Pattern and Management of Intracranial Haemorrhage with Rivaroxaban or Dabigatran Compared to Vitamin K Antagonists - Results of the Prospective Dresden Noac Registry

Background: In recent phase-III trials, the rate of intracranial haemorrhage (ICH) - the most feared complication of anticoagulant therapy - was around 0.8% per year for patients treated with vitamin K antagonists (VKA) and consistently lower (around 0.3-0.5%) for patients treated with the non-VKA oral anticoagulants (NOACs) rivaroxaban and dabigatran

However, patients in clinical trials present a selected population treated under a strict protocol in dedicated academic facilities. Consequently, the risk, management and outcome of ICH need to be evaluated in cohorts of patients treated with NOACs or VKA in daily care.

Aim: To evaluate the rate of ICH in patients treated with NOAC compared to VKA patients.

Patients and methods: The prospective NOAC registry was initiated in November 2011. A network of more than 230 physicians in the district of Saxony, Germany, enrol up to 3000 patients in the registry, which are prospectively followed by the central registry office for up to 60 months with central outcome event adjudication. For this analysis, every intracranial haemorrhage was identified in the database and ICH management and outcome were evaluated.

Results: Until January 31^th 2015, 2682 patients were registered and treated with dabigatran (348, mostly treated for atrial fibrillation), Rivaroxaban (1907; 1204 treated for atrial fibrillation and 703 for venous thromboembolism) or vitamin K antagonists (427; treated for atrial fibrillation). VKA patients had lower HAS-BLED scores compared to NOAC patients and were excellently managed with a time-in-therapeutic-range of 75%.

During follow-up (mean follow-up duration 25.6 months) ICH occurred in 7/427(1.6%) of VKA treated patients and in 14/2255 (0.6%) of patients treated with NOAC, which translated into an annualized rate of 1.3 events/100 pt. years (95%-CI 0.5-2.7) for VKA and 0.4 events/100 pt. years (95%-CI 0.2-0.6) for NOAC (p=0.0039).

Treatment of ICH consisted of PCC in 10 cases, plasma in 3 cases and surgical or interventional therapy in 7 cases (table 1, multiple treatments possible). As indicated, use of factor concentrates, plasma or other hemostatic agents or surgery was much more frequent in VKA patients compared to NOAC patients.

At Day 90 after ICH, 7/21 patients were dead (2/7 or 28.6% of VKA patients and 5/14 or 35.7% of NOAC patients). The surviving 14 patients received the following antithrombotic agents: 5 (35.7%) rivaroxaban, 3 (21.4%) heparin, 1 (7.1%) apixaban, 1 (7.1%) VKA, 3 (21.4%) aspirin, 1 (7.1%) none.Following ICH, oral anticoagulation therapy was either interrupted (n=7; 6 NOAC vs. 1 VKA) or permanently discontinued (n=10; 6 NOAC vs. 4 VKA).

Conclusion: Despite low bleeding risk and excellent INR control in our VKA cohort, the rate of ICH was higher than that of VKA patients treated in recent phase-III trials. Furthermore, ICH rates in our VKA cohort were significantly higher than those seen in our NOAC cohorts, which represented more patients with relevant comorbidities and higher bleeding risk. Consequently, the risk of ICH remains high even in "stable" VKA patients with good INR control and a preventive switch from VKA to NOAC may help to reduce ICH risk and should be discussed with the patient.