Efficacy and Safety of Long‐Term Antithrombotic Strategies in Patients With Chronic Coronary Syndrome: A Network Meta‐analysis of Randomized Controlled Trials

Abstract

Background

Long‐term antithrombotic strategies for patients with chronic coronary syndrome with high‐risk factors represent an important treatment dilemma in clinical practice. Our aim was to conduct a network meta‐analysis to evaluate the efficacy and safety of long‐term antithrombotic strategies in patients with chronic coronary syndrome.

Methods and Results

Four randomized studies were included (n=75167; THEMIS [Ticagrelor on Health Outcomes in Diabetes Mellitus Patients Intervention Study], COMPASS [Cardiovascular Outcomes for People Using Anticoagulation Strategies], PEGASUS‐TIMI 54 [Prevention of Cardiovascular Events in Patients With Prior Heart Attack Using Ticagrelor Compared to Placebo on a Background of Aspirin–Thrombolysis in Myocardial Infarction 54], and DAPT [Dual Anti‐platelet Therapy]). The odds ratios (ORs) and 95% CIs) were calculated as the measure of effect size. The results of the network meta‐analysis showed that, compared with aspirin monotherapy, the ORs for trial‐defined major adverse cardiovascular and cerebrovascular events were 0.86; (95% CI, 0.80–0.93) for ticagrelor plus aspirin, 0.89 (95% CI, 0.78–1.02) for rivaroxaban monotherapy, 0.74 (95% CI, 0.64–0.85) for rivaroxaban plus aspirin, and 0.72 (95% CI, 0.60,–0.86) for thienopyridine plus aspirin. Compared with aspirin monotherapy, the ORs for trial‐defined major bleeding were 2.15 (95% CI, 1.78–2.59]) for ticagrelor plus aspirin, 1.51 (95% CI, 1.23–1.85) for rivaroxaban monotherapy, and 1.68 (95% CI, 1.37–2.05) for rivaroxaban plus aspirin. For death from any cause, the improvement effect of rivaroxaban plus aspirin was detected versus aspirin monotherapy (OR, 0.76; 95% CI, 0.65–0.90), ticagrelor plus aspirin (OR, 0.79; 95% CI, 0.66–0.95), rivaroxaban monotherapy (OR, 0.82; 95% CI, 0.69–0.97), and thienopyridine plus aspirin (OR, 0.58; 95% CI, 0.41–0.82) regimens.

Conclusions

All antithrombotic strategies combined with aspirin significantly reduced the incidence of major adverse cardiovascular and cerebrovascular events and increased the risk of major bleeding compared with aspirin monotherapy. Considering the outcomes of all ischemic and bleeding events and all‐cause mortality, rivaroxaban plus aspirin appears to be the preferred long‐term antithrombotic regimen for patients with chronic coronary syndrome and high‐risk factors.

Nonstandard Abbreviations and Acronyms

CAPRIE

Clopidogrel Versus Aspirin in Patients at Risk of Ischaemic Events

CCS

chronic coronary syndrome

COMPASS

Cardiovascular Outcomes for People Using Anticoagulation Strategies trial

DAPT

Dual Anti‐platelet Therapy study

GUSTO

Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries criteria

MACEs

major adverse cardiovascular and cerebrovascular events

PEGASUS‐TIMI 54

Prevention of Cardiovascular Events in Patients with Prior Heart Attack Using Ticagrelor Compared to Placebo on a Background of Aspirin–Thrombolysis in Myocardial Infarction 54 trial

PLATO

Platelet Inhibition and Patient Outcomes trial

POPular AGE

Clopidogrel Versus Ticagrelor or Prasugrel in Patients Aged 70 Years or Older With Non–ST‐Elevation Acute Coronary Syndrome: The Randomised, Open‐Label, Non‐Inferiority Trial

SUCRA

surface under the cumulative ranking

THEMIS

Ticagrelor on Health Outcomes in Diabetes Mellitus Patients Intervention Study

Clinical Perspective

What Is New?

• For all‐cause mortality, the improvement effect of rivaroxaban plus aspirin was detected versus aspirin monotherapy, ticagrelor plus aspirin, rivaroxaban monotherapy, and thienopyridine plus aspirin regimens in patients with chronic coronary syndrome.

What Are the Clinical Implications?

• Rivaroxaban plus aspirin appears to be the preferred long‐term antithrombotic regimen for patients with chronic coronary syndrome with high‐risk factors.

