Atrial fibrillation and ischemic stroke
Atrial fibrillation (AF) is a major risk factor for ischemic stroke, associated with three to fivefold increased risk. Reference Wolf, Abbott and Kannel1 The rate of AF-associated ischemic stroke has tripled over the last few decades, and it is predicted to continue increasing in the future. Reference Yiin, Howard and Paul2 It is established that patients with AF who have suffered an ischemic stroke are at high risk for recurrence and require long-term anticoagulation. The risk of recurrent ischemic stroke ranges from 0.5% to 1.3% per day within the first 14 days after the index event based on retrospective observational studies. Reference Hart, Coull and Hart3 AF-related ischemic stroke is disabling in 60% and fatal in 20% of cases. Reference Gladstone, Bui and Fang4 A European epidemiological observational study demonstrated cardioembolic ischemic stroke is associated with the highest recurrence rate (22%; 95% CI 14–30) and lowest 2-year survival (55%; 95% CI 0.47–0.63) compared to other etiologies. Reference Kolominsky-Rabas, Weber, Gefeller, Neundoerfer and Heuschmann5 In patients with acute ischemic stroke and either known or newly diagnosed AF, the usual practice is to bridge with an antiplatelet agent until the patient is anticoagulated. Reference Gladstone, Lindsay and Douketis6,Reference Kleindorfer, Towfighi and Chaturvedi7 The timing of OAC initiation is often highly variable and opinion rather than evidence based. While earlier anticoagulation may reduce early recurrent ischemic stroke, it may also increase the risk of hemorrhagic transformation (HT), a serious early complication of ischemic stroke. Predictors of recurrent ischemic stroke in patients with AF include atrial thrombus, left atrial enlargement, left ventricular dysfunction, older age, larger infarct volume, and increasing CHA2DS2-VASc score, which is a clinical stroke risk score for patients with AF that includes heart failure, hypertension, age, diabetes mellitus, prior ischemic stroke, female sex, and other vascular diseases. Reference Paciaroni, Agnelli and Falocci8–Reference Hankey11 Most of these predictors are of limited use in informing the timing of anticoagulation, as they are also associated with an increased risk of HT. The competing rationales for early versus late anticoagulation make the optimal timing of anticoagulation after an ischemic stroke a persisting area of clinical equipoise.
Hemorrhagic Transformation
HT is a spectrum of ischemia-related brain hemorrhage, which varies from subtle heterogenous leakage of blood within the infarction to extensive hemorrhage within and beyond the infarction with and without mass effect. Reference Hornig, Dorndorf and Agnoli12 HT can lead to clinical deterioration from increasing edema, mass effect, intraventricular extension, and hydrocephalus and ultimately can result in death. Reference D'Amelio, Terruso and Famoso13
Pathophysiology
Understanding the mechanism of HT is a key element for predicting, preventing, treating, and prognosticating HT. The entire pathophysiology is still unclear. However, breakdown of the blood–brain barrier (BBB) is an essential component in the development of HT in ischemic stroke. Reference Warach and Latour14,Reference del Zoppo, von Kummer and Hamann15 BBB disruption results from a series of cellular, metabolic, and inflammatory events led by reduction in energy and failure in the Na+-K+ ATPase activity, causing injuries to cerebral endothelial cells and impairment in autoregulation of the cerebral blood vessels. Reference Rossi, Brady and Mohr16,Reference Khatri, McKinney, Swenson and Janardhan17 It has also been suggested that the BBB disruption is time-dependent, and that the mechanism of early HT (≤24 hours) may be different than the late HT (>24 hours). Reference Jickling, Liu and Stamova18 Reactive oxygen species, blood derived matrix metalloproteinases (MMP)-9, and the brain-derived MMP-2 may play critical roles in the mechanism of early HT. On the other hand, multiple factors could contribute to the late HT. This includes brain-derived MMP-9, MMP-3, inflammatory responses, vascular remodeling processes, and other proteases. Reference Jickling, Liu and Stamova18
Classification
