Intracranial Hemorrhage After Stroke Thrombolysis: More Questions Than Answers

Approximately, 40,000 stroke patients are treated with intravenous alteplase annually in the United States.¹ Symptomatic intracranial hemorrhage (sICH) remains the most feared complication of stroke thrombolysis. Despite a 5% prevalence of sICH in modern registries,² there has been disproportionately limited data on standardized diagnosis and evidence-based management and treatment. While current American Heart Association/American Stroke Association (AHA/ASA) guidelines have included statements such as recommending use of platelet and cryoprecipitate transfusions in the event of sICH,³ the rationale for this is thin and is opinion more than evidence-based. The previous recommendations, including the timing of CT imaging, whether to reverse asymptomatic or minimally symptomatic hemorrhages, and which therapeutic approaches to utilize, are largely based on expert opinion, anecdotes, personal experiences, and best guesses.

Given the morbidity and mortality associated with this grave complication, it is surprising that attempts to standardize definitions and processes of care across stroke centers and to test potential treatment paradigms for sICH in clinical trials have not yet occurred. In this statement, Dr. Yaghi et al. provide a detailed review of the available literature and a roadmap to take important steps in elucidating mechanisms of sICH, optimizing definitions and management, and identifying potential targets for reversal agents to mitigate the significant harm associated with this complication.⁴ While this new scientific statement from the AHA/ASA relies equally on expert opinion of the writing group, having such a document could help both clinicians and researchers as we move from anecdotes to evidence-based medicine for the management of sICH.

Apples to apples: Since the range of clinical and radiographic presentations from ICH after alteplase treatment can vary from asymptomatic and petechial to catastrophic and massive leading to imminent death, there is an urgent need to classify the condition using both elements with precision and accuracy. An estimated 2,000 patients experience sICH in the US while up to 5 times more experience asymptomatic ICH. As with most descriptive classification schemes that are born out of a need to catalogue and bin cases based on visual rating of the severity of injury, the currently available schemas do not attempt to create phenotypes based on prognostic value. For example, of the 4 sICH types, only 1 meaningfully predicts poor outcome.^5,6 Thus, there is a major opportunity to redefine groups based on meaningful links to outcomes. Storing and analyzing clinically acquired imaging data in such patients⁷ may be another powerful way to analyze and understand the radiographic contributions to morbidity and mortality after sICH. Besides a graded categorical classification based on appearance on brain imaging, more granular details, such as volume and location of hemorrhage, midline shift, and degree of deterioration using quantitative measurement tools and linkage to clinical outcomes, may help provide opportunities to better predict which hemorrhages worsen outcomes and should be treated and which ones are safe to be left alone.

Needle in a haystack: Since sICH is a relatively rare occurrence, predicting which patients will develop it and when it will occur are major challenges. Risk scores for development of sICH are only performed modestly.⁸ Here again, current prediction tools based on crude regression approaches may be missing the boat. Novel statistical approaches such as random forests and machine learning approaches may help radically change the way we predict outcomes.⁹ These may lend themselves to further personalization of care such that those at high risk of its occurrence receive more proactive surveillance than those at low risk. With these reliable tools in hand, it is hoped that standards of timing of surveillance imaging based on predicted risk of sICH can be developed. For example, the optimal frequency of neurologic monitoring may need to be modified to match the temporal probability of sICH as suggested by the authors.

Comparative effective trials: Prior small studies have indicated that treatment after sICH is highly variable and not clearly associated with improved outcomes.^{10, 11} These single-center and multi-center registries have amassed only hundreds of patients, well below the requisite sample size to make meaningful inferences about treatment effects. Leveraging large scale registries may prove important here as well. Comparative effectiveness research studies assessing various approaches, including those summarized by Yaghi et al., may be feasible in large million patient registries such as the Get With The Guidelines Stroke or Safe Implementation of Thrombolysis in Stroke registries. Increasingly, for efficiency, cost, and sample size, medicine has looked upon registry-based randomized clinical trials and comparative effective studies to answer pressing questions in the most pragmatic way possible.^{12, 13} Given the incidence of sICH is low (and becoming lower with improved protocol adherence and reduced onset-to-treatment times)¹⁴ and the data on effect size from various treatments is unknown, it may be unwise to consider relatively small multi-center trials in the usual manner that has been the backbone of stroke trials for over 2 decades. Instead, harnessing active national and multi-national registries and implementing simple randomization schemes (e.g., aminocaproic acid vs. prothrombin complex concentrate) could produce important data and results to guide further guidelines in the management of sICH.

A call for an antidote? Lastly, the authors allude to a possible sea change in sICH management: an antidote for alteplase. As we have seen in recent years with novel anticoagulant drug development, the tremendous anxiety regarding bleeding spurred the design, production, and testing of anticoagulant antidotes such as idarucizumab for dabigatran.¹⁵ The publication of its efficacy in reversing the effects of dabigatran and the subsequent Food and Drug Administration approval may allay fears about uncontrollable bleeding and reduced enthusiasm for dabigatran prior to the availability of an antidote. Though major guidelines explicitly make clear that alteplase should not be withheld on the basis of perceived sICH risk, there remains reluctance in emergency rooms across the US to administer the drug precisely because of the concern about hemorrhage. Imagine what having an antidote would do to those fears, rational or otherwise. While aminocaproic acid, an antifibrinolytic drug, has the most plausible direct inhibitory effect in the thrombolytic pathway, it has significant risk of side effects, including increased risk of thrombosis and infarction in patients with subarachnoid hemorrhage where we have the most experience with its use.¹⁶ Though there have been prior animal studies of antidotes to alteplase,¹⁷ the development of new targeted antidotes to alteplase at sites of potential or actual bleeding would be a paradigm shift.

In summary, a dire scenario that we all have encountered at some point in our careers taking care of acute stroke patients treated with alteplase finally has answers to several vexing questions. It will be no easy task to move the field forward from anecdotes and dogma to more organized, and hopefully, standardized and evidence-based approaches that actually reverses or prevents sICH in the first place.