Abstract
To assess the association of anticoagulation in patients with cardioembolic acute ischemic stroke (CES) who develop hemorrhagic transformation (HT) and its impact on neuroimaging and functional outcomes. This retrospective study enrolled patients presenting with CES within 48 h at a tertiary stroke center between January 2011 and August 2023. Patients who developed HT during hospitalization and underwent follow-up imaging within 1 week were included, focusing on those with hemorrhagic infarction or parenchymal hematoma type 1. Primary outcomes were HT exacerbation on follow-up imaging and 3-month modified Rankin Scale (mRS) distribution shift, comparing anticoagulation therapy (AC), antiplatelet therapy (APT), and drug discontinuation (DDDD). The safety outcome was the occurrence of symptomatic intracerebral hemorrhage (sICH), which was defined as a hemorrhage concomitant with neurological deterioration. Among 763 patients with HT (age 74.6 ± 8.9 years, 48.1% male), AC was associated with a higher incidence of HT exacerbation compared to APT (adjusted OR 0.48, 95% CI 0.29–0.80, p-value = 0.005). AC associated with a better 3-month mRS compared to both APT (adjusted OR 0.63, 95% CI 0.43–0.92, p-value = 0.017) and DD (adjusted OR 0.38, 95% CI 0.26–0.55, p-value < 0.001). sICH occurred in 5%, with rates of 1.5%, 2.1%, and 11.7% in the AC, APT, and DD groups, respectively (adjusted OR for DD vs. AC: 3.93, 95% CI 1.18–13.16, p-value = 0.026). Anticoagulation in CES patients with HT was associated with a better functional outcome and radiological exacerbation, without a significant increase in sICH risk. These findings suggest that the presence of HT should not necessarily preclude the use of anticoagulation therapy in this patient population. However, our study should be interpreted as hypothesis-generating, and confirmation from future prospective studies is warranted.
Keywords: Hemorrhagic transformation, Anticoagulation, Acute ischemic stroke, Cardioembolism
Introduction
Hemorrhagic transformation (HT) is recognized as a natural phenomenon in the course of acute ischemic stroke and is frequently used as an indicator of clinical outcomes in stroke patients [1]. Certain subtypes of HT, such as hemorrhagic infarction and parenchymal hematoma type 1, typically do not significantly affect neurological outcomes or mortality[2–4]. However, parenchymal hematoma type 2 is typically strongly associated with neurological deterioration and increased mortality rates [2–4].
Despite advancements in stroke management, HT remains a potentially prognostically detrimental complication, with reported incidence ranging from 13% to 46%[5, 6]. HT occurs more frequently and severely in acute cardioembolic stroke (CES) than in strokes of other etiologies. This is likely due to the distinct characteristics of CES. First, it often involves the sudden occlusion of a major cerebral artery by a large clot, which leads to a larger volume of acute ischemia. The subsequent restoration of blood flow into the ischemic territory—either spontaneously or following recanalization therapy—can cause more extensive hemorrhage due to the compromised integrity of already-damaged and weakened blood vessels. Additionally, CES is frequently associated with more severe underlying hypoperfusion, which contributes to greater infarct growth and, consequently, a higher risk of severe HT[6, 7].
Clinical guidelines on anticoagulation timing in CES aim to balance the risk of recurrent embolism against bleeding complications [8, 9]. These recommendations also emphasize tailoring the start of anticoagulation based on the likelihood of hemorrhagic conversion. Current guidelines acknowledge the presence of HT as an important factor in timing decisions but provide no specific recommendations, generally suggesting a 4–14 day delay based mainly on observational data [9]. In real-world practice, the presence of HT often leads clinicians to delay anticoagulation initiation; one study reported an average delay of 12 days, which was not associated with a higher risk of recurrent stroke [10]. In contrast to these practices, recent trials suggest early anticoagulation may have potential benefits over delayed therapy in atrial fibrillation patients with stroke [11–14]. However, most of these studies do not specifically address whether anticoagulation remains safe when HT is already present. Consequently, many clinicians are reluctant to administer anticoagulation once HT has been detected, given the lack of robust evidence evaluating its safety in an ongoing hemorrhagic process [15–17]. The presence of HT, therefore, is also a crucial factor to consider when determining the most appropriate timing to initiate medication. The lack of comprehensive data is compounded by the fact that existing studies are primarily based on small patient cohorts or subgroup analyses, making it difficult to draw definitive conclusions or establish clear clinical guidelines [18, 19].
To address this critical knowledge gap, our study aims to investigate the impact of anticoagulation on HT exacerbation, as visualized on follow-up imaging, and on clinical outcomes in patients with CES. We have specifically excluded patients with parenchymal hematoma type 2 from our analysis, given its well-established association with poor neurological outcomes. By focusing on patients with less severe forms of HT, we aim to challenge the prevailing guideline recommendation to delay anticoagulation in this population, while providing more nuanced insights into the safety and optimal timing of therapy.
Subjects/Materials and Methods
Subjects
This retrospective study included patients with acute ischemic stroke who presented within 48 h of symptom onset at a tertiary stroke center between January 2011 and August 2023 (n = 12,658). Among them, CES was classified per the Trial of Org 10,172 in Acute Stroke Treatment criteria [20], based on the attending physician’s clinical and radiological judgement (n = 2,820).Only patients who developed HT based on imaging performed during hospitalization were included (n = 1,050). Those for whom changes in HT could not be assessed due to a lack of follow-up imaging after the initial HT were excluded (n = 896). Additionally, patients with an initial diffusion-weighted magnetic resonance imaging (DWI) Alberta Stroke Program Early CT Score (ASPECTS) score of 0–2, indicating high stroke severity (n = 800), and those with an initial HT type classified as parenchymal hematoma type 2 or higher were excluded, resulting in a final analysis of 763 patients.
