023 (Guide)	China	91	33	1 year	64.8 ± 11.3	66.1 ± 9.8	77(84.6)	29(87.9)	ACS	Potency	MACE + BARC	RCT
Honda-2021	Japan	110	2494	10 months	68.0 ± 12.7	66.8 ± 13.0	87(79.1)	1953(78.3)	AMI	Potency	MACE + BARC	NRCT
Lee-2023	Korean	115	657	1 year	61.3 ± 10.5	60.3 ± 11.6	88 (76.5)	557 (84.8)	ACS	Dose	MACE + Non-BARC Bleeding	NRCT
Li-2018	China	152	161	1 year	59.3 ± 13.4	59.8 ± 11.2	115 (77.0)	136 (83.6)	STEMI	Potency	MACE + BARC	NRCT
Li-2025	China	166	549	1 year	58.0 ± 10.8	58.1 ± 10.7	117(70.5)	420(76.5)	ACS	Potency	MACE + Non-BARC Bleeding	NRCT

ACS Acute Coronary Syndrome, AMI Acute Myocardial Infarction, STEMI ST-segment Elevation Myocardial Infarction, BARC Bleeding Academic Research Consortium, C Control, T Treatment, MACE Major Adverse Cardiovascular Events, BARC Bleeding Academic Research Consortium, NRCT Non-Randomized Controlled Trial, RCT Randomized Controlled Trial

Risk of bias

Risk of bias for the five included RCTs was assessed using the Cochrane RoB 2.0 tool (Fig. 2). Overall, Kim-2020 and Kim-2021 were judged as low risk of bias [25, 26], Gong-2021 and Zhao-2023 were rated as having some concerns [24, 28], and Wang-2022 was assessed as high risk of bias [27]. The four cohort studies were evaluated for methodological quality using the NOS. All four cohort studies achieved scores between 7 and 8 (out of a maximum of 9), indicating they were of high quality (Supplementary Table 2).

Fig. 2.

Risk of bias assessment for the included RCTs(RoB 2.0)

Meta-analysis

MACE and bleeding outcomes

The composite endpoint of MACE was reported in eight studies [25–32], BARC type 1 bleeding in three studies [24, 25, 29], BARC type ≥ 2 bleeding in seven studies [24–29, 31], and BARC type ≥ 3 bleeding in six studies [25–29, 31],. Low heterogeneity was observed across studies for these outcomes (all I² < 50%); consequently, a fixed-effect model was applied to the pooled analysis. Pooled analysis indicated a 9% reduction in MACE risk in the de-escalation group in contrast to standard group, though the difference did not attain statistical significance (RR, 0.91; 95% CI, 0.70–1.19; p = 0.497; Fig. 3A). We did not pool potency and dose de-escalation separately because ischemic events were infrequent and the dose de-escalation evidence base was small. By comparison, the de-escalation approach was linked to a markedly reduced risk of bleeding. Specifically, it reduced the risk of BARC type 1 bleeding by 48% (RR, 0.52; 95% CI, 0.32–0.86; p = 0.011; Fig. 3B), BARC type ≥ 2 bleeding by 53% (RR, 0.47; 95% CI, 0.37–0.60; p < 0.001; Fig. 3C), and BARC type ≥ 3 bleeding by 39% (RR, 0.61; 95% CI, 0.40–0.94; p = 0.026; Fig. 3D).

Fig. 3.

Forest plots: A MACE; B Minor Bleeding Events (BARC Type 1); C Clinically Relevant Bleeding Events (BARC ≥ 2); D Major Bleeding Events (BARC ≥ 3)

Random-effects sensitivity analyses yielded effect estimates consistent with the primary fixed-effect results(Supplementary Fig. 1). For MACE, the random-effects model showed no significant difference (RR 0.97, 95% CI 0.69–1.38; I²=27.0%), consistent with the fixed-effect estimate. For bleeding outcomes, the random-effects results were materially unchanged: BARC type 1 (RR 0.52, 95% CI 0.31–0.86; I²=0%), BARC ≥ 2 (RR 0.48, 95% CI 0.37–0.61; I²=0%), and BARC ≥ 3 (RR 0.63, 95% CI 0.40–0.98; I²=0%). Overall, the conclusions remained robust with slightly wider confidence intervals under random effects.