Chronic coronary syndrome (CCS) includes patients with suspected or definite stable coronary artery disease and acute coronary syndrome (ACS) entering the chronic phase. According to the guidelines of the European Society of Cardiology published in 2019, ¹ CCS is divided into 6 types, of which type 3 is defined as asymptomatic and symptomatic patients with stabilized symptoms <1 year after ACS or patients with recent revascularization; type 4 is defined as asymptomatic and symptomatic patients >1 year after the initial diagnosis or revascularization; and type 6 CCS is defined as asymptomatic subjects in whom coronary artery disease is detected at screening. Briefly, all 3 types are considered as definite coronary artery disease that is currently in a chronic or stable state. Although patients with CCS have a lower incidence of ischemic events or recurrence than patients with ACS, these patients are still at risk of myocardial infarction (MI), ischemic stroke, and cardiovascular death. For patients with stable coronary artery disease, dual antiplatelet therapy is generally administered for 6 months after percutaneous coronary intervention (PCI), and the treatment duration for patients with ACS is generally 12 months. ² However, for patients with CCS, an important challenge in clinical practice and research is to develop strategies that will achieve fewer ischemic events without increasing bleeding events.

Aspirin irreversibly inhibits platelet cyclooxygenase‐1, thereby preventing thromboxane production, ³ a question worth exploring is whether patients undergoing PCI or ACS continue to use aspirin monotherapy after the default dual antiplatelet treatment period and what treatments are considered the optimal choice. To date, many large‐scale clinical studies have selected various antithrombotic strategies for patients with chronic coronary syndrome, such as P2Y12 inhibitor monotherapy, ⁴ , ⁵ , ⁶ dual antiplatelet regimens, ⁷ , ⁸ , ⁹ , ¹⁰ anticoagulant monotherapy, ¹¹ and an anticoagulant combined with antiplatelet regimen, ¹² , ¹³ all of which exhibit good efficacy in preventing ischemic events. However, these strategies increase the incidence of bleeding events to some extent compared with aspirin monotherapy, and a mutual comparison of these antithrombotic strategies has not been conducted. Therefore, the aim of this network meta‐analysis is to evaluate the efficacy and safety of long‐term antithrombotic strategies in patients with chronic coronary syndrome.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta Analyses guidelines (Table S1) were used in this systematic review and network meta‐analysis. ¹⁴ The data that support the findings of this study are available from the corresponding author on reasonable request.

Data Sources

The Medline, EMBASE, and Cochrane database were independently searched by 2 reviewers. The search terms were “coronary artery disease,” “coronary heart disease,” “chronic coronary syndrome,” “myocardial infarction,” “acute coronary syndrome,” “percutaneous coronary intervention,” “coronary artery bypass grafting,” “coronary stenting” paired with “aspirin,” “clopidogrel,” “prasugrel,” “ticagrelor,” “P2Y12 inhibitors,” “thienopyridine,” “warfarin,” “vitamin K antagonists,” “dabigatran,” “rivaroxaban,” “apixaban,” “edoxaban,” “factor Xa inhibitor,” or “new oral anticoagulants.” Searches for studies published up to August 2020 were conducted using subject heading terms, key words, and titles or abstracts, and all identified abstracts were reasonably screened (Table S2).

Study Selection

An initial eligibility screen of all retrieved titles and abstracts was conducted, and original studies were included in our network meta‐analysis if they met the following criteria: (1) randomized controlled trial (RCT) accompanied by 2 or more arms; (2) subjects who experienced coronary revascularization <1 year but were asymptomatic or stable; (3) subjects analyzed >1 year after revascularization; (4) asymptomatic or stable subjects who underwent coronary angiography and showed at least one vessel with stenosis >50% but did not undergo revascularization; (5) antithrombotic therapy, including anticoagulant or antiplatelet therapy; (6) reported major cardiovascular and cerebrovascular events (MACEs) and major bleeding accompanied by follow‐up events for more than 12 months.

The following exclusion criteria were used: (1) the default dual antiplatelet treatment duration was not completed, and (2) subjects who used oral anticoagulants or low molecular weight heparin for a long time before grouping.

Data Extraction

The methods of data extraction were outlined in our previous study. ¹⁵ All selected papers were reviewed by 2 reviewers who independently extracted the data. The following data were extracted: the study design, baseline characteristics, interventions, and outcomes. Following the extraction of relevant data by the 2 reviewers, data were examined for possible inconsistencies that were then resolved by discussion, and if consensus was unable to be reached, a third author was consulted. Studies were not conducted directly on humans, and ethical approval was therefore unnecessary.

Quality Assessment

Two reviewers used the 7 domains of the Cochrane risk‐of‐bias tool to evaluate the quality of the included studies on the basis of the following criteria: randomization sequence generation, concealment of randomization sequence, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Studies were classified as having a low, high, or unclear risk of bias for each item, as suggested in the Cochrane Handbook. ¹⁶

Outcome Measures

The primary efficacy outcome was trial‐defined MACEs, which was often defined as a combination of death from any cause or cardiovascular death, MI, and stroke; secondary efficacy outcomes were individual components of MACEs. The primary safety outcomes were trial‐defined major bleeding events and Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries (GUSTO) criteria–defined severe bleeding; secondary safety outcomes were trial‐defined minor bleeding events, GUSTO‐defined moderate bleeding, and intracranial hemorrhage.

