Oral anticoagulation in patients with left ventricular thrombus: a systematic review and network meta-analysis with reconstructed time-to-event data

Sebastian Heyne,Sascha Macherey-Meyer,Max M Meertens,Karl Finke,Stephan Baldus,Christoph Adler,Samuel Lee
DOI: https://doi.org/10.1007/s00392-024-02547-7
2024-09-16
Abstract:Background: Left ventricular thrombus (LVT) is associated with high rates of systemic embolism. Vitamin K antagonists (VKAs) are the only approved treatment for LVT. Although evidence suggests direct oral anticoagulant (DOACs) to be at least equally effective in general, the efficacy of individual DOACs remains unclear. Methods: A literature search was performed in EMBASE, MEDLINE and Web of Science looking for randomized controlled trials (RCTs) and non-randomized controlled studies of interventions (NRSI) comparing individual DOACs to VKAs for the treatment of LVT. Individual patient data was reconstructed and incorporated in a Bayesian network meta-analysis (NMA) and a Cox frailty regression model. Results: A total of 2545 patients across 19 studies (4 RCTs, 15 NRSI) were included. 1738 received VKAs, 581 received Rivaroxaban, 226 received Apixaban, 82 received Dabigatran and 2 received Edoxaban. LVT resolution was less likely with VKAs compared to Rivaroxaban in the time-to-event analysis (HR 0.66, 95% CI [0.49; 0.91], p = 0.01). There was no difference for other DOACs compared to VKAs. Rivaroxaban reduced ischemic stroke compared to VKAs (OR 0.18, 95% CrI [0.05; 0.49]), other DOACs did not. Conclusion: In this NMA, Rivaroxaban showed faster LVT resolution and consecutively lower odds of ischemic stroke than VKAs while Apixaban and Dabigatran showed at least equal efficacy. Given the quality and size of the available studies, these differences between individual DOACs should be acknowledged as hypothesis generating only. Future adequately powered randomized controlled trials are needed to assess possible time-varying effects between individual DOACs.
What problem does this paper attempt to address?