Anticoagulation in atrial fibrillation and liver disease: a pooled-analysis of >20 000 patients.

Aims: Anticoagulation for atrial fibrillation patients with liver disease represents a clinical dilemma. We sought to evaluate the efficacy/safety of different anticoagulation, i.e. vitamin K antagonist (VKA) and non-VKA oral anticoagulants (NOACs) in such patient group.

Methods and results: This was a pooled-analysis enrolling up-to-date clinical data. Two subsets: subset A (VKA vs. Non-Anticoagulation) and subset B (NOACs vs. VKA) were pre-specified. The study outcomes were ischaemic stroke (IS)/thromboembolism (TE), major bleeding (MB), intracranial bleeding (ICB), gastrointestinal bleeding (GIB), and all-cause mortality. A total of 20 042 patients' data were analysed (subset A: N = 10 275, subset B: N = 9767). Overall mean age: 71 ± 11 years, mean CHA2DS2-VASc score: 4.0 ± 1.8, mean HAS-BLED score: 3.6 ± 1.2. The majority of the patients had Child-Pugh category (A-B). As compared with Non-Anticoagulation, VKA seemed to reduce the risk of IS/TE [odds ratio (OR): 0.60, P = 0.05], but heighten the risk of all-bleeding events including MB (OR: 2.81, P = 0.01), ICB (OR: 1.60, P = 0.01), and GIB (OR: 3.32, P = 0.01). When compared with VKA, NOACs had similar efficacy in reducing the risk of IS/TE (OR: 0.82, P = 0.64), significantly lower risk of MB (OR: 0.54, P = 0.0003) and ICB (OR: 0.35, P < 0.0001), and trend towards reduced risk of GIB (OR: 0.72, P = 0.12) and all-cause mortality (OR: 0.79, P = 0.35). The favourable effects were maintained in subgroups of individual NOAC.

Conclusions: VKA appears to reduce the risk of IS/TE but increase all-bleeding events. NOACs have similar effect in reducing the risk of IS/TE and have significantly lower risk of MB and ICB as compared with VKA. NOACs seem to be associated with better clinical outcome than VKA in patients with mild-moderate liver disease.