Definition of HT and HT Exacerbation
HT was classified radiologically according to the European Cooperative Acute Stroke Study criteria [3]. Based on this classification, HT is typically categorized radiologically into two main subtypes: hemorrhagic infarction and parenchymal hematoma. Hemorrhagic infarction type 1 is defined as small petechial hemorrhage within the infarcted area without mass effect, while hemorrhagic infarction type 2 is characterized by confluent petechial hemorrhage within the infarcted area, also without mass effect. Parenchymal hematoma type 1 is defined as a hematoma involving less than 30% of the infarcted area, accompanied by mild mass effect. Parenchymal hematoma type 2, on the other hand, involves hematoma in more than 30% of the infarcted area with definite mass effect. In our hospital, where DWI is extensively used, HT was detected using DWI and susceptibility-weighted imaging. Imaging assessments were conducted by a committee of three neurologists specializing in stroke care. Discrepancies in interpretation were resolved through consensus during regular meetings. Exacerbation of HT was defined as an increase in HT severity by one or more stages (e.g., from hemorrhagic infarction type 2 to parenchymal hematoma type 1) or as an expansion of the hemorrhagic lesion within the same HT type, which was defined as any increase in the extent of hemorrhage or the appearance of a new HT area on follow-up imaging. Symptomatic intracerebral hemorrhage (sICH) was defined as either parenchymal hematoma type 2 or any HT associated with neurological aggravation on follow-up imaging.
Imaging Protocol and Analysis
The imaging protocol for acute stroke patients at our center is as follows: DWI and magnetic resonance angiography are conducted within a few hours of arrival at the emergency department unless contraindicated. When intravenous thrombolysis or endovascular treatment is performed, DWI is typically conducted 1 day after admission to assess the presence of lesions. A follow-up DWI is conducted on day 3–5, with additional tests performed as deemed necessary by the attending clinician. Patients included in this study were those whose HT was confirmed through imaging performed within 1 week of hospitalization after the index stroke. Only those patients who underwent follow-up imaging to assess changes in HT were analyzed to evaluate the impact of treatment. Accordingly, we aimed to evaluate the exacerbation of HT. When measuring the ASPECTS score, the small focal embolic lesion was excluded from scoring. Only confluent lesions within the scoring region were counted and subsequently analyzed.
Data Collection
Demographic, clinical, imaging, and laboratory data were analyzed retrospectively. Imaging tests were conducted at least twice during hospitalization, with the HT type confirmed at each time point. The ASPECTS score was measured using DWI performed at the time of initial hospitalization, with separate scores calculated for the anterior and posterior circulation based on previous studies [21, 22]. Patients were classified according to the treatment method used following the initial confirmation of HT: (1) anticoagulation therapy group (AC), (2) antiplatelet therapy group (APT), and (3) no medication and drug discontinuation therapy group (DD). For example, patients were assigned to the AC group if anticoagulation was continued or initiated after HT detection, and to the DD group if medication was discontinued or not resumed. The AC included patients treated with direct oral anticoagulants or vitamin K antagonists, without distinctions regarding dosage. The APT consisted of patients who received aspirin, clopidogrel or cilostazol, regardless of the type, dose, or combination of drugs. The DD included patients who discontinued their anticoagulant or antiplatelet therapy or did not resume these medications after HT if they were not already on them. In addition to examining the presence or absence of medication at the time of HT occurrence, antiplatelet and anticoagulation use at discharge was also analyzed.
Outcome
The primary outcomes were the occurrence of HT exacerbation on follow-up imaging and the shift in modified Rankin Scale (mRS) distribution at 3 months. Safety outcomes included the occurrence of sICH on follow-up imaging. Secondary outcomes included the proportion of patients with a good mRS (0–2) at 3 months, and the evaluation of composite events at 3 months, such as recurrent stroke (either hemorrhagic or ischemic), myocardial infarction, and all-cause mortality, as well as stroke recurrence.
Statistical Analysis
Baseline characteristics and outcomes were compared according to treatment methods using the chi-square test, ANOVA, Cochran-Mantel-Haenszel shift test, or Kruskal-Wallis test, depending on the type of variable. The presence of HT exacerbation, sICH, good functional outcomes at 3 months, and vascular outcomes by treatment were analyzed using Kaplan-Meier methods, chi-square tests, and log-rank tests. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated using a logistic regression model to evaluate the impact on HT exacerbation, 3-month mRS, and safety outcomes. Hazard ratios (HRs) and 95% confidence intervals (95% CIs) for 3-month vascular outcomes and stroke recurrence were analyzed using the Cox proportional hazards model for each treatment method. Predefined variables considered clinically relevant were adjusted for: age, sex, National Institutes of Health Stroke Scale (NIHSS) score, previous mRS, ASPECTS, HT type, history of stroke, coronary heart disease, hypertension, diabetes, hyperlipidemia, transient cerebral ischemia, peripheral arterial disease, prior diabetes treatment, prior anticoagulation treatment, acute thrombolytic treatment, white blood cell count, total cholesterol, low-density lipoprotein-C, blood urea nitrogen, creatinine, and fasting glucose. We adjusted for clinical and laboratory factors results that either showed differences among the treatment groups in the general characteristics or were deemed to be associated with stroke prognosis. Ordinal shifts toward better outcomes were analyzed using common ORs with 95% CIs. Two-sided p-values < 0.05 were considered significant. Statistical analyses were conducted using the R software (version 3.6.0, R Foundation, Vienna, Austria).