Classification of HT is based on two components, the radiographic features of the hemorrhage and associated clinical changes. The term hemorrhagic infarction (HI) has emerged to describe subtle or confluent heterogenous blood within the infarcted tissue without mass effect. The term parenchymal hematoma (PH) describes the extensive homogamous hematoma within and beyond the infraction borders with mass effect. Reference Pessin, Del Zoppo and Estol19,Reference del Zoppo, Poeck and Pessin20 In 1999, Fiorelli et al proposed the European Cooperative Acute Stroke Study (ECASS) classification system, which includes two subtypes of HI (HI-1: small petechiae along the margins of the infarct, and HI-2: confluent petechiae within the infarcted area but no space-occupying effect) and two subtypes of PH (PH-1: hematoma in 30% or less of the infarcted area with some slight space-occupying effect, and PH-2: hematoma in more than 30% of the infarcted area with substantial space-occupying effect). Reference Fiorelli, Bastianello and von Kummer21 The ECASS criteria do not clearly differentiate between PH within the area of infarction and PH remote from the infarction, nor do they include other types of hemorrhages such as intraventricular hemorrhage, subarachnoid hemorrhage, and subdural hemorrhage. The Heidelberg bleeding classification is a classification tool to grade HT that expands beyond the ECASS system by including and categorizing these previously non-classified hemorrhages. Reference von Kummer, Broderick and Campbell22 The single standard definition of symptomatic HT in ischemic stroke has yet to emerge. As a result, variability in how HT is defined in stroke studies has impacted the reporting rate of symptomatic HT and makes comparing rates of HT between studies challenging. Reference von Kummer, Broderick and Campbell22–Reference Wahlgren, Ahmed and Eriksson27
Risk of HT after Acute Ischemic Stroke
Most of the older studies on HT in patients with acute ischemic stroke were of limited sample size and/or study design, or derived from studies on thrombolysis. Additionally, the reported rates of HT after ischemic stroke are variable in different studies. This variability could be related to the differences in study design, the included populations, the definitions of HT, type and frequency of imaging, timing and sequence of scan used, and method of assessing and defining clinical worsening. The reported incidence of HT is up to 70% in autopsy studies. Reference Fisher and Adams28–Reference Lodder31 On the other hand, the incidence of HT in computed tomography (CT) studies was variously reported over the last four decades, from few to 43% of consecutive patients. Reference Hornig, Dorndorf and Agnoli12,Reference Lodder, Krijne-Kubat and Broekman30–Reference Ott, Zamani, Kleefield and Funkenstein35 A more recent large prospective study that examined the risk of early HT in patient with acute ischemic stroke by using systematic brain CT at baseline and 5 ± 2 days after stroke onset was conducted. Reference Paciaroni, Agnelli and Corea36 Early HT in patients with acute ischemic stroke, including both AF and non-AF-related infarcts, was observed to be about 9% within 5–7 days from the index event and, of these HT events, 3.2% were PH. The early Recurrence and cerebral bleeding in patients with Acute ischemic stroke and atrial Fibrillation (RAF) study was a prospective observational study which specifically examined the risk of HT in patients with AF-related ischemic stroke. The rate of HT was found to be higher, 13%: 8.8% HI and 4.2% PH. Reference Paciaroni, Agnelli and Falocci8 A recent combined analysis from two large observational studies included 2183 patients and found that HT occurred in 11% on repeated imaging, with 3.1% of these being PH. Reference Paciaroni, Bandini and Agnelli37 The reported incidence of HT varies based on type of study, type and frequency of imaging, and sequence of scan used.
Clinical Predictors of HT
Several clinical factors have been described in association with development of HT. Recognizing these clinical variables might be helpful for clinicians to anticipate and stratify the risk of HT in individual patients. However, it is unknown if early anticoagulation in the presence of these clinical parameters further increases the risk of new and/or progressive HT. The size of infarction is independently associated with HT. Reference Tan, Wang, Liu, Zhang, Wu and Liu38 Larger ischemic infarct volume is associated with mass effect and vascular compression, both of which increase vascular permeability and therefore HT risk. Higher National Institutes of Health Stroke Scale (NIHSS) scores indicate severe stroke and larger infarct which increase the risk of HT. Reference Kidwell, Saver and Carneado39 Further, cardioembolic stroke and AF-related ischemic stroke tend to be severe and related to large vessel occlusion (LVO). Reference Lin, Wolf and Kelly-Hayes40 Compared to other subtypes, cardioembolic stroke associated with AF has been associated with the highest risk of HT. Reference Alvarez-Sabin, Maisterra, Santamarina and Kase41 Additionally, hyperglycemia may worsen the BBB disruption by inducing systemic stress and enhancing circulating factors that damage BBB which