Individual components of MACE

Data on cardiovascular mortality were available from nine studies [24–32], all-cause mortality from four studies [25, 26, 30, 31], non-fatal MI from eight studies [25–32], and non-fatal stroke from nine studies [24–32]. Low statistical heterogeneity was observed for all of these endpoints (all I² < 50%). Accordingly, a fixed-effect model was applied for the analysis. The meta-analysis revealed that the risk of cardiovascular mortality did not differ significantly between the de-escalation group and those receiving standard DAPT (RR = 0.81; 95% CI: 0.47–1.40; p = 0.454; Fig. 4A). Similarly, no significant difference was found in all-cause mortality (RR = 0.90; 95% CI: 0.55–1.47; p = 0.676; Fig. 4B). Similarly, there was no meaningful difference in the incidence of non-fatal MI between the de-escalation and standard DAPT strategies (RR = 0.94; 95% CI: 0.64–1.38; p = 0.736; Fig. 4C), nor did the risk of non-fatal stroke (RR = 1.02; 95% CI: 0.64–1.64; p = 0.920; Fig. 4D).

Fig. 4.

Forest plots: A Cardiovascular Mortality; B All-Cause Mortality; C Non-fatal Myocardial Infarction; D Non-fatal Stroke

Unplanned revascularization and Non-BARC bleeding events

Data for unplanned revascularization were reported in seven studies [24–27, 29, 30, 32]. The analysis for this outcome revealed substantial heterogeneity (I² = 50.1%). A leave-one-out sensitivity analysis suggested that this heterogeneity was largely driven by the STEMI-only cohort study by Li et al. (2018). Consequently, the analysis was conducted using a random-effects model. The pooled results suggested no significant difference in the rate of unplanned revascularization between patients receiving de-escalation therapy and those on standard treatment (RR, 1.01; 95% CI, 0.64–1.57; p = 0.982; Fig. 5A). Four studies [24, 27, 30, 32] reported on non-BARC bleeding events. The analysis for this outcome indicated low heterogeneity (I² = 0.0%), and a fixed-effect model was selected for analysis. The combined analysis revealed a significant 48% reduction in the risk of non-BARC bleeding in the de-escalation group compared to the standard therapy group (RR = 0.52; 95% CI: 0.37–0.74; p < 0.001; Fig. 5B).

Fig. 5.

Forest plots: A Unplanned Revascularization; B Non-BARC Bleeding Events

Subgroup and meta-regression analyses

To explore possible contributors to the considerable heterogeneity observed in the outcome of unplanned revascularization (I² = 50.1%), subgroup analyses and meta-regression were performed. These analyses considered factors that could potentially influence the results, including study design, type of de-escalation strategy, publication language, and population ethnicity. However, neither the subgroup analysis (Supplementary Fig. 2) nor the meta-regression (Supplementary Fig. 3) successfully identified a clear source of this heterogeneity. This may be attributable to several factors, such as the imbalanced number of participants across subgroups, low event rates in some trials, and insufficient adjustment for confounding variables in the primary studies.

Sensitivity analysis

A leave-one-study-out sensitivity analysis was conducted to explore possible contributors to heterogeneity in outcomes with elevated I² values and to evaluate the stability of the pooled estimates. The sensitivity analysis indicated that removing the study by Li et al. (2018) [29]markedly reduced heterogeneity in the outcome of unplanned revascularization, with I² dropping from 50.0% to 0.0% (P = 0.577). Nevertheless, the comparison between the two groups for this outcome remained statistically non-significant (RR, 0.85; 95% CI, 0.65–1.12; p = 0.244; Fig. 6). This result suggests that the heterogeneity observed was primarily driven by the study conducted by Li et al., likely owing to clinical differences within the patient cohort. Specifically, this study focused solely on individuals with STEMI, whereas the other studies enrolled a more heterogeneous group of ACS patients. STEMI patients typically represent a higher acute thrombotic-risk presentation and may undergo more staged or target-vessel procedures during follow-up, which could influence revascularization event patterns. In addition, as a real-world non-randomized cohort, Li et al. (2018) may be more susceptible to residual confounding, which could also contribute to an outlying effect estimate for this endpoint. For all other outcomes, no single study significantly influenced the pooled estimates, reinforcing the consistency and reliability of our findings (Supplementary Figs. 4–6).

