Efficacy and risk of bleeding of new oral anticoagulants in treatment of acute venous thromboembolism: a network meta-analysis

Abstract

Objective To systematically evaluate the efficacy and bleeding risk of new oral anticoagulants (NOAC) in treatment for acute venous thromboembolism (VTE). Methods Randomized controlled trials (RCT) about NOAC treatment for VTE in related databases collected up to December 2017 were searched. The literature in accordance with the inclusion criteria were selected and evaluated. The outcome measures included recurrence rate of VTE, all-cause mortality, incidence of hemorrhea or incidence of clinically relevant non-major (CRNM) bleeding. The results of RCT in accordance with inclusion criteria were combined using network meta-analysis. The effect measures were expressed in odds ratio (OR) with its corresponding 95% confidence interval (CI). Results A total of 9 RCT comprising 27 827 patients and 5 anticoagulant drugs (rivaroxaban, apixaban, edoxaban, dabigatran, and warfarin) were enrolled in the study. The quality evaluation showed that 4 of 9 RCT′s bias risks were low, 2 were uncertain, and 3 were high. Network meta-analysis results showed that neither the differences of recurrence rate of symptomatic VTE nor that of all-cause mortality among the patients with acute VTE using 5 different anticoagulant drugs were statistically significant. The results of surface under the cumulative ranking probability (SUCRA) showed that apixaban had the highest probability of having the optimal curative effect in decreasing symptomatic recurrence rate and all-cause mortality (SUCRA: 69.0%, 77.1%), followed by rivaroxaban (SUCRA: 60.7%, 51.4%). About the safety of the drugs, the risk of hemorrhoea induced by apixaban was lower than that induced by dabigatran and edoxaban (dabigatran vs. apixaban, OR=2.36, 95%CI: 1.16-4.83; edoxaban vs. apixaban, OR=2.64, 95%CI: 1.36-5.13). The risk of CRNM bleeding associated with apixaban was lower than that associated with rivaroxaban and edoxaban (edoxaban vs. apixaban, OR=1.67, 95%CI: 1.26-2.20; rivaroxaban vs. apixaban, OR=2.09, 95%CI: 1.59-2.74). The risk of CRNM bleeding associated with dabigatran was lower than that due to edoxaban and rivaroxaban (edoxaban vs. dabigatran, OR=1.36, 95%CI: 1.01-1.84; rivaroxaban vs. dabigatran, OR=1.71, 95%CI: 1.27-2.30). The results of SUCRA showed that the probability of hemorrhoea induced by apixaban was the lowest (SUCRA: 98.2%), followed by rivaroxaban, dabigatran, edoxaban, and warfarin; the probability of CRNM bleeding induced by apixaban was the lowest (97.0%), then followed by dabigatran, edoxaban, rivaroxaban, and warfarin. Conclusions For adult patients with new-onset or relapsed acute VTE, the efficacy of NOAC and warfarin were similar. The risk of bleeding with NOAC treatment was significantly lower than that with warfarin, of which, apixaban had the highest probability of having the best efficacy and the lowest risk of bleeding. Key words: Anticoagulants, oral administration; Venous thromboembolism; Hemorrhage; Network meta-analysis