increases the risk of HT. Reference Kase, Furlan and Wechsler42–Reference Poppe, Majumdar and Jeerakathil44 The effect of high blood pressure (BP) on the risk of HT has been documented, especially in patients with larger infarcts who received thrombolytic therapy. Reference Butcher, Christensen and Parsons45 High BP is considered an modifier of the risk of HT through its interaction with other predictors. Reference Alvarez-Sabin, Maisterra, Santamarina and Kase41 Another factor found to be associated with HT is high body temperature in the first 24 hours after stroke onset. Reference Azzimondi, Bassein and Nonino46,Reference Leira, Sobrino and Blanco47 Moreover, low-density lipoprotein cholesterol level with and without statin has correlated with the risk of HT. Reference Bang, Saver and Liebeskind48,Reference D'Amelio, Terruso, Famoso, Ragonese, Aridon and Savettieri49 Finally, reperfusion therapies, including thrombolysis and endovascular therapy, are all associated with an increased rate of HT. Reference Alvarez-Sabin, Maisterra, Santamarina and Kase41,Reference Furlan, Higashida and Wechsler50–Reference Maros, Brekenfeld and Broocks54 However, a recent prospective observational study has shown that acute reperfusion therapies using thrombolysis and/or EVT prior to initiating anticoagulation did not influence the risk of recurrent ischemic stroke(s) and HT. Reference Giustozzi, Acciarresi and Agnelli55
Radiographic Predictors of HT
Non-contrast CT/CT Angiography (CTA)/CT Perfusion (CTP)
Early ischemic changes, including loss of gray–white matter differentiation, hypodensity (hypoattenuation) of brain parenchyma, presence of edema or mass effect, and low Alberta Stroke Program Early CT Score (ASPECTS) of ≤7 on baseline CT scan, appear to increase the risk of HT. Reference Larrue, von Kummer, Muller and Bluhmki56–Reference Hill, Rowley and Adler59 Another CT marker associated with HT is the hyperdense middle cerebral artery sign. Reference Zou, Churilov, He, Campbell, Davis and Yan60 Patients with hyperdense middle cerebral artery sign, LVO, poor collateral flow, lower clot burden scores, and severe hypoperfusion with large core in CTP tend to have severe stroke and early ischemic changes, and thus, higher risk of HT. Reference Lin, Zink, Tsiouris, John, Tekchandani and Sanelli61–Reference Adebayo and Culpan63
Magnetic Resonance Imaging
Magnetic resonance imaging (MRI) is more sensitive than CT for detection of HT after acute ischaemic stroke, Reference Arnould, Grandin, Peeters, Cosnard and Duprez64,Reference Renou, Sibon and Tourdias65 especially with adding the susceptibility weighted imaging, which is highly accurate in detecting blood products. A low volumetric apparent diffusion coefficient and large diffusion weight imaging on baseline MRI are associated with an increased risk of HT. Reference Butcher, Christensen and Parsons45,Reference Tong, Adami, Moseley and Marks66–Reference Tong, Adami, Moseley and Marks68 Additionally, multiple radiological markers in MRI have been associated with HT including poor fluid suppression on fluid-attenuated inversion recovery (FLAIR) from extravasation of contrast into CSF leading to hyperintensity of the CSF space, which has been labeled Hyperintense Acute Reperfusion Marker (HARM), Reference Mathews, Caldemeyer, Lowe, Greenspan, Weber and Ulmer69 sulcal hyperintensity on FLAIR, Reference Kim, Kim, Na, Roh, Ryoo and Kim70 parenchymal enhancement on post-contrast T1, Reference Vo, Santiago, Lin, Hsu, Lee and Lee71–Reference Kim, Na, Kim, Lee, Ryoo and Kim73 signs of disrupted BBB, Reference Bang, Buck and Saver74,Reference Lee, Saver and Alger75 and finally, cerebral microbleeds (CMBs). A recent meta-analysis has shown that patients with ≥5 CMBs are at higher risk of hemorrhage than those with fewer or no CMBs – patients with <5 CMBs (2.48%; 95% CI 1.2–6.2; p = 0.001). Reference Charidimou, Karayiannis and Song76 At this point, however, MRI screening is not routinely performed prior to initiation of anticoagulation and this finding has not changed clinical practice.
Serological Biomarkers Predictive of HT
Matrix Metalloproteinases
Several studies have demonstrated that increased expression of MMP-9 is associated with an increased risk of HT after reperfusion. Reference Sumii and Lo77–Reference Tang, Li, Li, Mu, Yang and Xie79 Similar studies have shown that elevated MMPs correlate with disrupted BBB and HT in experimental stroke models regardless of whether or not thrombolysis is received. This observation is likely secondary to the fact that MMP has the potential to degrade the basal lamina of the vascular endothelium. Reference Visse and Nagase80 However, the role of MMPs in the setting of anticoagulant-associated bleeding is unknown.
