Danny R. Rose, Jr., MD

Murthy SB, Gupta A, Merkler AE, Navi BB, Mandava P, Iadecola C, et al. Restarting Anticoagulant Therapy After Intracranial Hemorrhage: A Systematic Review and Meta-Analysis. Stroke. 2017

For most clinicians, intracerebral hemorrhage (ICH) is the most feared potential complication of anticoagulation therapy, carrying significant morbidity and mortality. Clinical decision-making regarding the resumption of anticoagulation for patients for whom it is indicated is complex, as many of these patients have significant risk of ischemic and hemorrhagic events. Traditionally, providers have been reluctant to restart anticoagulation after ICH, especially for patients with atrial fibrillation as compared to mechanical valves.

AHA guidelines reflect the lack of clarity on this matter, with current recommendations (Class IIa) to avoid anticoagulation after spontaneous lobar ICH in patients with non-valvular atrial fibrillation (NVAF), and to consider resuming antiplatelet therapy after all ICH and anticoagulation in patients with non-lobar ICH (Class IIb). Selection and timing for resumption of anticoagulation has been the topic of ongoing research, with the results of pertinent studies presented at the International Stroke Conference, as well as the European Stroke Organisation Conference in 2017. In addition to this, a recent meta-analysis published in Stroke by Drs. Murthy et al. sought to address this clinical challenge by reviewing available studies with respect to the safety and efficacy of restarting anticoagulation after ICH. To understand the significance of this study in the context of evolving concepts regarding anticoagulation after ICH, we will start by reviewing a previous study with similar aims.

The CHIRONE study (Poli et al., Neurology. 2014 Mar 25;82(12):1020-6) was the first prospective observational cohort study looking at patients who were restarted on vitamin K antagonists after an intracranial hemorrhage. This study included patients with a variety of indications for anticoagulation, with slightly less than half having a diagnosis of atrial fibrillation and around one-third of patients having prosthetic heart valves. In addition, the index ICH in this study included both traumatic and spontaneous hemorrhages. These patients were followed for a median time of two years, and a rate of recurrence of 2.56 x 100 patient-years was recorded, a twofold to fivefold higher risk as compared to patients without a history of ICH. About 40% of these recurrences were fatal or severely disabling.

An accompanying editorial by Alejandro Rabinstein (Neurology 2014;82:1016–1017) commented on the various limitations of this study, especially in the context of applying it to risk-benefit determinations. Patients included were only those whose clinician determined that the risk-benefit assessment favored resumption of anticoagulation. In addition, the index ICH event was traumatic in the majority of cases, and the information regarding ischemic complications was lacking. The relatively small number of recurrent hemorrhages also limited the power of their analysis with respect to identifying risk factors significantly associated with recurrent ICH. Dr. Rabinstein aptly subtitled this editorial “A risky decision with no recipe,” and this was especially true in patients with atrial fibrillation, who are felt to represent a less “strict” indication for anticoagulation as compared to patients with mechanical valves.

The design and study selection in the meta-analysis by Drs. Murthy et al. addresses several of these limitations. They only included studies that tracked patients for which there was clear documentation of whether or not anticoagulation was restarted. In addition, they only included studies where both thromboembolic (ischemic stroke, myocardial infarction) outcomes and recurrent ICH were documented. Similar to the CHIRONE study, patients with a variety of indications for anticoagulation were included. Atrial fibrillation was the most common indication in the included studies, with prosthetic heart valve being the second most common.

Six studies with 2044 patients were included in the analysis of thrombotic complications. Roughly 40% of patients included in the above studies were restarted on anticoagulation therapy. Those patients had a significantly smaller rate of thromboembolic events (6.7% versus 17.6%) as compared to patients that were not restarted on anticoagulation. This resulted in an inverse association between anticoagulation and the risk of thromboembolic events with a pooled RR of 0.34. No statistically significant heterogeneity or publication bias was noted.

Eight studies were included in the meta-analysis for ICH recurrence. This included 5306 ICH patients of whom roughly 36% were restarted on anticoagulation. Recurrence of ICH was observed in 8.7% of patients on anticoagulation and 7.8% of patients not on anticoagulation, with no difference in risk between the two groups. There was significant heterogeneity in this sample, but no evidence of publication bias.

Limitations of the study include the possibility that patients selected for resumption of anticoagulation may have been lower risk, but this could not be verified due to incomplete data regarding hematoma size and other factors. The study also did not account for lobar hemorrhages, which have been associated with a higher risk of recurrence in some studies, presumably due to the likelihood of underlying cerebral amyloid angiopathy. However, this study is an important part of the growing body of evidence supporting resumption of anticoagulation therapy in selected patients.

Another study presented by Alessandro Biffi at ISC 2017 and Joji Kuramatsu at ESOC 2017 will likely provide further evidence, specifically in patients with NVAF. They presented results from a meta-analysis of the RETRACE, MGH and ERICH studies and suggest decreased mortality and improved functional outcomes at one year after anticoagulation resumption in both lobar and non-lobar ICH. Eleni Korompoki presented another pertinent meta-analysis showing statistically significant reduction in ischemic stroke risk and similar ICH recurrence in patients for whom anticoagulation was resumed.

Randomized prospective trials would be the logical next step to confirm these findings and allow for more detailed guidelines, and several are ongoing. It will be interesting to see data including newer oral anticoagulants, given the lower ICH rate in many of these drugs as compared to warfarin. The timing of anticoagulation resumption is another important component that has yet to be adequately clarified in the literature. The tides appear to be shifting in favor of restarting anticoagulation in more patients after ICH, but the publication of the first well-designed RCT will likely be the first step in effecting large-scale changes in practice.