Statistical Analysis

A standard paired meta‐analysis was performed using the DerSimonian‐Laird random‐effects model. The odds ratios (ORs) and 95% CIs served as a summary statistic. Statistically significant results were those results where the 95% CI did not include 1. The heterogeneity test was completed using the χ²‐based Q‐test, and a P value <0.1 was considered to indicate heterogeneous results, whereas a P value >0.1 was considered to indicate a lack of heterogeneity. If heterogeneity was observed in the results, the degree of heterogeneity was determined using the I² test (I²=0–25%, no heterogeneity; I²=25–50%, moderate heterogeneity; I²=50–75%, substantial heterogeneity; and I²=75–100%, extreme heterogeneity).

A network meta‐analysis was performed using the frequentist approach. The OR (95% CI) served as a summary statistic. We calculated the surface under the cumulative ranking (SUCRA) value to evaluate the rankings of treatment strategies. SUCRA values are presented as the percentage of the area under the cumulative rank probability curve and the entire plane of the plot. A smaller SUCRA value resulted in a lower incidence of adverse outcomes, indicating better efficacy of the treatment regimen. An examination of the assumption in the network meta‐analysis includes homogeneity, transitivity, and consistency. The examination of the homogeneity assumption was performed through direct treatment comparisons, and thus the χ²‐based Q‐test and I² test were used for the analysis. The transitivity assumption was assessed by comparing the distribution of clinical variables, which were considered interfering factors that might affect the outcomes. The consistency assumption was tested to verify the feasibility of mixed comparisons (ie, no inconsistency in the evidence between direct and indirect treatment comparisons). A design‐by‐treatment approach was used to assess inconsistency in the entire analytical network, ¹⁷ and a loop‐specific approach and node‐splitting approach were used to assess local inconsistency. In addition, subgroup network meta‐analyses were conducted of subjects with or without a history of PCI (prespecified) or with or without a history of prior MI to assess whether the results of the study were affected by the study characteristics (effect modifiers). The verified data were analyzed using Stata software (version 15.0; Stata Corporation, College Station, TX), REVMAN software (version 5.3; Cochrane Collaboration, Oxford, UK) and Word Processing System (version 2.5; Beijing, China).

Results

Literature Search

The literature search identified 945 records in Medline, 1564 records in EMBASE, and 9 records in the Cochrane database (Figure S1). After checking for duplicates, 18 unique and full‐text published articles remained. A brief review of the abstract and manuscript of these 18 articles resulted in 4 studies that were appropriate for a detailed review; all 4 studies (THEMIS [Ticagrelor on Health Outcomes in Diabetes Mellitus Patients Intervention Study], COMPASS [Cardiovascular Outcomes for People Using Anticoagulation Strategies], PEGASUS‐TIMI 54 [Prevention of Cardiovascular Events in Patients With Prior Heart Attack Using Ticagrelor Compared to Placebo on a Background of Aspirin–Thrombolysis in Myocardial Infarction 54], and DAPT [Dual Anti‐platelet Therapy]) were included in the network meta‐analysis. ⁷ , ⁹ , ¹⁰ , ¹³ The remaining 14 articles were excluded because they did not meet the inclusion criteria or met the exclusion criteria (Table S2). Briefly, 2 studies ¹⁸ , ¹⁹ were excluded because they described an unplanned post hoc analysis that undermined the randomization principle; 5 studies ²⁰ , ²¹ , ²² , ²³ , ²⁴ were excluded because the end point of interest was not reported; 1 study ²⁵ was excluded because some of the included subjects were patients without stable coronary artery disease; 5 studies ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ were excluded because patients with ACS were not rerandomized at follow‐up after the end of default dual antiplatelet therapy, that is, the follow‐up process after the nodes meeting this inclusion criterion had disrupted the randomization principle; and 1 study ³¹ was excluded because of noncompliance with the standard dual antiplatelet principle.

Characteristics of the Included Studies and Patients

The main characteristics of these studies are reported in Table S3. These RCTs were published between 2014 and 2019; the follow‐up period ranged from 18.0 to 39.9 months, and all 4 RCTs included in this network meta‐analysis were randomized, double‐blind, placebo‐controlled trials. In the THEMIS and PEGASUS‐TIMI 54 studies, ticagrelor was used as an intervention drug. Rivaroxaban was used as the intervention in the COMPASS study. In the DAPT study, thienopyridines were used as intervention drugs, including clopidogrel and prasugrel. All patients were treated with aspirin, except for patients included in one arm of the COMPASS trial, who were treated with rivaroxaban monotherapy.