Fig. 6.

Unplanned Revascularization (Excluding the Li-2018)

Sensitivity analysis restricted to randomized trials

In a sensitivity analysis including only the randomized controlled trials (i.e. excluding observational studies), the pooled results for all major outcomes remained directionally consistent with those of the primary analysis(Supplementary Fig. 7–9). De-escalation of DAPT continued to significantly reduce clinically relevant bleeding (BARC type ≥ 2) and major bleeding (BARC type ≥ 3) compared to standard therapy, with RR of approximately 0.47 (95% CI: 0.36–0.60) for BARC ≥ 2 and 0.61 (95% CI: 0.38–0.97) for BARC ≥ 3 -virtually identical to the overall analysis. Similarly, there was still no significant difference in the risk of MACE between de-escalation and standard DAPT (RR: 077, 95% CI spanning 1.0, p > 0.05). No appreciable between-group differences were observed for other ischemic endpoints, including cardiovascular mortality (RR: 0.65, 95% CI: 0.31–1.41), all-cause mortality (RR: 0.87, 95% CI: 0.49–1.57), non-fatal MI (RR: 0.76, 95% CI: 0.47–1.22), non-fatal stroke (RR:0.85, 95% CI: 0.50–1.45), or unplanned revascularization (RR: 0.83, 95% CI: 0.62–1.12); all remained statistically non-significant, consistent with the primary analysis. Importantly, the only notable change in this RCT-only analysis was observed for minor bleeding events (BARC type 1).Whereas de-escalation was associated with a significantly lower risk of BARC 1 bleeding in the overall analysis (previously RR: 0.52, 95% CI: 0.32–0.86), the RCT-restricted analysis still showed a trend toward reduced minor bleeding with de-escalation but this difference was no longer statistically significant (RR: 0.65, 95% CI: 0.28–1.50, p = 0.311).This attenuation of significance is most likely due to the reduced sample size and statistical power after excluding the observational cohorts, rather than a true reversal of effect.Overall, the RCT-only sensitivity analysis confirms the robustness of our primary findings. The direction and magnitude of treatment effects remained largely unchanged, and heterogeneity (I²) measures for each outcome were not substantially affected by the exclusion of non-randomized studies.

Publication bias

This meta-analysis incorporated nine studies in total. Since the number of included studies did not reach the commonly accepted threshold of ten for reliable detection of publication bias, formal assessments—such as funnel plot inspection or Egger’s test—were not conducted.

GRADE evidence quality

The quality of evidence for each outcome was assessed using the GRADE methodology [33], with detailed results presented in Table 2. The evaluation revealed that the evidence supporting cardiovascular-related outcomes—such as MACE, mortality, and stroke—was primarily rated as ‘low’ to ‘very low’ in quality. In contrast, bleeding-related outcomes were supported by stronger evidence, with most being graded as ‘moderate’, indicating a comparatively higher level of confidence. However, these findings still warrant confirmation through future high-quality RCTs.

Table 2.