Results
General Characteristics
The average follow-up period for each patient was 86.4 ± 18.5 days. Table 1 summarizes the baseline characteristics of 763 HT patients (mean age 74.6 ± 8.9 years, 48.1% male). After developing HT, 324 patients (42.4%) were in the AC, 191 patients (25.0%) were in the APT, and 248 patients (32.5%) were in the DD. Regarding HT type, 347 patients (45.5%) were classified as hemorrhagic infarction type 1, 322 patients (42.2%) as hemorrhagic infarction type 2, and 94 patients (12.3%) as parenchymal hematoma type 1. The AC group had a higher proportion of hemorrhagic infarction type 1 cases (54.4%) and a lower proportion of parenchymal hematoma type 1 cases (10.0%). In contrast, the DD group had a lower proportion of hemorrhagic infarction type 1 cases (17.6%) and a higher proportion of parenchymal hematoma type 1 cases (73.4%), showing a significant difference (p < 0.001). The median initial NIHSS score was 11 (IQR 6–15), with lower scores in the AC (9, IQR 3–13) than in the DD (12, IQR 8–15) (p < 0.001). Anterior circulation infarcts accounted for 86.2% of cases. Among the patients who were not on any medication at the time of HT, 29.3% were discharged on anticoagulation, whereas 82.4% of those who were initially on anticoagulation continued their treatment (p < 0.001).
Table 1.
General characteristics of study population according to the treatment after HT
| All (n = 763) | AC (n = 324) | APT (n = 191) | DD (n = 248) | p-value | |
|---|---|---|---|---|---|
| Age | 74.6 ± 8.9 | 74.5 ± 9.0 | 74.5 ± 8.7 | 74.9 ± 9.0 | 0.824 |
| Male | 367 (48.1%) | 149 (46.0%) | 85 (44.5%) | 133 (53.6%) | 0.100 |
| HT Type | < 0.001 | ||||
| HI-1 | 347 (45.5%) | 189 (58.3%) | 97 (50.8%) | 61 (24.6%) | |
| HI-2 | 322 (42.2%) | 126 (38.9%) | 78 (40.8%) | 118 (47.6%) | |
| PH-1 | 94 (12.3%) | 9 (2.8%) | 16 (8.4%) | 69 (27.8%) | |
| initial NIHSS score | 11 (6–15) | 9 (3–13) | 12 (7–15) | 12 (8–15) | < 0.001 |
| previous mRS score | 0 (0–1) | 0 (0–1) | 0 (0–1) | 0 (0–0) | 0.405 |
| BMI | 23.3 ± 3.1 | 23.2 ± 3.1 | 23.0 ± 2.9 | 23.7 ± 3.3 | 0.056 |
| ASPECT DWI score | 7 (5–8) | 7 (6–8) | 7 (5–8) | 7 (5–8) | < 0.001 |
| Anterior (86.2%) | 7 (5–8) | 7 (6–8) | 7 (5–8) | 7 (5–8) | < 0.001 |
| Posterior (13.8%) | 8 (7–8) | 8 (7–8) | 8 (7–8) | 8 (7–9) | 0.524 |
| WBC | 8.6 ± 3.1 | 8.1 ± 3.0 | 9.2 ± 3.3 | 9.0 ± 3.1 | < 0.001 |
| Total cholesterol | 161.0 ± 35.9 | 158.3 ± 35.5 | 167.3 ± 36.3 | 159.8 ± 35.7 | 0.017 |
| BUN | 18.4 ± 8.0 | 17.6 ± 7.8 | 18.8 ± 8.4 | 19.1 ± 7.9 | 0.069 |
| Creatinine | 0.86 ± 0.40 | 0.86 ± 0.34 | 0.81 ± 0.40 | 0.92 ± 0.47 | 0.015 |
| Hemoglobin | 13.4 ± 1.8 | 13.3 ± 1.8 | 13.3 ± 1.8 | 13.5 ± 1.8 | 0.483 |
| Fasting glucose | 135.3 ± 47.9 | 131.3 ± 50.3 | 132.3 ± 45.9 | 142.8 ± 45.5 | 0.010 |
| Platelet count | 204.5 ± 59.2 | 204.7 ± 59.5 | 210.3 ± 60.5 | 199.9 ± 57.5 | 0.189 |
| LDL-C | 97.5 ± 28.8 | 96.6 ± 27.7 | 100.7 ± 30.6 | 96.1 ± 28.9 | 0.196 |
| history of TIA | 12 (1.6%) | 8 (2.5%) | 0 (0.0%) | 4 (1.6%) | 0.061 |
| history of stroke | 186 (24.4%) | 93 (28.7%) | 51 (26.7%) | 42 (16.9%) | 0.004 |
| history of PAD | 11 (1.4%) | 1 (0.3%) | 3 (1.6%) | 7 (2.8%) | 0.037 |
| history of CHD | 63 (8.3%) | 24 (7.4%) | 19 (9.9%) | 20 (8.1%) | 0.594 |
| history of HTN | 486 (63.7%) | 200 (61.7%) | 132 (69.1%) | 154 (62.1%) | 0.198 |
| history of DM | 203 (26.6%) | 85 (26.2%) | 43 (22.5%) | 75 (30.2%) | 0.188 |
| history of DL | 93 (12.2%) | 39 (12.0%) | 26 (13.6%) | 28 (11.3%) | 0.757 |
| Smoking | 0.774 | ||||
| Current smoking | 72 (9.4%) | 34 (10.5%) | 18 (9.4%) | 20 (8.1%) | |
| Ex-smoking | 65 (8.5%) | 27 (8.3%) | 19 (9.9%) | 19 (7.7%) | |
| Never | 626 (82.0%) | 263 (81.2%) | 154 (80.6%) | 209 (84.3%) | |
| history of AC | 165 (21.6%) | 97 (29.9%) | 29 (15.2%) | 39 (15.7%) | < 0.001 |
| history of statin | 119 (15.6%) | 56 (17.3%) | 31 (16.2%) | 32 (12.9%) | 0.345 |