The main clinical features of the patients are shown in Table S4. A total of 75 167 patients were included in this network meta‐analysis, and the sample size of a single RCT ranged from 9961 to 21 162. The overall prevalence of a history of PCI ranged from 57.8% to 100%. The mean age ranged from 61.6 to 69.0 years, 20.0% to 31.6% of patients were female, and 62.9% to 91.4% were White. With the exception of the DAPT trial, which did not report the number of vessels involved, more than 55% of patients had multiple coronary artery diseases, and most patients were at high risk of thromboembolic and bleeding events. Risk factors for atherosclerotic cardiovascular disease, such as hypertension, dyslipidemia, and diabetes mellitus, were common in the analyzed patients.

Quality of Studies

The quality assessment of the included studies is presented in Figure S2 and Table S5. All studies mentioned the use of randomized allocation, and the use of computer or network system method for randomization of the groups was considered a low risk of bias. The allocation concealment method of all studies was completed through an interactive voice response or network response system and was considered a low risk of bias. All trials were double‐blind, the outcome indicators were objective end points, a subjective evaluation was not performed, and the implementation of the blinding method would not be destroyed; thus, the performance bias and the detection bias of all trials were considered low risks. Completion rates for all trials were >90%, and missing data were adequately explained; therefore, incomplete outcome data for all trials were considered a low risk of bias. Although all trials were sponsored by pharmacists, these individuals were not involved in the analysis of the data, and thus the possible effect of the pharmacists on the results of all trials was considered a low risk of bias. All RCTs were judged to be at a low risk of bias.

Results of Homogeneity, Transitivity, and Consistency Analyses

The homogeneity assumption was P_{heterogeneity}=0.440 (I²=0%) for trial‐defined MACEs (Figure S3), P_{heterogeneity} = 0.724 (I²=0%) for all‐cause death, P_{heterogeneity}=0.111 (I²=60.6%) for cardiovascular death, P_{heterogeneity}=0.954 (I²=0%) for MI, P_{heterogeneity}=0.824 (I²=0%) for stroke, P_{heterogeneity}=0.646 (I²=0%) for trial‐defined major bleeding events, and P_{heterogeneity}=0.520 (I²=0%) for intracranial hemorrhage. The results of the transitivity assessment, which are presented in a combined histogram, showed that with the exception of the proportion of PCI history, the mean age, the proportion of hypertension, the proportion of diabetes mellitus, and the proportion of multivessel coronary arteries were relatively similar across compared treatment groups (Figure S4). In this network analysis, because all arms in the included studies were directly compared and a mixed comparison was not conducted (ie, no source of inconsistency was identified), global inconsistency testing and the node‐splitting approach were not necessary. In addition, only 1 loop was present in the structure of this network meta‐analysis, and the loop belonged to a multiarm trial of the same study; thus, evidence inconsistency did not exist.

Structure of Network Meta‐Analysis

Figure 1 shows the network of treatment regimens used in the analysis of the major efficacy outcome and major safety outcomes. We compared 5 treatment strategies: aspirin, ticagrelor plus aspirin, rivaroxaban plus aspirin, rivaroxaban, and thienopyridine plus aspirin. We used aspirin as a reference because all 5 RCTs studied this regimen.

Figure 1. Evidence structure of eligible comparisons for network meta‐analysis.

Lines connect the interventions that have been studied in head‐to‐head (direct) comparisons in the eligible RCTs. The width of the lines represents the cumulative number of RCTs for each pairwise comparison and the size of every node is proportional to the number of randomized participants (sample size). GUSTO indicates Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries criteria; MACEs, major adverse cardiovascular and cerebrovascular events; and RCTs, randomized controlled trials.