Leukocyte RNA
Inflammation and immune response after ischemic stroke may also influence the risk of HT by promoting peripheral leukocytes activation, adhesion, migration, and potentially BBB disruption. Reference Allport, Ding, Collins, Gerritsen and Luscinskas81,Reference Rosenberg82 Preliminary data suggest the risk of HT in patients with stroke can be stratified by RNA expressed in circulating leukocytes within 3 hours of stroke onset. A panel of six genes associated with subsequent HT has been identified. Reference Jickling, Liu and Stamova18,Reference Jickling, Ander and Stamova83 Of note, these data are driven from patient with early HT after thrombolysis, thus the role of leukocyte RNA in the setting of anticoagulation-related bleeding remains unclear.
Natural History and Outcomes of HT
HT is part of the spectrum of ischemia-related brain hemorrhage associated with a wide range of clinical significance. Most studies of the significance of HT come from thrombolysis trials. The relationship between PH-2 and clinical and functional outcomes has been established, but this is not the case for HI-1, HI-2, and PH-1. Reference Lees, Bluhmki and von Kummer84 A retrospective study indicated that any PH independently predicted mortality at day 30 and day 90. Reference D'Amelio, Terruso and Famoso13 . Conversely, a post hoc analysis of European Cooperative Acute Stroke Study I (ECASS I) indicated that only PH-2 was associated with an increased risk of neurological deterioration at 24 hours (OR 18.0; 95% CI 6.0–56.0) and 90 day mortality (OR 11.4; 95% CI 3.7–36.0). Reference Fiorelli, Bastianello and von Kummer21 Another post hoc analysis of the ECASS II study demonstrated that PH-2 was associated with approximately 50% mortality. Reference Berger, Fiorelli and Steiner85 This analysis and others revealed that HI-1 and HI-2 were not associated with unfavourable outcomes. Reference Fiorelli, Bastianello and von Kummer21,Reference Berger, Fiorelli and Steiner85,Reference Trouillas and von Kummer86 Similarly, another prospective study assessed the outcomes of early HT after ischemic stroke and found that only PH was independently associated with a higher risk of mortality and functional disability. Reference Paciaroni, Agnelli and Corea36 Conversely, a more recent prospective study revealed that both PH (OR 1.79; 95% CI 1.00–3.27; p = 0.05) and HI (OR 1.75; 95% CI 1.21–2.53; p = 0.003) were associated with death and disability. Reference Paciaroni, Bandini and Agnelli37 However, quantifying the impact of HT, especially HI, on functional outcome is challenging as the independent contribution of HT to clinical worsening remains uncertain, Reference Kablau, Kreisel and Sauer87 and evidence from high quality studies is lacking.
Early Anticoagulation after AF-related Ischemic Stroke
Older studies focused on heparins for early anticoagulation, perhaps due to its rapid onset of action compared with vitamin K antagonists (VKA) which can take several days to reach to therapeutic level. However, in the last decade, four direct oral anticoagulants (DOACs) have been demonstrated to have a lower long-term risk of intracranial hemorrhagic complications compared to older anticoagulants, and these are now the standard of care for stroke prevention in non-valvular AF. Reference Connolly, Ezekowitz and Yusuf88–Reference Giugliano, Ruff and Braunwald91 In a meta-analysis, DOACs were associated with a 52% significant reduction of intracranial hemorrhage compared to warfarin. Reference Ruff, Giugliano and Braunwald92 In the following sections, we discuss studies of anticoagulation initiation within 14 days of an ischemic stroke or transient ischemic attack.
Early Heparin Initiation after Acute Ischemic Stroke
Among patients who were not on any antithrombotic therapy in the International Stroke Trial, 4.9% developed recurrent ischemic events within 2 weeks of the index stroke onset. Reference Saxena, Lewis, Berge, Sandercock and Koudstaal93 A dose-dependent reduction in the recurrent ischemic strokes was noted in the unfractionated heparin (UFH) groups, but that benefit was offset by an increase in HT. A meta-analysis indicated that initiating low molecular weight heparin (LMWH) within 24–72 hours after ischemic stroke (regardless of the mechanism) for 7–30 days was associated with no reduction in recurrent ischemic stroke rates, but a trend to more symptomatic HT, and a significant increase in major extracranial bleeding. Reference Bath, Iddenden and Bath94 The Heparin in Acute Embolic Stroke Trial (HAEST) compared LMWH and aspirin initiation within 30 hours of stroke onset. Recurrent ischemic events within 14 days of the index stroke occurred 7.5% and 8.5% in patients that received LMWH and aspirin, respectively. Reference Berge, Abdelnoor, Nakstad and Sandset95 Starting LMWH within 30 hours of AF-related ischemic stroke was associated with an increased rate of symptomatic HT and extracranial hemorrhage. Reference Berge, Abdelnoor, Nakstad and Sandset95 A meta-analysis of seven trials of parenteral anticoagulants (UFH, LMWH, heparinoids) started within 48 hours of acute cardioembolic stroke indicated recurrent ischemic events within 7–14 days were similar to those in patients treated with aspirin or placebo (3.0% vs. 4.9%; OR 0.68; 95% CI 0.44–1.06; p = 0.09), but symptomatic HT was more frequent (2.5% vs 0.7%; OR 2.89; 95% CI 1.19–7.01; p = 0.02), (Figure 1). Reference Paciaroni, Agnelli, Micheli and Caso96
Absolute event rates of recurrent ischemic stroke and symptomatic HT associated with early LMWH after acute ischemic stroke. HT=hemorrhagic transformation; ICH=intracerebral hemorrhage; LMWH=low molecular weight heparin.
