Anticoagulation Resumption After Intracranial Haemorrhage With Mechanical Valves: A Data-free Zone

Freek W.A. Verheugt


Eur Heart J. 2018;39(19):1724-1725. 

Intracranial haemorrhage (ICH) is the most feared complication of oral anticoagulation for patients on vitamin K antagonists (VKAs). This concerns mostly patients with atrial fibrillation or venous thrombo-embolism, or those carrying mechanical heart valves. The incidence of VKA-associated ICH is ~0.7% per year with atrial fibrillation[1] and 0.5% per year in the patients with a mechanical heart valve.[2] Although often fatal, VKA-induced ICH can be effectively treated with a reversal strategy using prothrombin concentrate complex. Yet, the surviving patients are at risk for thrombo-embolic complications and, thus, in need of resumption of therapeutic anticoagulation, either with heparin followed by a VKA, or with just a VKA. Alternatively, non-vitamin K oral anticoagulants (NOACs) could be chosen, since they induce much less ICH,[3] but this only applies for patients with atrial fibrillation or venous thrombo-embolism, and not for carriers of mechanical heart valves.[4] The question remains, however, of when to restart therapeutic anticoagulation in mechanical heart valve patients surviving ICH. Proper studies on the subject are scarce[5,6] and, thus, guidelines do not address this issue specifically.[4,7,8] Therefore, the area is somewhat data free.

In this issue of the journal, a co-operative study group from Germany collected, in 22 tertiary care centres over 10 years, 2504 patients with oral anticoagulation-related ICH, of whom 137 had mechanical heart valves.[9] The authors studied the international normalized ratio (INR) at baseline which was, as expected, well within the target range. INR reversal was attempted in the management of ICH, but was unsuccessful in the majority of patients. Finally, the moment of restart of therapeutic anticoagulation, either parenteral, oral, or both, was studied in retrospect. Therapeutic anticoagulation was started in only 48% of patients. After anticoagulation resumption, bleeding occurred in 15% of cases within 50 days and was more common than thrombo-embolic events (5%). In comparison with no resumption, restart of anticoagulation resulted in a significant increase in bleeding (from 6% to 26%), but also in a reduction of thrombo-embolic events (from 10% to 2%). Of the bleeding events, ICH recurrence was the most common complication, and the majority of thrombo-embolic events were intracranial. The most interesting findings of the study, however, were the timing plots of bleeding on one hand and thrombo-embolism on the other. Resumption of therapeutic anticoagulation earlier than 14 days after ICH was associated with more bleeding compared with restart after 14 days, whereas the combination of bleeding and ischaemic complications was also in favour of anticoagulation resumption later than 14 days. These findings also agreed well with the final neurological outcome. The combination of timing of anticoagulation restart on one hand and the neurological recovery on the other is the centrepiece of the study.

The authors must be complimented on their extensive and thorough work on the vexing issue of ICH in patients with a mechanical heart valve. In particular, the per 3 day balancing of the bleeding risk vs. the risk of thrombosis has been well worked out and illustrated. Also, the correction for treatment crossover was factored in. However, their work is not unique. In a previous systematic review on 88 cases of ICH in carriers of mechanical heart valves, no clear advice regarding restart of VKAs could be given.[5] There has been another study that was reported as a letter, where there was no correction for treatment crossovers and no data on neurological outcome were given. In that study in 141 patients, the advice was that restart of anticoagulation was optimal between 7 and 10 days after ICH.[6] In the current study, such a period proved to be too short and resulted in a worse neurological outcome.

Yet, as pointed out by the authors, their data are observational and retrospective over 10 years. Such analyses have their inherent shortcomings. A randomized trial would be ideal, but ethically difficult to perform. The data from the current study will end up in future guidelines, but will not change them.

In absolute numbers, VKA-related ICH is far more common in atrial fibrillation than in patients with a mechanic heart valve, and represents a huge problem. Clearly today's data are not applicable for atrial fibrillation patients. Patients with atrial fibrillation who are not anticoagulated have a risk of thrombo-embolism similar to that of untreated patients with mechanical heart valves.[10,11] Therefore, the large majority of patients with atrial fibrillation need oral anticoagulation, not the least because they have more co-morbidities leading to stroke, such as hypertension.[12] For resumption of oral anticoagulation after ICH in atrial fibrillation, there are good data showing that restart should take place at 7–8 weeks after ICH.[13] Not starting anticoagulation altogether or alternatively starting antiplatelet therapy in atrial fibrillation after ICH has a poor long-term mortality outcome.[14] Unlike for carriers of mechanical heart valves, for atrial fibrillation patients there are effective alternatives to reduce the risk for ICH, such as the use of NOACs or the insertion of a left atrial appendage-occluding devices (Table 1). Although attractive, the efficacy and safety of these strategies are currently under investigation in randomized trials, but unfortunately these options are not applicable for patients with mechanical heart valves for the reasons mentioned above. Finally, replacing the mechanical valve by a bioprosthesis after ICH seems unattractive given the risk of reoperation in general, and the risk of recurrent ICH in the peri-procedural phase in particular.

In conclusion, VKA-related ICH in patients with mechanical valves is not uncommon, and its optimal management remains problematic. Resumption of therapeutic anticoagulation can be postponed for at least 2 weeks, as can be learned from the extensive study in this issue.