Efficacy Outcomes

The results of the network meta‐analysis showed that, with the exception of rivaroxaban monotherapy (OR, 0.89; 95% CI, 0.78–1.02), ticagrelor plus aspirin (OR, 0.86; 95% CI, 0.80–0.93), rivaroxaban plus aspirin (OR, 0.74; 95% CI, 0.64–0.85), and thienopyridine plus aspirin (OR, 0.72; 95% CI, 0.60–0.86) regimens all significantly reduced the trial‐defined MACEs compared with aspirin monotherapy (Figure 2 and Table S6). In addition, rivaroxaban plus aspirin significantly reduced the incidence of MACEs compared with rivaroxaban monotherapy (OR, 0.83; 0.72–0.96) regimen. For death from any cause, the improvement effect of rivaroxaban plus aspirin was detected versus aspirin monotherapy (OR, 0.76; 95% CI, 0.65–0.90), ticagrelor plus aspirin (OR, 0.79; 95% CI, 0.66–0.95), rivaroxaban monotherapy (OR, 0.82; 95% CI, 0.69–0.97), and thienopyridine plus aspirin (OR, 0.58; 95% CI, 0.41–0.82) regimens. Rivaroxaban plus aspirin seemed to reduce the incidence of cardiovascular death compared with aspirin monotherapy (OR, 0.75; 95% CI, 0.55–1.01) and rivaroxaban monotherapy (OR, 0.78; 95% CI, 0.58–1.06) regimens, although not statistically significant. Compared with aspirin monotherapy, both ticagrelor plus aspirin (OR, 0.83; 95% CI, 0.74–0.92) and thienopyridine plus aspirin (OR, 0.48; 95% CI, 0.38–0.62) regimens significantly reduce MI. Compared with aspirin monotherapy, ticagrelor plus aspirin (OR, 0.80; 95% CI, 0.69–0.93) and rivaroxaban plus aspirin (OR, 0.56; 95% CI, 0.42–0.75) regimens significantly reduced stroke, and the improvement effect of rivaroxaban plus aspirin was detected as compared with ticagrelor plus aspirin (OR, 0.70; 95% CI, 0.51–0.97) and rivaroxaban monotherapy (OR, 0.70; 95% CI, 0.52,0.94) regimens.

Figure 2. Forest plots for efficacy outcomes.

A, Trial‐defined MACEs. B, Death from any cause. C, Cardiovascular death. D, Myocardial infarction. E, Stroke. MACEs indicates major adverse cardiovascular and cerebrovascular events; and OR, odds ratio.

Safety Outcomes

The results of the network meta‐analysis showed that, compared with aspirin monotherapy, all exploratory strategies including ticagrelor plus aspirin (OR, 2.15; 95% CI, 1.78–2.59]), rivaroxaban monotherapy (OR, 1.51; 95% CI, 1.23–1.85) and rivaroxaban plus aspirin (OR, 1.68; 95% CI, 1.37–2.05) increased major bleeding events (Figure 3 and Table S6). However, rivaroxaban (OR, 1.54; 95% CI, 0.95–2.50), rivaroxaban plus aspirin (OR, 1.06; 95% CI, 0.63–1.79), and thienopyridine plus aspirin (OR, 1.45; 95% CI, 0.85–2.45) regimens did not increase the incidence of GUSTO‐defined severe bleeding compared with the aspirin monotherapy. Compared with aspirin monotherapy, ticagrelor plus aspirin (OR, 3.39; 95% CI, 2.07–5.57), rivaroxaban monotherapy (OR, 1.56; 95% CI, 1.38–1.77) and rivaroxaban plus aspirin (OR, 1.77; 95% CI, 1.57–2.00) regimens increased minor bleeding events. And rivaroxaban monotherapy (OR, 1.63; 95% CI, 1.07–2.49), rivaroxaban plus aspirin (OR, 2.05; 95% CI, 1.20–3.49), and thienopyridine plus aspirin (OR, 1.68; 95% CI, 1.16–2.43) regimens all increased GUSTO‐defined moderate bleeding events versus aspirin monotherapy. With the exception of rivaroxaban plus aspirin (OR, 1.12; 0.64–1.97), both ticagrelor plus aspirin (OR, 1.41; 95% CI, 1.05–1.90) and rivaroxaban monotherapy (OR, 1.88; 95% CI, 1.13–3.12) regimens significantly increased the intracranial hemorrhage compared with aspirin monotherapy.

Figure 3. Forest plots for safety outcomes.

A, Trial‐defined major bleeding. B, GUSTO major bleeding. C, Trial‐defined minor bleeding. D, GUSTO moderate bleeding. E, Intracranial hemorrhage. GUSTO indicates Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries criteria; and OR, odds ratio.

Ranking of Treatment Strategies

Table 1 shows the SUCRA values for efficacy outcomes and safety outcomes. A smaller SUCRA value indicates a lower incidence of adverse outcomes, indicating better efficacy of the treatment regimen. The rivaroxaban plus aspirin regimen achieved the greatest number of best performance rankings for all efficacy outcomes (SUCRA value, 15.9 for MACEs, 9.8 for cardiovascular death, 51.0 for MI, and 2.3 for stroke), while aspirin unexpectedly achieved the greatest number of worst performance rankings (98.8 for MACEs, 74.6 for cardiovascular death, 94.0 for MI, and 93.0 for stroke). Among all safety outcomes, aspirin was again unexpectedly the regimen with the largest number of best performance rankings (0 for major bleeding, 18.0 for severe bleeding, 0 for minor bleeding, 0.6 for moderate bleeding, and 12.0 for intracranial hemorrhage), while the ticagrelor plus aspirin regimen appeared to be the regimen with the largest number of worst performance rankings (98.5 for major bleeding, 99.8 for minor bleeding, and 64.1 for intracranial hemorrhage). Notably, the best regimen for preventing all‐cause mortality was the rivaroxaban plus aspirin regimen (0,4), while the worst regimen was the thienopyridine plus aspirin regimen (97.5). The cumulative rank probability plots showing the efficacy outcomes and safety outcomes for each treatment regimen are shown in Figure S5.