Early VKA Initiation after Acute Ischemic Stroke
Randomized controlled trials (RCTs) and observational studies of early anticoagulation after stroke, including patients initiated on VKA and DOACs, are summarized in Table 1. Data from RCTs evaluating the timing of VKA initiation after ischemic stroke in patients with AF are limited. The Stroke Acute Management with Urgent Risk-factor Assessment and Improvement (SAMURAI) study was a prospective observational study of initiating anticoagulant after stroke/TIA in patients with AF. Reference Toyoda, Arihiro and Todo97 A total of 650 patients were started on warfarin at a median of 3 days after stroke/TIA onset. There was no reported ICH prior to hospital discharge; however, systematic neuroimaging was not performed. The early Recurrence and cerebral bleeding in patients with Acute ischemic stroke and atrial Fibrillation (RAF) study was a prospective observational study. Reference Paciaroni, Agnelli and Falocci8 VKA alone was initiated in 37% of patients, with another 36% started on LMWH therapy before receiving VKA. Timing of anticoagulation was at the discretion of the treating physician, varying from 1 to 90 days after stroke. The optimal net clinical benefit of composite outcomes for anticoagulation initiation was 4–14 days after stroke onset. Conversely, data from 1644 patients in the Virtual International Stroke Trials Archive (VISTA) indicated that early VKA initiation, within 2–3 days after stroke onset, was associated with lower recurrent ischemic events without an increase in symptomatic ICH. Reference Abdul-Rahim, Fulton and Frank98 Finally, the Clinical Relevance Of Microbleeds In Stroke-2 (CROMIS-2) study assessed the effect of oral anticoagulant timing in patients with AF and stroke. Reference Wilson, Ambler and Banerjee99 Timing of anticoagulation was determined by the treating physicians and then retrospectively dichotomized into early (0–4 days) and late (≥5 days or never started) groups. Of 1355 patients prescribed an oral anticoagulant, 26% started early and 74% started late. Both groups had similar rates of recurrent ischemic events and ICH. Most patients (65%) were treated with warfarin, and 24% received bridging heparin therapy.
Studies of early initiation of oral anticoagulants after atrial fibrillation-related ischemic stroke
N=number; NIHSS=National Institutes of Health Stroke Scale; HT=hemorrhagic transformation; DOAC=direct oral anticoagulant; TIA=transient ischemic attack; CT=computed tomography; MRI=magnetic resonance imaging; ASA=aspirin; RR=relative risk; aHR=adjusted hazard ratio; cm=centimeter.
* Patients without atrial fibrillation.
