Clinical manifestations

Presentation and course

Hemorrhagic transformation can be defined both radiographically and clinically. Hemorrhagic transformation encompasses a broad spectrum of secondary bleeding, ranging from small areas of petechial hemorrhage to massive space-occupying hematomas (116).

Radiographically, the European Cooperative Acute Stroke Study (ECASS) investigators classified hemorrhagic transformation into hemorrhagic infarction (petechial infarction without space-occupying effect) and parenchymal hematoma (hemorrhage with mass effect). Hemorrhagic infarctions were further subdivided into hemorrhagic infarction 1 (small petechiae) and hemorrhagic infarction 2 (more confluent petechiae). Similarly, parenchymal hematomas were further subdivided into parenchymal hematoma 1 (less than 33% of the infarcted area with some mild space-occupying effect) and parenchymal hematoma 2 (greater than 33% of the infarcted area with significant space-occupying effect or clot remote from the infarcted area). In the ECASS I analysis, hemorrhagic infarction 1, hemorrhagic infarction 2, and parenchymal hematoma 1 did not modify the risk of early neurologic deterioration, death, and disability, whereas parenchymal hematoma 2 had a devastating impact on early neurologic course and on 3-month death (50). The interrater agreement for the diagnosis of hemorrhagic transformation based on the ECASS criteria was estimated in 18 raters from eight comprehensive stroke centers that participated in a multicenter trial of endovascular therapy for large-vessel occlusion in which 30 patients with a mean age of 74 years and a median NIHSS score of 18 at admission were selected for analysis (58). Sixty percent of these patients had intravenous thrombolysis before mechanical thrombectomy, and 80% had a successful recanalization. The raters all showed moderate interrater agreement (kappa 0.55) for the presence of hemorrhagic transformation, whereas interrater agreement for the ECASS categories was only fair (kappa 0.27). Substantial interrater agreement (kappa 0.72) was observed when analysis was dichotomized between parenchymal hematoma type 2 and the other ECASS classification criteria (58).

A “significant” hemorrhage can be defined by volume and size; a study discovered that a hemorrhage greater than 25 mL will produce a more clinically significant outcome in terms of a worsening NIH stroke scale at the time of a hospital discharge than those hemorrhages less than 25 mL (34).

Clinically, a second approach to classification describes hemorrhagic transformation as either symptomatic or asymptomatic. The National Institute of Neurological Disorders and Stroke (NINDS) tPA Stroke Study Group recognized that small hemorrhages in critical brain areas could also be devastating, leaving clinicians as the most qualified judges of significant hemorrhagic transformation. If a decline in the patient’s condition was temporally associated with hemorrhage on the CT scan, the hemorrhagic transformation was classified as symptomatic (132). The NINDS group found that the variables independently associated with an increased risk of symptomatic intracerebral hemorrhage were the severity of neurologic deficit as measured by the National Institutes of Health Stroke Scale and brain edema (acute hypodensity) or mass effect by CT before treatment. However, their multivariate regression model correctly predicted tPA-treated patients who would or would not have a symptomatic hemorrhage only 57% of the time. Symptomatic hemorrhagic transformation occurred in 6.4% (n=22) of patients treated with tPA compared to 0.6% of placebo-treated patients. Nevertheless, in the subgroup of patients with a severe deficit, tPA-treated patients were more likely than those receiving placebo to have a favorable 3-month outcome, with a decrease in the absolute risk of mortality of 4%. The authors concluded that despite the higher rate of intracerebral hemorrhage, patients with severe strokes or edema or mass effect on the baseline CT were still reasonable candidates for tPA if administered within 3 hours of onset.

Of the 624 patients enrolled in the NINDS trial, half were treated with tPA and half with placebo. Twenty-two developed clinically significant hemorrhage, 20 (6.4%) in the tPA group and two (0.6%) in the placebo-treated group. Four symptomatic intracranial hemorrhages occurred outside of the vascular distribution of the presenting ischemic stroke (20% of all tPA-related symptomatic intracranial hemorrhages and 1.3% of all tPA-treated patients). Of the 10 patients with fatal hemorrhages, eight (7 tPA- and one placebo-treated patient) had symptom onset within the first 12 hours, and all had symptom onset within the first 24 hours. The presenting signs and symptoms of symptomatic intracranial hemorrhage among the 22 patients included deterioration in the level of consciousness in 20, increased weakness in 16, headache in five, and increased blood pressure or pulse in 11.