GRADE quality of evidence for outcomes

Outcome	No. of studies	Study design	Risk of Bias	Inconsistency	Indirectness	Imprecision	Publication Bias	Quality of Evidence
MACE	8	RCTs + Cohort studies	Serious	Not serious	Not serious	Serious	Not assessed	Low
BARC type 1	3	RCTs + Cohort studies	Serious	Not serious	Not serious	Not serious	Not assessed	Moderate
BARC ≥ type 2	7	RCTs + Cohort studies	Serious	Not serious	Not serious	Not serious	Not assessed	Moderate
BARC ≥ type 3	6	RCTs + Cohort studies	Serious	Not serious	Not serious	Not serious	Not assessed	Moderate
Cardiovascular Mortality	9	RCTs + Cohort studies	Serious	Not serious	Not serious	Very serious	Not assessed	Very Low
All-cause Mortality	4	RCTs + Cohort studies	Serious	Not serious	Not serious	Serious	Not assessed	Low
Non-fatal MI	8	RCTs + Cohort studies	Serious	Not serious	Not serious	Serious	Not assessed	Low
Non-fatal Stroke	9	RCTs + Cohort studies	Serious	Not serious	Not serious	Serious	Not assessed	Low
Unplanned Revascularization	7	RCTs + Cohort studies	Serious	Serious	Not serious	Serious	Not assessed	Very Low
Non-BARC bleeding events	4	RCTs + Cohort studies	Serious	Not serious	Not serious	Not serious	Not assessed	Moderate

MACE Major Adverse Cardiovascular Events, BARC Bleeding Academic Research Consortium, MI Myocardial Infarction, RCT Randomized Controlled Trial, GRADE Grading of Recommendations Assessment, Development and Evaluation

Discussion

This meta-analysis assessed DAPT de-escalation after PCI in East Asian patients with ACS. Compared with standard potent DAPT, de-escalation was not associated with a statistically significant increase in ischemic risk. It was associated with fewer bleeding events. This suggests that in the East Asian ACS population, a judicious reduction in DAPT intensity can achieve a substantial decrease in bleeding complications without a statistically significant increase in ischemic events. Our results are consistent with recent RCTs from East Asia. TALOS-AMI evaluated an unguided switch from ticagrelor to clopidogrel in stabilized AMI patients [25].HOST-REDUCE-POLYTECH-ACS evaluated prasugrel dose reduction [26]. Together, these data support tailored antiplatelet therapy in East Asian patients. However, potency de-escalation (switching to clopidogrel) and dose de-escalation (reduced-dose potent P2Y12 inhibition) are pharmacodynamically distinct and may yield different levels of platelet inhibition. Therefore, the absence of an ischemic signal in our pooled analysis should not be interpreted as uniform ischemic neutrality across de-escalation modalities. Given limited power for modality-specific comparisons, divergent effects cannot be excluded, and future adequately powered randomized trials are warranted to directly compare de-escalation strategies. Collectively, our results offer supportive evidence-based support for optimizing antiplatelet therapy in the context of the ‘East Asian paradox’—the unique phenomenon wherein this population exhibits a lower ischemic yet higher bleeding risk on standard antithrombotic therapy [34]—and indicate that a DAPT de-escalation approach may be a more suitable therapeutic strategy for these patients. Bleeding risk is heterogeneous and may be influenced by patient-level characteristics such as age and sex. In our evidence base, participants were predominantly male and relatively young, which may limit generalizability to elderly and female patients. Some included studies reported baseline bleeding-risk stratification tools (e.g., ARC-HBR major criteria or PARIS bleeding score); however, subgroup-specific bleeding outcomes by age, sex, or risk strata were inconsistently reported and rarely available in an extractable format across studies. Therefore, a quantitative stratified meta-analysis was not feasible, and future trials and/or individual-participant-data meta-analyses with standardized bleeding risk assessment and reporting are warranted. For MACE, the pooled estimate was RR 0.91 with a 95% CI of 0.70 to 1.19. This interval remains compatible with a clinically relevant increase in risk. Therefore, the finding should be interpreted as a non-significant difference rather than evidence of equivalence or non-inferiority. Importantly, the primary inference regarding ischemic safety is anchored in the randomized evidence. Randomized controlled trials constitute the highest level of evidence for causal assessment of ischemic outcomes after ACS-PCI, minimizing confounding and treatment-selection bias. Accordingly, our interpretation of ischemic safety relies primarily on the directionally consistent findings observed in the RCT-only sensitivity analysis, while observational cohorts are considered supportive context that increases external generalizability but should not be weighted equivalently for causal inference. In this context, the RCT-only sensitivity analysis is particularly informative for the clinical question of ischemic safety. When restricted to randomized trials, the pooled estimates for MACE and individual ischemic endpoints remained non-significant and directionally consistent with the primary analysis, with no signal of increased myocardial infarction, stroke, or cardiovascular mortality. While imprecision persists due to low event counts, the randomized evidence provides the strongest available reassurance that de-escalation does not materially increase ischemic risk in stabilized East Asian ACS patients after PCI.