Table  .

SUCRA Values^* for Each Treatment Regimen and Outcomes

Value	Treatment Regimen
	Aspirin	Ticagrelor + Aspirin	Rivaroxaban + Aspirin	Rivaroxaban	Thienopyridine + Aspirin
Efficacy outcome
Trial‐defined MACEs	98.8	57.6	15.9	66.1	11.6
Cardiovascular death	74.6	51.5	9.8	59.8	54.4
MI	94.0	40.1	51.0	64.9	0
Stroke	93.0	46.3	2.3	49.9	58.5
All‐cause death	66.8	45.9	0.4	39.3	97.5

Safety outcome
Major †	Severe bleeding ‡	0	18.0	98.5	NA	63.7	28.6	37.7	81.9	NA	71.5
Minor †	Moderate bleeding ‡	0	0.6	99.8	NA	66.4	85.3	33.9	52.9	NA	61.2
Intracranial hemorrhage		12.0		64.1		30.5		93.4		NA

MACEs indicates major adverse cardiovascular and cerebrovascular events; MI, myocardial infarction; NA, not available; and SUCRA, surface under the cumulative ranking.

^*SUCRA values are presented as percentage of area under the cumulative rank probability curve and the entire plane of the plot. The smaller the SUCRA value, the less incidence of adverse outcomes, which means the better the treatment regimen performance.

^†Major/minor bleeding is defined by the respective trials.

^‡Severe/moderate bleeding is defined by the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries (GUSTO) criteria.

Subgroup Analysis

The subgroup analysis of whether subjects undergoing PCI (Figure S6 and Table S7) confirmed that in the PCI subgroup, ticagrelor plus aspirin (OR, 0.84; 95% CI, 0.73–0.96), ³² , ³³ rivaroxaban plus aspirin (OR, 0.72; 95% CI, 0.60–0.87), ³⁴ and thienopyridine plus aspirin (OR, 0.72; 95% CI, 0.60–0.86) regimens still reduced the occurrence of MACEs versus aspirin monotherapy, and the efficacy ranking based on SUCRA values was the same as the overall population. In the PCI subgroup, ticagrelor plus aspirin (OR, 1.82; 95% CI, 1.33–2.49), and rivaroxaban plus aspirin (OR, 1.73; 95% CI, 1.34–2.24) regimens still increased the incidence of major bleeding events versus aspirin monotherapy, and the efficacy rankings based on SUCRA values were also the same as the overall population. In the non‐PCI subgroup, compared with aspirin monotherapy, the rivaroxaban plus aspirin (OR, 0.76; 95% CI, 0.59–0.98) regimen still reduced the occurrence of MACEs, the ticagrelor plus aspirin (OR, 0.90; 0.77–1.06) regimen did not significantly reduce the occurrence of MACEs. The safety rankings based on SUCRA values were also the same as the overall population. In the non‐PCI subgroup, ticagrelor plus aspirin (OR, 2.32; 1.67–3.22) and rivaroxaban plus aspirin (OR, 1.61; 1.16–2.21) regimens still increased the incidence of major bleeding events versus aspirin monotherapy. The rankings of SUCRA values was also the same as in the overall population.

The subgroup analysis of whether subjects had a history of prior MI (Figure S7 and Table S8) confirmed that in the prior MI subgroup, both ticagrelor plus aspirin (OR, 0.84; 95% CI, 0.71–0.99), rivaroxaban plus aspirin (OR, 0.74; 95% CI, 0.60–0.91) and thienopyridine plus aspirin (OR, 0.69; 95% CI, 0.48–0.98) regimens still reduced the occurrence of MACEs versus aspirin monotherapy, and the efficacy ranking based on SUCRA values was the same as the overall population. In the prior MI subgroup, ticagrelor plus aspirin (OR, 2.27; 95% CI, 1.67–3.09), and rivaroxaban plus aspirin regimens (OR, 1.64; 95% CI, 1.27–2.11) still increased the incidence of major bleeding events versus aspirin monotherapy, and the efficacy rankings based on SUCRA values were also the same as the overall population. In the non–prior MI subgroup, compared with aspirin monotherapy, the ticagrelor plus aspirin (OR, 0.89; 95% CI, 0.68–1.16), rivaroxaban plus aspirin (OR, 0.74; 95% CI, 0.52–1.03) and thienopyridine plus aspirin (OR, 0.74; 95% CI, 0.50–1.07) regimens appeared to reduce the incidence of MACEs, although not statistically significant. The safety rankings based on SUCRA values were also similar to the overall population. In the non–prior MI subgroup, ticagrelor plus aspirin (OR, 2.08; 95% CI, 1.50–2.87) and rivaroxaban plus aspirin (OR, 1.76; 95% CI, 1.20–2.58) regimens still increased the incidence of major bleeding events versus aspirin monotherapy. The rankings of SUCRA values were also the same as in the overall population. Overall, the results of the subgroup analysis were generally consistent with the overall population.