| history of DM treat | 154 (20.2%) | 64 (19.8%) | 30 (15.7%) | 60 (24.2%) | 0.087 |
| history of HTN treat | 396 (51.9%) | 157 (48.5%) | 109 (57.1%) | 130 (52.4%) | 0.165 |
| Acute treatment | 0.009 | ||||
| IVT only | 157 (20.6%) | 45 (13.9%) | 51 (26.7%) | 61 (24.6%) | |
| IAT only | 115 (15.1%) | 56 (17.3%) | 28 (14.7%) | 31 (12.5%) | |
| IVT + IAT | 93 (12.2%) | 40 (12.3%) | 22 (11.5%) | 31 (12.5%) | |
| None | 398 (52.2%) | 183 (56.5%) | 90 (47.1%) | 125 (50.4%) | |
| Discharge APT | 259 (33.9%) | 72 (22.2%) | 123 (64.4%) | 64 (25.8%) | < 0.001 |
| Discharge AC | 396 (51.9%) | 267 (82.4%) | 56 (29.3%) | 73 (29.4%) | < 0.001 |
| mean SBP | 135.6 ± 14.4 | 134.6 ± 14.3 | 137.3 ± 14.2 | 135.5 ± 14.5 | 0.117 |
† P-value by Chi-square test, Fisher’s exact test, Cochran-Mantel-Haenszel shift test, ANOVA and Kruskal-Wallis Test
AC, anticoagulation; APT, antiplatelet; DD, no or quit medication; HT, hemorrhagic transformation; HI, hemorrhagic infarction; PH, parenchymal hematoma; NIHSS, National institute of health stroke scale; BMI, body mass index; WBC, white blood cell; BUN, blood urea nitrogen; LDL-C, low density lipoprotein; TIA, transient cerebral ischemia; PAD, peripheral arterial disease; CHD, coronary heart disease; HTN, hypertension; DM, diabetes mellitus; DL, dyslipidemia; IVT, intravenous thrombolysis; IAT, intra-arterial thrombectomy; SBP, systolic blood pressure
Primary Outcome: HT Exacerbation and mRS Shift
HT exacerbation was observed in 28.7% of patients based on imaging findings, with rates of 30.9% in the AC, 21.5% in the APT, and 31.5% in the DD (Table 2). n the unadjusted model, HT exacerbation was more common in the AC than in the APT but did not differ significantly from the DD. After adjusting for predefined variables, the risk of HT exacerbation was lower in the APT than in the AC (adjusted OR 0.48, 95% CI 0.29–0.80, p = 0.005). The DD group also showed a lower risk of HT exacerbation, this finding was not statistically significant (adjusted OR 0.86, 95% CI 0.53–1.37, p = 0.521). Regarding the shift toward a better outcome in mRS scores at 3 months, both the APT (adjusted OR 0.63, 95% CI 0.43–0.92, p = 0.017) and DD (adjusted OR 0.38, 95% CI 0.26–0.55, p < 0.001) had a significantly lower likelihood of achieving a good mRS outcome compared to the AC (Table 3). Figure 1 illustrates the mRS distribution, showing significantly better outcomes in AC than in APT and DD. Although HT occurred relatively frequently in the AC, it was more likely to be associated with a good mRS outcome.
Table 2.
Event rates of outcomes according to the treatment after HT
| Event Rate | All | AC | APT | DD | p |
|---|---|---|---|---|---|
| Number | 763 | 324 | 191 | 248 | |
| HT exacerbation in imaging | 0.038 | ||||
| Number of events | 219 | 100 | 41 | 78 | |
| Event rates (%, 95% CI) |
28.7 (25.5–32.1) |
30.9 (25.9–36.2) |
21.5 (15.9–28.0) |
31.5 (25.7–37.6) |
|
| 3 months good mRS (0–2) | < 0.001 | ||||
| Number of events | 208 | 125 | 42 | 41 | |
| Event rates (%, 95% CI) |
27.3 (24.1–30.6) |
38.6 (33.3–44.1) |
22.0 (16.3–28.5) |
16.5 (12.1–21.8) |
|
| Symptomatic ICH | < 0.001 | ||||
| Number of events | 38 | 5 | 4 | 29 | |
| Event rates (%, 95% CI) | 5.0 (3.5–6.8) | 1.5 (0.5–3.6) | 2.1 (0.6–5.3) | 11.7 (8.0–16.4.0.4) | |
|
Recurrence of stroke in 3-month |
0.047 | ||||
| Number of events | 18 | 5 | 9 | 4 | |
| Event rates (%, 95% CI) | 2.4 (1.3–3.4) | 1.5 (0.2–2.9) | 4.7 (1.7–7.7) | 1.6 (0.0–3.2.0.2) | |
|
Composite vascular outcome in 3-month |
< 0.001 | ||||
| Number of events | 95 | 24 | 22 | 49 | |
| Event rates (%, 95% CI) |
12.5 (10.1–14.8) |
7.4 (4.6–10.3) |
11.5 (7.0–16.0) |
19.8 (14.8–24.7) |
P-value by Chi-square test and log rank test
APT, antiplatelet; AC, anticoagulation; DD, no or quit medication; HT, hemorrhagic transformation; mRS, modified Rankin Scale; ICH, intracerebral hemorrhage
Table 3.