Beyond tailoring the intensity of P2Y12 inhibition (i.e., de-escalation), DAPT can also be individualized by adapting treatment duration. Standard strategies often apply a fixed duration (e.g., 12 months after ACS-PCI) irrespective of the patient’s competing ischemic and bleeding risks. In the PARTHENOPE randomized trial, a DAPT score–guided approach that assigned 3-, 6-, or 24-month DAPT according to the patient’s risk profile was superior to a uniform 12-month strategy in terms of net clinical benefit, supporting a risk-adapted paradigm in which patients with higher ischemic risk (DAPT score ≥ 2) may benefit from prolonged therapy, whereas those with lower risk may be candidates for shorter or less intensive strategies [35]. In contemporary PCI practice, especially in the setting of complex PCI where thrombotic risk is intrinsically higher, antithrombotic therapy should be framed as a risk-tailoring exercise rather than a ‘one-size-fits-all’ approach. Recent state-of-the-art evidence highlights that in complex PCI, the choice and duration of therapy should be guided primarily by bleeding risk, with consideration of prolonged antithrombotic regimens (e.g., extended DAPT or dual-pathway inhibition) when ischemic risk predominates and bleeding risk is low [36]. At the same time, selected contemporary strategies such as P2Y12 inhibitor monotherapy after short DAPT have been evaluated in complex PCI populations and, in several datasets, were not associated with an excess of ischemic events. Although our meta-analysis focused on aspirin-based DAPT de-escalation (and therefore excluded aspirin-free monotherapy strategies by design), these evolving data underscore the broader principle that optimal post-PCI antithrombotic management—particularly in complex PCI—requires individualized balancing of ischemic versus bleeding risk.

Compared to meta-analyses based predominantly on Western cohorts, the findings of our study are more specifically applicable to the East Asian population. In Western populations, a higher baseline ischemic risk often necessitates more cautious de-escalation approaches. For instance, the TROPICAL-ACS trial required guidance from platelet function testing (PFT) to ensure that antithrombotic efficacy was not unduly compromised post-de-escalation, thereby mitigating any potential increase in ischemic risk [9]. In contrast, our analysis includes multiple trials that suggested the practicality of unguided de-escalation within East Asian populations, highlighting a fundamental divergence in the ischemia–bleeding balance between Eastern and Western cohorts. This discrepancy suggests that a direct application of the potent DAPT regimens recommended in Western guidelines may lead to ‘over-treatment’ in East Asian patients, exposing them to an unnecessary bleeding hazard. A de-escalation strategy, therefore, may represents an effective correction for this risk-benefit imbalance.

A growing body of research suggests that East Asian individuals, when treated with antiplatelet agents, tend to have a reduced rate of ischemic complications but an elevated bleeding risk compared to Western populations [37–39]. This phenomenon is may be influenced by multiple mechanisms, including distinct pharmacogenetic profiles and coagulation physiology [40, 41]. From a pharmacogenetic perspective, East Asian populations have a substantially higher frequency of CYP2C19 loss-of-function alleles than Western populations, with estimates ranging from approximately 50–60% versus 25–30%, respectively [42], which can be associated with reduced bioactivation of clopidogrel. Conversely, multiple studies have confirmed that East Asian patients are more sensitive to standard doses of ticagrelor or prasugrel, exhibiting significantly higher plasma concentrations and levels of platelet inhibition than their Western counterparts, which may contribute to their heightened bleeding tendency [43–45]. Thus, transitioning from a potent P2Y12 inhibitor to clopidogrel effectively shifts the patient from a state of excessive platelet inhibition and elevated bleeding risk to one that, despite potentially increasing on-treatment platelet reactivity (HPR), may better align with the lower ischemic threshold observed in East Asian populations. Beyond these differences in drug metabolism, research also suggests that East Asian populations generally exhibit a lower intrinsic thrombogenic tendency [46]. Evidence indicates that, compared to Caucasians, individuals of East Asian descent have lower levels of inflammatory markers, reduced coagulation factor activity, and distinct patterns of endogenous fibrinolytic system activity [47–49]. This inherent ‘antithrombotic’ predisposition may provide a biological cornerstone for the potential safety of de-escalation strategies. It implies that when faced with a similar thrombotic stimulus, such as atherosclerotic plaque rupture, East Asian patients have a lower propensity for thrombus formation and propagation and, consequently, may not require the same intensity of antiplatelet inhibition as their Western counterparts. Importantly, these pharmacogenetic and pathophysiologic considerations are drawn from prior literature and were not directly evaluated in the included studies (e.g., no genotype- or pharmacokinetic-stratified outcome analyses); therefore, they should be interpreted as hypothesis-generating rather than causal explanations.