Early DOACs Initiation after Acute Ischemic Stroke
RCTs and observational studies of early anticoagulation after stroke, including patients initiated on VKA and DOACs, are summarized in Table 1. In the pivotal phase III DOAC trials, more than 70,000 patients were randomized; however, patients were not eligible immediately after ischemic stroke (exclusion ranged from 7 to 30 days after onset). Reference Connolly, Ezekowitz and Yusuf88–Reference Giugliano, Ruff and Braunwald91 Moreover, among all the randomized patients, only 30% had previous stroke or TIA and the number of patients randomized early after stroke has never been published but is likely to be small. Reference Ruff, Giugliano and Braunwald92 DOACs are now the standard of care for stroke prevention in AF patients. Reference Gladstone, Lindsay and Douketis6,Reference Kleindorfer, Towfighi and Chaturvedi7 The optimal timing after ischemic stroke is a highly relevant clinical question, given the anticoagulant effect of these drugs begins within hours of administration. Reference Julia and James100
Randomized studies of early DOAC use post-stroke: There are only three published randomized trials assessing the safety of early DOAC administration after ischemic stroke. The Triple AXEL (Acute Stroke With Xarelto to Reduce Intracranial Hemorrhage, recurrent Embolic Stroke, and hospitaL stay) compared rivaroxaban (n = 101) to warfarin (n = 94) initiation within 5 days of cardioembolic stroke (median NIHSS score of 2 in both arms). Reference Hong, Kwon and Lee101 This study revealed similar recurrent ischemic stroke and HT rates in both groups. Incident radiographic HT detected on follow-up MRI was seen in 49.5% and 54.5% of patients receiving rivaroxaban and warfarin, respectively. The higher frequency of HT compared to previous studies is likely related to the higher sensitivity of MRI for petechial bleeding. The DATAS II trial (Dabigatran in Acute Transient Ischemic Attack and minor Stroke) randomized 305 patients without AF to dabigatran or aspirin within 72 hours of mild ischemic symptom onset. Reference Butcher, Ng and Sheridan102 There were no symptomatic HT events in either group, but asymptomatic HT detected with MRI was reported in 7.8% of the dabigatran group and 3.5% of the aspirin group (relative risk 2.22, [0.79, 6.21]). More recently, the Apixaban for Early Prevention of Recurrent Embolic Stroke and Hemorrhagic Transformation (AREST) trial randomized patients to apixaban or warfarin, but the timing of initiation was according to the infarct size. Reference Labovitz, Rose and Fradley103 The study was stopped early in 2019, after 91 patients had been randomized as DOACs have become the standard of care for most patients with AF-related stroke. One case of symptomatic HT occurred in the warfarin group. Incident radiographic HT was detected in 12.2% and 10.6 % of the apixaban and warfarin groups, respectively. Recurrent ischemic events were more common in both the apixaban (14.6%) and warfarin (19.2%) groups.
Prospective, non-randomized studies of early DOAC use post-stroke: The SAMURAI study included a total of 466 patients who were started on DOAC at a median of 4 days after stroke/TIA onset. Reference Toyoda, Arihiro and Todo97 In those patients, the recurrent stroke/systemic embolism rate was 2.84% and the rate of major bleeding was 1.1% at day 90. Reference Arihiro, Todo and Koga104 The Early Recurrence and Major Bleeding in Patients With Acute Ischemic Stroke and Atrial Fibrillation Treated With Non-Vitamin-K Oral Anticoagulants (RAF-NOACs) study examined 1127 patients, 80% of whom received DOAC within 15 days of stroke. Reference Paciaroni, Agnelli and Falocci105 Recurrent ischemic events within 90 days occurred in 2.8% of patients, which was more common than the symptomatic cerebral bleeding rate of 1.6%. However, the lack of serial imaging may have led to an under-estimation of the rate of asymptomatic HT. The ‘Novel Oral Anticoagulants in Ischemic Stroke Patients’ (NOACISP) registry did not report an increased risk of symptomatic HT or recurrent stroke in 100 patients initiated on DOAC within 7 days of stroke onset. Reference Seiffge, Traenka and Polymeris106 In another study, the risk of recurrent stroke and hemorrhagic complication was found to be similar in patients with anterior and posterior circulation infarcts. Reference Paciaroni, Agnelli and Giustozzi107 Symptomatic HT rate was higher in this cohort (4.2% and 3.7% in the anterior and posterior circulation, respectively), and potentially explained by the fact that warfarin and/or bridging heparin prior to DOAC administration was used.