Associations of the treatment after HT with outcomes
| unadjusted OR (95% CI) | p-value | adjusted OR (95% CI) | p-value | |
|---|---|---|---|---|
| Primary outcome | ||||
| HT exacerbation in imaging | ||||
| AC | 1(Ref) | 1(Ref) | ||
| APT | 0.61 (0.40–0.93) | 0.022 | 0.48 (0.29–0.80) | 0.005 |
| DD | 1.03 (0.72–1.47) | 0.880 | 0.86 (0.53–1.37) | 0.521 |
| mRS score shifta | ||||
| AC | 1(Ref) | 1(Ref) | ||
| APT | 0.43 (0.32–0.60) | < 0.001 | 0.63 (0.43–0.92) | 0.017 |
| DD | 0.26 (0.19–0.36) | < 0.001 | 0.38 (0.26–0.55) | < 0.001 |
| Secondary and safety outcome | ||||
| 3-month good mRS (mRS 0–2) | ||||
| AC | 1(Ref) | 1(Ref) | ||
| APT | 0.45 (0.30–0.68) | < 0.001 | 0.55 (0.31–0.96) | 0.036 |
| DD | 0.32 (0.21–0.47) | < 0.001 | 0.39 (0.22–0.69) | 0.001 |
| Symptomatic ICH | ||||
| AC | 1(Ref) | 1(Ref) | ||
| APT | 1.36 (0.36–5.15) | 0.646 | 1.03 (0.24–4.47) | 0.965 |
| DD | 8.45 (3.22–22.17) | < 0.001 | 3.93 (1.18–13.16) | 0.026 |
a The ordinal shift across the range of modified Rankin scale scores toward a better outcome, for which the treatment effect is reported as a common OR with the 95% CI
Adjusted variables: age, male, previous_mrs, ASPECT score, hemorrhagic transformation, initial NIHSS, history of CHD, history of HTN, history of DM, history of DL, WBC, Total cholesterol, BUN, Creatinine, fasting glucose, history of stroke, history of PAD, history of TIH, history of DM, history of AC, acute thrombolytic treatment, discharge APT, discharge AC, mean SBP
AC, anticoagulation; APT, antiplatelet; DD, no or quit medication; HT, hemorrhagic transformation; HI, hemorrhagic infarction; PH, parenchymal hematoma; NIHSS, National institute of health stroke scale; BMI, body mass index; WBC, white blood cell; BUN, blood urea nitrogen; LDL-C, low density lipoprotein; TIA, transient cerebral ischemia; PAD, peripheral arterial disease; CHD, coronary heart disease; HTN, hypertension; DM, diabetes mellitus; DL, dyslipidemia; IVT, intravenous thrombolysis; IAT, intra-arterial thrombectomy; SBP, systolic blood pressure
Fig. 1.
3-month mRS distribution according to the treatment after HT
Safety Outcome and Secondary Outcomes
sICH occurred in 5% of cases, with incidence rates of 1.5% in the AC, 2.1% in the APT, and 11.7% in the DD (p < 0.001) (Table 2). The unadjusted model showed DD had an 8.45-fold higher risk of sICH than AC, and this trend persisted even after adjusting for variables (adjusted OR 3.93, 95% CI 1.18–13.16, p = 0.026) (Table 3). At 3 months, the proportion of patients with favorable outcomes (mRS 0–2) was higher in the AC (38.6%) than in the DD (16.5%) (Table 2). The AC had significantly higher odds of achieving a good functional outcome—45% higher than the APT group and 61% higher than the DD group (Table 3). At 3 months, the composite vascular outcome occurred in 12.5% of all cases, with statistical differences observed at 7.4% in the AC, 11.5% in the APT, and 19.8% in the DD (p < 0.001). Stroke recurrence at 3 months occurred in 2.4% of patients overall, with rates of 1.5% in the AC, 4.7% in the APT, and 1.6% in the DD, showing differences in incidence between groups (p = 0.047) (Table 2). The DD group had a higher 3-month composite outcome rate compared to the AC group, a difference that remained significant after adjustment (adjusted HR 2.01, 95% CI 1.09–3.72, p = 0.025). After adjusting for variables, there was no significant difference in stroke recurrence among the three groups (Table 4). The Kaplan-Meier survival curve for these outcomes is presented in Fig. 2, demonstrating differences in stroke recurrence and composite vascular outcomes across treatment groups.
Table 4.
Association of the treatment after HT with vascular outcomes
| unadjusted HR (95% CI) | p-value | adjusted HR (95% CI) | p-value | |
|---|---|---|---|---|
| Secondary outcome | ||||
| 3-month stroke recurrence | ||||
| AC | 1(Ref) | 1(Ref) | ||
| APT | 3.09 (1.04–9.23) | 0.043 | 2.86 (0.75–10.91) | 0.124 |
| DD | 1.05 (0.28–3.91) | 0.942 | 1.38 (0.30–6.24) | 0.677 |
| 3-month composite vascular outcome | ||||
| AC | 1(Ref) | 1(Ref) | ||
| APT | 1.59 (0.89–2.83) | 0.118 | 1.25 (0.64–2.45) | 0.518 |
| DD | 2.88 (1.77–4.69) | < 0.001 | 2.01 (1.09–3.72) | 0.025 |
Adjusted variables: age, male, previous_mrs, ASPECT score, hemorrhagic transformation, initial NIHSS, history of CHD, history of HTN, history of DM, history of DL, WBC, Total cholesterol, BUN, Creatinine, fasting glucose, history of stroke, history of PAD, history of TIH, history of DM, history of AC, acute thrombolytic treatment, discharge APT, discharge AC, mean SBP
AC, anticoagulation; APT, antiplatelet; DD, no or quit medication; HT, hemorrhagic transformation; HI, hemorrhagic infarction; PH, parenchymal hematoma; NIHSS, National institute of health stroke scale; BMI, body mass index; WBC, white blood cell; BUN, blood urea nitrogen; LDL-C, low density lipoprotein; TIA, transient cerebral ischemia; PAD, peripheral arterial disease; CHD, coronary heart disease; HTN, hypertension; DM, diabetes mellitus; DL, dyslipidemia; IVT, intravenous thrombolysis; IAT, intra-arterial thrombectomy; SBP, systolic blood pressure
Fig. 2.