Our study’s findings carry important clinical implications for the development of future antiplatelet therapy strategies in East Asian populations. First, DAPT de-escalation could potentially be considered a routine therapeutic option for East Asian ACS patients during the stable phase post-PCI (e.g., after the first month). Importantly, the current East Asian evidence base underlying this meta-analysis predominantly reflects de-escalation implemented after completion of the first month of potent DAPT (i.e., during a stabilized post-acute phase). Therefore, our findings should be interpreted primarily as supporting de-escalation around one-month post-PCI rather than during the immediate peri-procedural period. Second, the success of ‘unguided’ de-escalation strategies in our analysis suggests that de-escalation based on clinical judgment is a feasible and potentially safe approach in regions with limited resources or where PFT and genetic testing are not readily available, which enhances the accessibility and practical utility of the strategy. In resource-limited East Asian settings, the availability and cost of platelet function testing or genotyping may constrain guided approaches; therefore, clinically risk-stratified unguided de-escalation after the acute phase in stabilized patients may be a pragmatic option. Future trials in East Asian populations should directly compare de-escalation strategies versus P2Y12 inhibitor monotherapy to inform optimal strategy selection across different resource settings. Notably, similar bleeding-mitigation concepts through strategy selection have also been observed in other high-risk populations, such as ICI-treated cancer patients with VTE, where DOACs demonstrated comparable recurrent VTE risk but lower intracranial hemorrhage and mortality than LMWH [50]. Likewise, in cancer surgery settings, meta-analyses suggest DOAC prophylaxis is non-inferior to LMWH with broadly comparable bleeding outcomes, underscoring the value of tailoring antithrombotic strategies to population-specific bleeding risk [51].