Prospective studies including systematic brain imaging following OAC initiation post-stroke have generally been smaller. By using serial MRI pre- and post-treatment, a prospective assessment of the safety of rivaroxaban initiation at a median of 3 days after AF-related ischemic stroke demonstrated that asymptomatic petechial HT was common at baseline (25/60) and remained clinically silent despite immediate treatment with rivaroxaban. Reference Gioia, Kate and Sivakumar108 A recent prospective, multicenter registry of dabigatran initiation within 14 days of acute minor ischemic stroke/TIA (NIHSS ≤ 3) onset in patients with AF has also been published. Reference Alrohimi, Ng and Dowlatshahi109 A total of 101 patients (median NIHSS score was 1) were enrolled. The median time from ischemic symptom onset to dabigatran initiation was 2 days and this was not associated with any symptomatic HT, but asymptomatic HT evident on systematically acquired follow-up CT was seen in 6% of patients. Another recent prospective study of apixaban initiation within 14 days of TIA/acute ischemic stroke regardless the size and severity has been completed. Reference Alrohimi, Buck, Jickling, Shuaib, Thirunavukkarasu and Butcher110 A total of 100 patients (median NIHSS score was 4) were included. The median time from ischemic symptom onset to apixaban initiation 2 two days and this was not associated with any symptomatic HT, but incident radiographic HT evident on systematically acquired follow-up CT was seen in 3% of patients. Recurrent ischemic events occurred in 13 patients, 4 of which were associated with severe disability and 4 with mortality. One prospective study included 120 patients with AF-related stroke who received either DOAC or heparin/warfarin, 80% of which treated within 14 days. Reference Al Bakr, Al Omar and Nada111 HT occurred in 27%, including 7% whom developed a PH, but the majority of these were related to warfarin with/without heparin. A more recent study of 147 patients treated with DOAC within 7 days, HT was seen in 8 patients (asymptomatic in 7, and symptomatic in 1). Reference Cappellari, Carletti, Danese and Bovi112 Another small MRI-based prospective study of 41 patients was completed. Reference Shibazaki, Kimura, Aoki, Saji and Sakai113 Median NIHSS and time from onset to DOAC initiation were 3 and 2 days, respectively. Incident asymptomatic HT was observed in 11 patients, all of which were asymptomatic.
Guideline Statements and Current Practice
Current guidelines are inconsistent and provide limited advice with respect to the timing of DOAC initiation after AF-related ischemic stroke. Reference Gladstone, Lindsay and Douketis6,Reference Kleindorfer, Towfighi and Chaturvedi7,Reference Heidbuchel, Verhamme and Alings114–Reference Hersi, Alhebaishi and Hamoui118 The European Heart Rhythm Association of the European Society of Cardiology (EHRA-ESC) endorsed the “1–3–6–12 days rule,” which recommends timing of anticoagulation based on clinical severity and infarct size, but neither are well defined. Reference Heidbuchel, Verhamme and Alings114 The 2021 guidelines of the American Heart Association/American Stroke Association (AHA/ASA) recommend starting oral anticoagulation immediately after TIA and 2–14 days after the index event for ischemic stroke. Reference Kleindorfer, Towfighi and Chaturvedi7 In the same guideline statement, a delay in initiation beyond 14 days is recommended for patients considered to be at high risk of HT. Bridging therapy with heparins is considered a class III recommendation against (HARM) in the ESC guidelines. Reference Hindricks, Potpara and Dagres115 An online survey indicated that 95% of physicians in the UK were uncertain when to start oral anticoagulation after cardioembolic stroke. Reference Munn, Abdul-Rahim, Fischer, Werring, Robinson and Dawson119
Discussion
Standardized documentation of HT rates on follow-up imaging is missing from most studies of early anticoagulation. Studies including serial imaging data at baseline and following DOAC initiation have demonstrated the rate of incident asymptomatic HT to be as low as 3–6% in CT-based studies and up to 13% in MRI-based studies. Reference Butcher, Ng and Sheridan102,Reference Gioia, Kate and Sivakumar108–Reference Alrohimi, Buck, Jickling, Shuaib, Thirunavukkarasu and Butcher110,Reference Kate, Gioia and Buck120 Even in cases with baseline HT prior to anticoagulation, DOAC initiation did not precipitate symptomatic HT. Conversely, early recurrent ischemic events were more common, all of which were symptomatic. Most of the recurrent ischemic events occurred within the first 14 days, suggesting the risk of ischemic events is highest early after the index event. In contrast to HT, which was detected incidentally on follow-up imaging and was not observed to independently influence the functional outcomes, recurrent ischemic events were always clinically evident and associated with poor functional outcomes.
The results from most observational studies suggest that symptomatic HT rates will likely be low in randomized trials of DOAC initiation post AF-related stroke. The low baseline event rate makes it difficult to predict the effect of DOAC initiation and also complicates studies aimed at risk stratification of patients with AF-related stroke. Conversely, recurrent ischemic stroke will potentially be the clinical outcome of interest in these RCTs. However, HT remains a critical complication to consider, as even a slight increase in frequency may outweigh any benefits of early anticoagulation. Although incident radiographic HT is asymptomatic and only detected incidentally on follow-up imaging, it likely shares a common mechanism and pathophysiological pathway with more severe symptomatic HT. Reference Warach and Latour14,Reference del Zoppo, von Kummer and Hamann15 Clinicians tend to delay DOAC initiation in patients with larger infarcts, irrespective of the presence or absence of HT, Reference Paciaroni, Agnelli and Falocci8,Reference Alrohimi, Ng and Dowlatshahi109-Reference Al Bakr, Al Omar and Nada111 consistent with expert recommendations. Reference Diener, Selim, Molina and Greenberg121 This practice of delaying DOAC administration due to concerns that patients with large infarcts are more prone to symptomatic HT appears reasonable at present. The absolute risk of HT, however, and the optimal timing in individual patients remains unknown. A follow-up neuroimaging scan before initiating DOAC is often used in clinical practice to assess infarct progression or hematoma expansion but whether these factors should guide decision, and how, remains unknown. Serial and systematic collection of HT rates constitute objective criteria which may be important surrogate markers even in larger trials, where the expected absolute number of symptomatic HT cases is likely to remain low.
