Network Meta-Analysis of Ticagrelor for Stroke Prevention in Patients at High Risk for Cardiovascular or Cerebrovascular Events

Alexandra Bálint, MD; Dániel Tornyos, MD; Oumaima El Alaoui El Abdallaoui, MS; Péter Kupó, MD; András Komócsi, MD, DSc


Stroke. 2021;52(9):2809-2816. 

In This Article


The literature search resulted in 1335 citations of which 26 were compliant with the inclusion criteria (Figure I in the Data Supplement). Twenty-six randomized controlled trials involving 124 495 patients (range, 48–21 162) were included in the analysis. Clinical characteristics of the populations and procedural data are shown in Table I in the Data Supplement. Table II in the Data Supplement presents the main characteristics of the involved trials.

Patients were recruited to the trials due to nonsevere ischemic stroke or TIA,[10,19,20] acute coronary syndrome,[7,21–34] high risk for acute coronary syndrome,[35,36] peripheral artery disease,[37] coronary artery bypass graft surgery,[38,39] known coronary artery disease,[40,41] or transcatheter aortic valve implantation.[42] Six treatment arms were identified using aspirin or the P2Y12 inhibitors clopidogrel, prasugrel, or ticagrelor in monotherapy or combined with aspirin. The 6 antiplatelet treatment arms allowed 15 possible pairwise comparisons of which 7 were implemented in the included trials (Figure 1).

Figure 1.

Network layout and the results of the primary end points.
Network graphs depict the overall structure of comparisons of primary end points in our network. The numbers and the thickness of the edges correspond to the number of studies a specific comparison was tested within. The blue triangles mark the multiarm trials of the network. A–C, Forest plots show the results of the random-effects network meta-analyses as comparisons with aspirin (A) monotherapy set as reference (D–F). C indicates clopidogrel; P, prasugrel; RR, risk ratio; and T, ticagrelor.

The dose of the long-term P2Y12 inhibitor treatment was different in the trials using 90 mg BID or 60 mg BID for ticagrelor, 75 mg OD for clopidogrel, and 10 mg OD for prasugrel. Aspirin was administered in a low dose (75–150 mg).

Analysis of bias showed high quality of the source information with a low probability of bias. No obvious publication bias was found in Figures II and III in the Data Supplement. Neither net heat plots nor net-splitting analyses revealed major inconsistencies between direct and indirect evidence (Figures IV and V in the Data Supplement).

Association Between Ticagrelor Treatment and Clinical Outcomes

In the 26 trials, 3035 (2.43%) stroke events were reported. Compared with aspirin monotherapy, stroke risk was significantly 23% lower with aspirin plus clopidogrel and 20% lower with aspirin plus ticagrelor (A+T) combinations. With ticagrelor alone and with the combination of aspirin and prasugrel, stroke risk was also lower (11% and 24%, respectively) but 14% higher with clopidogrel monotherapy; however, these latter results did not reach the level of significance (Figure 1; Table III in the Data Supplement). The data were consistent (I2, 0% [0.0%–34.2%]) and without significant heterogeneity neither within designs nor between designs (P=0.6828 and P=0.8351, respectively).

The risk of ischemic stroke was significantly reduced with ticagrelor plus aspirin (RR, 0.80 [95% CI, 0.71–0.89]). Ticagrelor monotherapy also resulted in a decreasing trend in the risk of ischemic stroke (RR, 0.88 [95% CI, 0.77–1.00]; P=0.05). In the case of hemorrhagic stroke, none of the treatments influenced the risk significantly. Combination ticagrelor to aspirin increased the risk of intracranial bleeding with 53% (RR, 1.53 [95% CI, 1.16–2.03]; P=0.05). Data of ischemic stroke were consistent and homogenous, while in the case of hemorrhagic stroke, moderate heterogeneity was seen (I2, 47%; Table 1).

Twenty-three trials reported 5194 (4.20 %) mortality events. Compared with aspirin, mortality was 20% higher with aspirin plus clopidogrel and showed a decreasing trend with aspirin plus prasugrel (RR, 0.78 [95% CI, 0.59–1.03]). With the other treatments, the difference remained <10% and did not reach the level of statistical significance (Figure 1). Low degree of heterogeneity was noted in mortality data (I2, 12.3% [0.0%–47.1%]).

Twenty-one trials reported 2811 (2.5%) major bleeding events classified by the individual trial definitions. Compared with aspirin alone, major bleeding was in similar ranges with antiplatelet monotherapies while the relative risk was twice higher with combined antiplatelet therapies (Figure 1). Low-degree inconsistency was noted for major bleeding data (I2, 10.2% [0.0%–45.9%]).

Analyses of the clinical outcomes suggested clustering of treatment arms with antiplatelet monotherapies separating from combination therapies (Figure 2). Metaregression analyses did not find a signal for an important interaction between the background risk of the included population and the risk of stroke or major bleeding using the different antiplatelet regimes (Figure VI in the Data Supplement). Subgroup analyses stratified according to the inclusion conditions showed data consistent in all strata with more effective stroke reduction of the ticagrelor-plus-aspirin combination, as well as the higher risk of bleeding. Net adverse clinical event data showed a higher level of inconsistency and variances with nonsignificant relations except for the benefit of ticagrelor plus aspirin in acute coronary syndrome trials (Table V in the Data Supplement). In accordance with the metaregression analysis, subgrouping, and multilevel analysis reflected consistent results (Figure VII in the Data Supplement). Further exploratory analyses attempting aggregate analyses of prevention and bleeding trade-off neither at the level of disabling conditions nor with multivariate analysis of direct comparisons explored significant differences (Figure VIII in the Data Supplement).

Figure 2.

Cluster analysis of the included treatment arms.
A depicts the relative risk of stroke and major bleeding with their respective CIs related to aspirin monotherapy. Both the risk ratio values and the P score values showed a strong negative correlation between stroke and major bleeding risk (R, −0.871; P=0.024 and R, −0.899; P=0.015, respectively). B shows the 3-dimensional projection of the predictor space of the nearest neighbor analysis derived from the analysis of the 11 analyzed predictors. The plot shows discernible clustering of combined therapies and monotherapies. A indicates aspirin; ASA, aspirin alone; C, clopidogrel; P, prasugrel; and T, ticagrelor.

Ranking of Treatment Strategies

Clopidogrel plus aspirin and ticagrelor plus aspirin were ranked as the most effective strategies for the prevention of stroke (P score, 0.79 and 0.73, respectively). For the prevention of ischemic stroke, the ranking for A+T was higher (P score, 0.72 and 0.81, respectively; Table IV in the Data Supplement). Ranking with regard to the major bleeding or stroke prevention showed opposite tendencies (R, −0.879; P=0.021; Figure 2). Regarding major bleeding, aspirin was ranked as the safest strategy (P score, 0.82; Table IV in the Data Supplement).

Effect of the Individual Antiplatelets in the Component NMA Models

The component analysis reflected that the use of each antiplatelet agent conveyed the reduction of stroke risk, but this effect reached the level of statistical significance only in the case of ticagrelor. An important increase in bleeding risk was characteristic for all drugs; however, no important change in mortality risk was detected (Table 2).