Associations of the treatment after HT with 3-month vascular outcomes. (A) Stroke recurrence, (B) composite vascular outcome
Discussion
Our study explored the radiological and clinical outcomes of anticoagulation in patients with CES who developed HT, analyzing treatment outcomes after HT occurrence. While anticoagulation was associated with a radiological worsening of HT, it did not increase sICH incidence. Moreover, patients receiving anticoagulation showed significantly better functional outcomes at three months. These findings are clinically important, as they highlight the potential benefits of carefully tailored therapy in supporting long-term recovery, despite the longstanding concern that HT could worsen with anticoagulation.
The high prevalence of HT in CES is associated with larger infarct volumes and impaired collateral circulation, with a greater incidence of parenchymal hematoma than other stroke types [10, 23]. Anticoagulation for secondary prevention further increases HT risk[24–26], though its development is multifactorial, influenced by systemic and neuroinflammatory factors [6, 27, 28]. it is generally considered that DOACs carry a lower risk compared with warfarin [16]. In addition, left atrial appendage closure has increasingly been utilized as an alternative strategy in selected patients and has been shown to markedly reduce the risk of intracranial hemorrhage compared with long-term oral anticoagulation [29]. Careful management of these factors is crucial in balancing ischemic prevention with hemorrhagic risks.
Our findings corroborate the results of several recent large-scale trials that advocate for early anticoagulation in patients with atrial fibrillation related stroke. Notably, ELAN, TIMING and ARESTtrials have demonstrated that early initiation of anticoagulation is non-inferior to delayed initiation in terms of both safety and efficacy, particularly in preventing recurrent ischemic events[10–12]. A subgroup analysis of the ELAN examined outcomes based on the presence of HT[18]. The results showed no significant difference between early and late initiation of anticoagulation in terms of efficacy and safety, regardless of HT status. For patients with HT, the adjusted risk difference for early treatment compared to late treatment was − 2.2% (95% CI −7.8% to 3.5%), with no difference in the incidence of sICH, suggesting that early treatment in the presence of HT does not significantly compromise safety and may even be beneficial. However, when focusing specifically on patients with parenchymal hematoma, early anticoagulation was associated with a 25.1% higher likelihood of a poor 3-month outcome (mRS 3–6). While our study differs in evaluating anticoagulation after HT occurrence rather than at drug initiation, both studies suggest anticoagulation can be relatively safe in selected cases, potentially improving prognosis when severe forms like parenchymal hematoma type 2 are excluded. Another study examined HT progression in 220 patients and reported HT worsening in 14.3% of anticoagulated patients versus 4.2% in the antiplatelet group, though without statistical significance [19]. However, its small warfarin cohort (n = 28) and lack of CES focus limit its conclusions. In contrast, our study analyzed nearly 700 CES patients, offering a more robust assessment.
Our findings indicate that HT worsened more in the AC than in the APT group, with no significant difference compared to DD. However, anticoagulation, despite radiological progression, was not associated with sICH risk; rather, DD patients had a 3.93-fold higher sICH risk. This suggests HT subtype may be more crucial than medication in determining sICH risk. Prior studies, including ELAN and meta-analyses, reported sICH rates under 1%[10, 11, 30], whereas our cohort’s rate was 5%, suggesting HT itself increases symptomatic hemorrhage risk. Nonetheless, in the AC group, sICH incidence was only 1.5%, reinforcing that anticoagulation can be used safely in specific HT subtypes. Anticoagulation following HT was associated with a lower risk of composite vascular events, including mortality, though its effect on stroke recurrence was not statistically significant, likely due to the low overall event rate (2.4%). However, AC showed a trend toward lower recurrence compared to APT. Our ischemic stroke rate (1.5%) was consistent with the ELAN trial and lower than the 3% reported in DOAC meta-analyses[11, 30], aligning our findings with prior research. The seemingly lower recurrence rate in DD likely reflects underdetection due to early mortality or sICH rather than true stroke risk reduction. Nevertheless, it is also possible that patients in the DD group had higher baseline stroke severity or other contraindications, which may have precluded the use of anticoagulation. These alternative explanations should also be considered. Kaplan-Meier analysis showed that most vascular events occurred early, suggesting the high sICH incidence in DD was due to more frequent parenchymal hematomas. On the other hand, the use of anticoagulation did not appear to significantly increase mortality when compared to findings from previous studies on in CES [30, 31]. These findings suggest anticoagulation after HT is relatively safe and may improve stroke prognosis if high-risk HT types are avoided.
A crucial consideration in this study is selection bias. The AC group had lower initial stroke severity and higher ASPECT scores, which could influence outcomes. However, our analysis adjusted for these factors to minimize confounding. To minimize these differences as much as possible, a regression analysis was conducted; however, residual confounding and bias cannot be completely ruled out. Further well-designed studies are needed to confirm these results.