Future Directions
The question of optimal timing to start DOAC will remain unanswered until randomized trials of early versus delayed DOAC are completed (Table 2). Reference Alrohimi, Jickling and Jeerakathil122–Reference Seiffge, Werring and Paciaroni125 Ongoing trials include LASER (Lixiana Acute Stroke Evaluation Registry, NCT03494530), TIMING (Timing of Oral Anticoagulant Therapy in Acute Ischemic Stroke With Atrial Fibrillation: a Prospective Multicenter Registry-based Non-inferiority Randomized Controlled Clinical Trial, NCT02961348), ELAN (Early Versus Late Initiation of Direct Oral Anticoagulants in Post-ischaemic Stroke Patients With Atrial fibrillation, NCT03148457), START (Optimal Delay Time to Initiate Anticoagulation After Ischemic Stroke in Atrial Fibrillation, NCT03021928) trials, and OPTIMAS (OPtimal TIming of Anticoagulation After Acute Ischemic Stroke, NCT03759938).
Ongoing randomized trials assessing the safety and efficacy of early versus delayed oral anticoagulant initiation after AF-related ischemic stroke
NCT=The National Clinical Trial number; h=hours; HT=hemorrhagic transformation; DOAC=direct oral anticoagulant; NT-proBNP=N-terminal pro b-type natriuretic peptide; GDF-15=growth/differentiation factor-15.
Patients in the above-named trials are randomized to a DOAC, initiated as early as 48 hours and up to 5 days after onset, or delayed to 6–14 days. The primary endpoint in these trials is the composite of recurrent ischemic stroke and symptomatic HT. While using a composite endpoint will increase the number of events and statistical power, the contrasting outcomes could potentially dilute the net treatment effect (i.e., early initiation might potentially reduce recurrent stroke but increase symptomatic HT, and vice versa) and lead to a neutral primary outcome result.
The observed current practice of delaying DOAC initiation in patients with larger infarcts is relevant to the ongoing RCTs, where randomization is either stratified by infarct volume/size Reference King, Lawrence, Milling and Warach123 ELAN, NCT03148457 and/or includes arms with relatively broad initiation time windows of several days, extending up to 14 days after symptom onset. Reference Alrohimi, Jickling and Jeerakathil122,Reference Asberg, Hijazi, Norrving, Terent, Ohagen and Oldgren124 OPTIMAS, NCT03759938 This form of stratification by infarct volume or by having broad initiation time windows where the decision to time the treatment at the treating physician’s discretion, even in the context of randomization, may play a role in introducing a bias. It may also limit the understanding of whether the timing of DOAC initiation should be adjusted based on infarct size and/or clinical severity of the index event to reduce the risk of HT. While these trials are expected to advance the field, the broad randomization windows, the contrast effect of used composite outcomes, and lack of systematic imaging post-treatment in most of the trials may leave some clinical equipoise and also facilitate continuation of current practice patterns based on expert opinion.
Conclusions
Currently available data suggest that early DOAC initiation after AF-related stroke is safe and perhaps even efficacious. However, this needs to be confirmed by ongoing randomized trials of early versus delayed DOAC initiation.
Acknowledgments
AA thanks King Saud University and the Saudi Arabian Ministry of Education for Residency and Fellowship funding. AA thanks The University of Alberta Hospital Foundation and the Neuroscience and Mental Health Institute for the Neurology Fellowship Award.
Statement of Authorship
AA conceived of the idea, acquired the data, and drafted the initial manuscript. GJ and BB made critical revisions of the manuscript. KB conceived of the idea and made critical revisions of the manuscript.
Disclosures
No conflicts of interest, financial or otherwise related to this manuscript, are declared by AA. KB received grant from Alberta Technologies Futures and honoraria from Boehringer-Ingelheim and Pfizer. GJ received grants from CIHR, Heart and Stroke Foundation of Canada, NIH, and University Hospital Foundation. BB is the ethic committee chair (Unpaid) for the Canadian Stroke Consortium.