Discussion

The main findings of this network meta‐analysis are listed below.

1. For regimens based on aspirin, the addition of any antithrombotic drug reduces the incidence of MACEs compared with aspirin alone.

2. In the safety evaluation, other antithrombotic regimens increase the risk of major bleeding events compared with aspirin monotherapy, and ticagrelor plus aspirin regimen appears to have low safety.

3. Considering the outcomes of all ischemic and bleeding events and all‐cause mortality, rivaroxaban plus aspirin appears to be the preferred long‐term antithrombotic regimen for patients with CCS and high‐risk factors. However, rivaroxaban monotherapy should be avoided.

Comparison of Antithrombotic Strategies With Anticoagulants and Antiplatelet Agents

Both anticoagulants and antiplatelet drugs play an important role in antithrombotic therapy. According to a previous meta‐analysis, ³⁵ the vitamin K antagonist (warfarin) plus aspirin regimen had an additional benefit in preventing ischemic events compared with the aspirin monotherapy. However, its clinical application was limited because anticoagulant therapy increased severe bleeding events, including intracranial hemorrhage. The COMPASS study ¹³ published in 2017 revealed advantages of the new oral anticoagulant (rivaroxaban) plus aspirin regimen compared with the aspirin‐only regimen in terms of preventing ischemic events, and although it also increased the risk of major bleeding, it did not increase intracranial hemorrhage and fatal bleeding. The results of this network meta‐analysis were based on odds ratios (95% CIs) and rankings of the efficacy and safety based on SUCRA values. Of all the antithrombotic strategies, except for the outcome of MI, rivaroxaban plus aspirin seemed to be the best in terms of anti‐ischemia. And it also reduced all‐cause mortality, suggesting that increased bleeding does not offset the benefits of reduced ischemic events. Rivaroxaban plus aspirin significantly reduced all‐cause mortality compared with all other antithrombotic regimens, which further indicated that the clinical net benefit of rivaroxaban combined with aspirin was the highest among all antithrombotic strategies. Moreover, in terms of the prevention of cerebrovascular events, the rivaroxaban plus aspirin regimen significantly reduced the incidence of ischemic events without increasing bleeding events compared with aspirin monotherapy. Unexpectedly, both GUSTO‐defined severe bleeding events and intracranial hemorrhage were more frequent with the rivaroxaban alone than with rivaroxaban plus aspirin, which is difficult to explain. This may be attributable to an imbalance in baseline levels of some factors between groups.

Antithrombotic Strategies With or Without Aspirin

Aspirin plays an important role in the secondary prevention of atherosclerotic cardiovascular disease. ³ , ³⁶ However, during the long‐term antithrombotic treatment of chronic coronary syndrome, controversy exists regarding whether aspirin is the first choice or whether aspirin must be included in dual antithrombotic therapy. A previous RCT (CAPRIE [Clopidogrel Versus Aspirin in Patients at Risk of Ischaemic Events]) showed that the long‐term use of clopidogrel in patients with atherosclerotic vascular disease was more effective at reducing the overall risk of ischemic stroke than aspirin. ⁵ However, a meta‐analysis of clopidogrel versus aspirin alone for stable coronary artery disease showed no significant differences in ischemic and bleeding events. ³⁷ A randomized multicenter trial ⁶ published in 2018 did not reveal an advantage of long‐term oral administration of ticagrelor alone compared with aspirin in ischemic and bleeding events in patients after stent implantation. In our network meta‐analysis, there is no direct comparison between aspirin and P2Y12 receptor inhibitors. Aspirin combined with P2Y12 receptor inhibitors improved vascular events compared with aspirin monotherapy, but the overall improvement seemed to be less effective than rivaroxaban combined with aspirin, with lower safety compared with aspirin alone and rivaroxaban combined with aspirin. The effect of rivaroxaban alone was similar to that of aspirin alone in anti‐ischemic events, while rivaroxaban alone significantly increased the incidence of bleeding events. However, rivaroxaban combined with aspirin caused fewer MACEs than rivaroxaban or aspirin monotherapy. Similar results occurred in all‐cause death, cardiovascular death, MI, and stroke, which suggested that combination of low‐dose of rivaroxaban on an aspirin basis may be a recommended option for patients with CCS with high‐risk factors. Furthermore, this prespecified subgroup analysis, based on whether or not patients undergoing PCI, suggested robustness of the overall outcome, while it revealed that rivaroxaban plus aspirin may be the recommended long‐term antithrombotic regimen for patients with CCS regardless of whether they are undergoing PCI.