At our center, calculating infarct size during image analysis was not available. Since most of our centers used DWI as the initial examination, we attempted to adjust for severity using the ASPECT score derived from DWI images. Additionally, the score was measured separately for the anterior and posterior circulation. However, due to the small proportion of cases involving the posterior circulation, we were unable to analyze differences in outcomes based on stroke location. Future large-scale multicenter studies will be needed to clarify whether outcomes differ according to stroke location. Additionally, While MRI is more sensitive than CT in detecting small HT, it may overestimate the extent of HT. Since the classification of HT is generally based on CT, our MRI-based study may have been more sensitive in detecting HT. However, excluding cases of small hemorrhagic infarctions, studies have reported a similar detection rate for HT between CT and MRI[32]. Therefore, we believe that our study on HT using MRI provides sufficiently reliable results.
Several limitations of our study warrant consideration. Firstly, despite analyzing a large number of HT patients, the single-center design introduces potential center bias, limiting generalizability. Secondly, the retrospective nature of the study presents inherent limitations, such as the inability to control for all confounding variables and the lack of temporal causality assessmentThirdly, the long study period (2011–2023) encompassed substantial advances in stroke care, including changes in imaging protocols and treatment strategies, which may have introduced heterogeneity and influenced outcomes. Fourthly, image analysis relied on individual expert judgment, which may introduce variability in interpretation. We aimed to mitigate this by conducting regular meetings to review and standardize analyses. Fourthly, our classification of treatment methods did not distinguish between different drug doses or types, necessitating further research with larger cohorts. Finally, treatment decisions—including the choice, dose, and duration of AC, APT, or DD, as well as the timing and frequency of follow-up imaging after HT—were made at the discretion of the attending physicians. Although this reflects real-world practice, the absence of a uniform protocol introduces variability and potential selection bias. Despite these limitations, our study remains significant as a relatively large-scale investigation focused exclusively on patients who developed HT. Future advances in artificial intelligence may help reduce interobserver variability in imaging assessment and support more personalized treatment strategies, which could address some of these limitations [33].
In conclusion, our study suggests that anticoagulation administration in CES patients with HT, excluding parenchymal hematoma type 2, may lead to radiological worsening of HT but does not increase sICH and is associated with improved functional outcomes. These findings may challenge the prevailing notion that the mere presence of HT should preclude anticoagulation therapy and underscore the importance of patient-tailored approaches that prevent embolic events while effectively managing hemorrhagic risk. However, our results should be interpreted with caution. This study should be regarded as hypothesis-generating, and confirmation from future large-scale prospective studies is warranted.
Acknowledgements
None.
Author Contributions
Authors’ contributions-Conceptualization: JT Kim, H Kim-Data curation: YE An, BS Seo, JM Kim, KH Choi, MS Park-Formal analysis: JS Lee-Methodology: H Kim, JT Kim, YE An, BS Seo, JM Kim, KH Choi, MS Park-Writing – original draft: H Kim-Writing – Review and editing: H Kim, JT Kim-Supervision: JT Kim.
Funding
This study was supported by a grant (BCRI24011) of Chonnam National University Hospital Biomedical Research Institute. This work was supported by Establishment of K-Health National Medical Care Service and Industrial Ecosystem funded by the Ministry of Science and ICT(MSIT, Korea) Balanced National Development Account. [Project Name: Establishment of K-Health National Medical Care Service and Industrial Ecosystem/Project Number : ITAH0603230110010001000100100]
Data Availability
Data used in this study are available upon reasonable request to the corresponding author.
Declarations
Ethics Approval
The current study was approved by the institutional review board of Chonnam National University Hospital (CNUH-2024-294).
Disclosure
Dr. H Kim reports no disclosures. Dr. JT Kim reports no disclosures. Dr. MS Park reports no disclosures. Dr. KH Choi reports no disclosures. Dr. JM Kim reports no disclosures. Dr. YE An reports no disclosures. Dr. BS Seo reports no disclosures. Dr. JS Lee reports no disclosures.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Toni D, Fiorelli M, Bastianello S, et al. Hemorrhagic transformation of brain infarct: predictability in the first 5 hours from stroke onset and influence on clinical outcome. Neurology. 1996;46:341–5. [DOI] [PubMed] [Google Scholar]
- 2.Berger C, Fiorelli M, Steiner T, et al. Hemorrhagic transformation of ischemic brain tissue -: asymptomatic or symptomatic? Stroke. 2001;32:1330–5. [DOI] [PubMed] [Google Scholar]
- 3.Fiorelli M, Bastianello S, von Kummer R, et al. Hemorrhagic transformation within 36 hours of a cerebral infarct - relationships with early clinical deterioration and 3-month outcome in the European cooperative acute stroke study i (ecass i) cohort. Stroke. 1999;30:2280–4. [DOI] [PubMed] [Google Scholar]
- 4.Molina CA, Montaner J, Abilleira S, et al. Timing of spontaneous recanalization and risk of hemorrhagic transformation in acute cardioembolic stroke. Stroke. 2001;32:1079–84. [DOI] [PubMed] [Google Scholar]