Likewise, stratified analyses by patient-level bleeding risk factors (e.g., age, sex, or baseline bleeding risk scores) were not feasible because subgroup-specific bleeding outcomes were not consistently reported across studies.Nevertheless, we acknowledge several limitations of our study. But the timing of de-escalation varied across studies, and very early de-escalation (in-hospital or within the first 7 days after PCI) remains insufficiently studied in East Asian ACS populations. Accordingly, our results should not be extrapolated to ultra-early de-escalation strategies, which warrant dedicated investigation. First, the inclusion of cohort studies may affect the robustness of our conclusions. Because observational studies are inherently susceptible to residual confounding and treatment-selection bias, we treated them as complementary evidence that enhances real-world relevance rather than as equivalent contributors to the causal inference on ischemic safety, which is best informed by randomized trials. Additionally, subgroup analysis by ACS subtype (STEMI vs. NSTE-ACS) may be clinically warranted; however, only one included study enrolled a STEMI-only population (Li et al., 2018), and most other studies did not report extractable subtype-stratified outcomes.Thus, a formal subtype-based subgroup meta-analysis would be underpowered and potentially unstable in the current evidence base. Second, while the GRADE evaluation rated the evidence for bleeding outcomes as moderate in quality, the supporting evidence for cardiovascular events was judged to be low or very low, reflecting imprecision driven by limited event counts and modest study sizes. Ischemic endpoints were relatively infrequent across the included studies, leading to a limited number of events and consequently wider confidence intervals for MACE and its components. In addition, only a small number of RCTs were available and several were modest in size, with ischemic outcomes often not being the primary powered endpoints, which further reduced statistical precision. Finally, cohort studies may be subject to treatment-selection bias and residual confounding, and in some studies the adjustment for confounders and outcome ascertainment may be insufficient, which could further contribute to uncertainty in ischemic estimates. Moreover, although formal funnel-plot or Egger-based assessments were not performed due to the limited number of studies, we cannot exclude the possibility of reporting or publication bias, particularly because several included trials were small and some were published in local-language sources; therefore, the observed directionally consistent bleeding benefit may be partly inflated by selective outcome reporting or preferential publication of positive findings. Low I² does not guarantee true homogeneity when event rates are low. Similar effects can occur by chance. Therefore, ischemic findings should be interpreted cautiously, and strategy-specific effects require future confirmation.This indicates that the existing evidence base remains insufficient to conclusively exclude a slight potential elevation in ischemic risk associated with de-escalation strategies. Third, there was heterogeneity across included studies in both the timing and operational definitions of de-escalation (e.g., when and how switching/dose reduction was implemented), which may have influenced between-study comparability. Looking ahead, there is a clear need for more large-scale, rigorously designed RCTs focusing specifically on East Asian populations. These studies should aim to provide definitive evidence regarding the ischemic safety of de-escalation strategies, including their potential non-inferiority or superiority. Timing of de-escalation may be an important effect modifier, but timing was inconsistently reported and largely clustered around ~ 1 month across included trials with limited extractable timing-stratified outcomes, precluding a robust timing-based moderator analysis in this meta-analysis. In addition, due to the lack of patient-level data, we were unable to evaluate the role of CYP2C19 genotyping or platelet function testing in guiding de-escalation decisions and outcomes, and thus mechanistic interpretations regarding pharmacogenetics should be considered hypothesis-generating rather than causal. Moreover, because potency and dose de-escalation have distinct biological and platelet-inhibitory profiles, pooled ischemic estimates may obscure modality-specific effects; thus, ischemic neutrality should not be assumed to be uniform across de-escalation strategies. In addition, several PCI trials preferentially report time-to-event effect estimates (hazard ratios); however, because HRs were not consistently available across included studies, we pooled risk ratios at fixed follow-up, which may not fully account for censoring or differences in follow-up time and could introduce some imprecision. Finally, since most included trials had follow-up durations limited to one year, the long-term outcomes beyond 12 months remain inadequately defined. As a result, late ischemic events—particularly late stent thrombosis—may have been under-captured in the current evidence base. This knowledge gap needs to be addressed through extended follow-up of existing trials or by conducting large-scale registry studies.

Conclusion

This meta-analysis suggests that in East Asian ACS patients undergoing PCI, DAPT de-escalation reduces bleeding risk. Ischemic outcomes did not differ significantly, but clinically relevant harm cannot be excluded due to imprecision. This interpretation is primarily supported by randomized trial evidence, whereas observational data provide supportive context and improve generalizability. These results suggest a potentially favorable risk-benefit profile, especially for those at high bleeding risk. Nonetheless, due to limitations such as study heterogeneity, short follow-up, and low-certainty evidence for some outcomes, further large-scale, long-term RCTs are needed to confirm these findings.

Authors’ contributions

Z.C. designed the study; Z.C., J.C., and Y.W. performed systematic search and selected the studies; K.Z. and JT.C. analyzed the data; H.J. and J.L. prepared the manuscript and critically reviewed the manuscript; and Z.C. had primary responsibility for final content. All authors have read and approved the final manuscript.

Funding

This research was funded by the Xiamen Municipal Health Commission, Grant No. 2024GZL-CX51. This work was supported by the Xiamen Municipal Health Commission (Grant No. XWZY-2025-0601) and theXiamen Municipal Science and Technology Plan Project (Grant No. 3502Z20224ZD1170).

Data availability

The data are available from the corresponding author upon reasonable request.

Ethics approval and consent to participate

None.

Competing interests

The authors declare no competing interests.