Antiplatelet Strategies With Thienopyridines Versus Nonthienopyridines

According to the PLATO (Platelet Inhibition and Patient Outcomes) study, ³⁸ ticagrelor has a higher priority than clopidogrel in dual antiplatelet therapy for patients with ACS. However, a recently published RCT, the POPular AGE (Clopidogrel Versus Ticagrelor or Prasugrel in Patients Aged 70 Years or Older With Non–ST‐Elevation Acute Coronary Syndrome: The Randomised, Open‐Label, Non‐Inferiority Trial) study, included elderly (>70 years) and nonischemic high‐risk patients with ACS who had a higher risk of bleeding and a lower risk of ischemia than subjects in the PLATO study, and the results favored clopidogrel over ticagrelor. ³⁹ In our network meta‐analysis, the thienopyridine plus aspirin seemed to be more safe and even more effective than ticagrelor plus aspirin regimen, which may also suggest that some patients who need prolonged dual antiplatelet therapy are more likely to choose thienopyridine plus aspirin in the choice of antithrombotic drugs. Notably, the thienopyridine plus aspirin regimen increased the incidence of all‐cause mortality compared with aspirin monotherapy, and the authors of the study explained this result as an imbalance in the cancer rates at baseline among the included patients, as the thienopyridine plus aspirin regimen group included more patients with cancer and an increased mortality rate, which the authors subsequently balanced. After correction, all‐cause mortality was not significantly different between the 2 groups.

Mechanistic Insights Into Antithrombotic Regimens

The most common cause of coronary heart disease is atherosclerosis, and any external factor may induce the rupture of vulnerable plaques or subintimal hemorrhage to promote platelet aggregation^26–28; therefore, antiplatelet therapy is essential. In addition, coronary stenosis caused by severe atherosclerotic plaques affects the velocity of intravascular blood flow, which tends to be as slow as venous flow, and thus anticoagulation is also needed. ⁴⁰ This network meta‐analysis suggests that the rivaroxaban plus aspirin regimen appears to have a good advantage in preventing ischemia that is even greater than the dual antiplatelet regimen, which may be attributed to the fact that approximately 60% of the patients with chronic coronary syndromes included in this analysis have multivessel coronary artery disease, which is likely to alter the coronary blood flow.

Limitations

The current analysis has some limitations. First, although clear statistical heterogeneity was not observed in our network meta‐analysis, some clinical heterogeneity was identified among the studies, with potential sources including inclusion and exclusion criteria for patients, dose and course of treatment with drugs, definition of outcomes, and follow‐up time, which may affect the interpretation of our results. Second, although the sample size was sufficient for the primary efficacy and safety outcomes, the majority of included patients were White, and thus these data may not apply to other races. Additional data on the efficacy in different races must be refined. In addition, the analysis of some rare and clinically interesting end points, such as stent thrombosis, was not included because these end points were not reported in half of the included studies. Finally, fewer trials with the same exploratory treatment group were included in this analysis, and more studies with more similarities are needed in the future to provide more robust results.

Conclusions

In terms of long‐term antithrombotic strategies for patients with chronic coronary syndrome and high‐risk factors, all antithrombotic strategies combined with aspirin significantly reduced the incidence of MACEs and increased the risk of major bleeding events compared with aspirin monotherapy. However, compared with aspirin monotherapy, rivaroxaban plus aspirin reduced all‐cause mortality and was the only strategy with a net clinical benefit in preventing cerebrovascular events. In addition, the net clinical benefit of rivaroxaban monotherapy and the ticagrelor plus aspirin regimen might be lower than other strategies. However, additional large‐scale clinical trials must be conducted to further determine the appropriate long‐term antithrombotic regimens for patients with CCS and high‐risk factors. On the basis of the available evidence, our results tend to support the hypothesis that the rivaroxaban plus aspirin regimen has a good application prospect.

Sources of Funding

This study was supported by the Zhejiang Health Commission (2021KY916), the Science Technology Department of Zhejiang Province (2019C03SA100640), and the Hangzhou Bureau of Science and Technology (A20200624). The sponsors played no role in the study design, data collection and analysis, or decision to submit the article for publication.

Disclosures

None.

Supporting information

Tables S1–S8

Figures S1–S7

Reference 41

(J Am Heart Assoc. 2021;10:e019184. DOI: 10.1161/JAHA.120.019184.)