- 5.Rothwell PM, Buchan A, Johnston SC. Recent advances in management of transient ischaemic attacks and minor ischaemic strokes. Lancet Neurol. 2006;5:323–31. [DOI] [PubMed] [Google Scholar]
- 6.Zhang J, Yang Y, Sun H, et al. Hemorrhagic transformation after cerebral infarction: current concepts and challenges. Ann Transl Med. 2014;2:81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Paciaroni M, Agnelli G, Corea F, et al. <article-title update="added">Early hemorrhagic transformation of brain infarction: rate, predictive factors, and influence on clinical outcome: results of a prospective multicenter study. Stroke. 2008;39:2249–56. [DOI] [PubMed] [Google Scholar]
- 8.Powers WJ, Rabinstein AA, Ackerson T, et al. Guidelines for the early management of patients with acute ischemic stroke: 2019. update… Stroke. 2019;50:e344–418. [DOI] [PubMed] [Google Scholar]
- 9.Klijn CJM, Paciaron M, Berge E, et al. Antithrombotic treatment… Eur Stroke J. 2019;4:198–223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Paciaroni M, Bandini F, Agnelli G, et al. Hemorrhagic transformation in patients with acute ischemic stroke and atrial fibrillation…. J Am Heart Assoc. 2018;7:e010133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Oldgren J, Asberg S, Hijazi Z, et al. Early versus delayed non-vitamin k antagonist oral anticoagulant therapy…. Circulation. 2022;146:E279–279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Fischer U, Koga M, Strbian D, et al. Early versus later anticoagulation for stroke with atrial fibrillation. N Engl J Med. 2023;388:2411–21. [DOI] [PubMed] [Google Scholar]
- 13.Labovitz AJ, Rose DZ, Fradley MG, et al. Early Apixaban use following stroke in patients with atrial fibrillation: results of the AREST trial. Stroke. 2021;52:1164–71. 10.1161/STROKEAHA.120.030042. [DOI] [PubMed] [Google Scholar]
- 14.Rose DZ, Meriwether JN, Fradley MG, et al. Protoc AREST… Front Neurol. 2019;10:975. 10.3389/fneur.2019.00975. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ruff CT, Giugliano RP, Braunwald E, et al. Comparison of the efficacy and safety of new oral anticoagulants with warfarin. Lancet. 2014;383:955–62. [DOI] [PubMed] [Google Scholar]
- 16.Seiffge DJ, Werring DJ, Paciaroni M, et al. Timing of anticoagulation after recent ischaemic stroke in patients with atrial fibrillation. Lancet Neurol. 2019;18:117–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wang X, Ouyang M, Yang J, et al. Anticoagulants for acute ischaemic stroke. Cochrane Database Syst Rev. 2021;10:CD000024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rohner R, Kneihsl M, Goeldlin MB et al. Early versus late initiation of direct oral anticoagulants… circulation. 2024;150:19–29. [DOI] [PubMed]
- 19.Kim JT, Heo SH, Park MS, et al. Use of antithrombotics after hemorrhagic transformation in acute ischemic stroke. PLoS One. 2014;9:e89798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Adams HP Jr, Bendixen BH, Kappelle LJ, et al. Classification of subtype of acute ischemic stroke. Stroke. 1993;24:35–41. [DOI] [PubMed] [Google Scholar]
- 21.Barber PA, Demchuk AM, Zhang J, et al. Validity and reliability of a quantitative computed tomography score…. Lancet. 2000;355:1670–4. [DOI] [PubMed]
- 22.Puetz V, Sylaja PN, Coutts SB, et al. Extent of Hypoattenuation on Ct angiography source images predicts functional outcome. Stroke. 2008;39:2485–90. [DOI] [PubMed] [Google Scholar]
- 23.Tu HTH, Campbell BCV, Christensen S, et al. Worse stroke outcome in atrial fibrillation…. Int J Stroke. 2015;10:534–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Marsh EB, Llinas RH, Hillis AE, et al. Hemorrhagic transformation in patients with acute ischaemic stroke…. Eur J Neurol. 2013;20:962–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Pande SD, Win MM, Khine AA, et al. Haemorrhagic transformation following ischaemic stroke: a retrospective study. Sci Rep. 2020;10:5319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Hong JM, Kim D, Kim M. Hemorrhagic transformation after ischemic stroke: mechanisms and management. Front Neurol. 2021. 10.3389/fneur.2021.703258. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Rosell A, Foerch C, Murata Y, et al. Mechanisms and markers for hemorrhagic transformation after stroke. Acta Neurochir Suppl. 2008;105:173. [DOI] [PubMed] [Google Scholar]
- 28.Carnicelli AP, Hong H, Connolly SJ et al. Direct oral anticoagulants versus warfarin in patients with atrial fibrillation… stroke. 2022;53:242–55. [DOI] [PMC free article] [PubMed]
- 29.Reddy VY, Doshi SK, Kar S, et al. 5-year outcomes after left atrial appendage closure: from the PREVAIL and PROTECT AF trials. J Am Coll Cardiol. 2017;70:2964–75. 10.1016/j.jacc.2017.10.021. [DOI] [PubMed] [Google Scholar]
- 30.Chun KH, Lee H, Hong JH, Seo KD. Prognosis of patients with ischemic stroke with prior anticoagulant therapy: direct oral anticoagulants versus warfarin. J Am Heart Assoc. 2024;13(15):e034698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Sakamoto Y, Okubo S, Nito C, et al. The relationship between stroke severity and prior direct oral anticoagulant therapy…. Eur J Neurol. 2017;24:1399–406. [DOI] [PubMed] [Google Scholar]
- 32.Hsia AW, Latour LL, Somani S, et al. Comparison of hemorrhage on ct versus mri after thrombectomy: the hecate study. Stroke Vasc Interv Neurol. 2024. 10.1161/SVIN.124.001441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Saini H, Rose DZ. The ghost in the machine: artificial intelligence in neurocardiology will advance stroke care. Neurohospitalist. 2024;0(0):1–2. 10.1177/19418744241288887. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data used in this study are available upon reasonable request to the corresponding author.