From the NINDS study, it should be noted that an additional 21 patients had asymptomatic intracranial hemorrhage during the first 36 hours, and five patients had symptomatic intracranial hemorrhage between 36 hours and 3 months. This is in keeping with prior literature describing that most hemorrhagic infarctions are asymptomatic (11). Their presence is often detected by serendipity on follow-up CT scans performed on patients who are stable or clinically improving (77; 136; 109; 166). Moreover, it is often difficult to attribute clinical deterioration to the presence of hemorrhagic transformation when other causes of stroke progression, such as evolving cytotoxic edema, brain herniation, or hypoperfusion coexist.

Several studies have given a range for elucidating the rate of symptomatic transformations reported from 0.6% to 6.5% (Table 1) of patients in the placebo arms of trials of thrombolytic therapy for acute ischemic stroke. A meta-analysis of thrombolytic trials of acute ischemic stroke reported a 7.7% overall incidence of symptomatic hemorrhages in the control patients (175).

When thrombolytics were added, the rates of symptomatic hemorrhagic transformation increased to between 6.4% and 19.8% with rtPA and between 8.0% and 21.2% with streptokinase (Table 2).

Table 1. Spontaneous Hemorrhage Rates in Placebo-treated Patients: Results of Thrombolytic Trials

The dynamic evolution of hemorrhagic transformation has important clinical implications initiating anticoagulation therapy or using thrombolytic agents for acute ischemic stroke. Hemorrhagic infarction is rarely detected in the first 6 hours following stroke onset and is present on initial CT scans done within the first 24 hours in only 5% of cardioembolic strokes (27; 66; 136). This provides a potentially safe "therapeutic window" for thrombolytic therapy administered within the first few hours of ischemia to salvage still viable tissue. The American Stroke Association guidelines recommend intravenous rtPA, particularly within 3 hours of stroke onset. In carefully selected patients, intravenous thrombolysis may also be administered up to 4.5 hours after stroke onset, whereas intra-arterial tPA within 6 hours of stroke onset may be used in selected patients with major ischemic stroke due to occlusion of the middle cerebral artery (143). The latest American Heart Association/American Stroke Association guidelines for patients with clinical stroke symptoms on awakening (“wake-up stroke”) or unclear symptom onset indicate that a diffusion-weighted imaging-fluid attenuated inversion recovery (DWI-FLAIR) mismatch between a positive diffusion-weighted image with no corresponding visible signal change on FLAIR sequences may be useful in identifying patients with acute ischemic stroke in whom reperfusion therapy may be beneficial. Patients exhibiting this DWI/FLAIR mismatch may also benefit from intravenous alteplase treatment (143). In a posthoc analysis of 503 patients enrolled in the WAKE-UP trial, hemorrhagic transformation occurred in about 30% of patients with unknown stroke onset treated with MRI-guided intravenous thrombolysis, compared to 18% of patients treated with placebo (85). In this study, intravenous thrombolysis, atrial fibrillation, higher median baseline NIHSS score at admission, higher serum glucose levels on admission, and higher lesion volume on DWI at baseline were found to be independent predictors of any hemorrhagic transformation. Based on the Heidelberg classification, parenchymal hematoma was the least common type of hemorrhagic transformation in these patients and was associated with poor functional outcomes. Intravenous thrombolysis, atrial fibrillation, and higher NIHSS score at admission were found to be predictors of parenchymal hematoma (85).

A quantitative systematic review of hemorrhagic transformation in ischemic stroke from 1985 to 2017 included 65 studies (7 randomized clinical trials and 58 observational studies) with 17,259 patients meeting inclusion criteria (75). The overall hemorrhagic transformation rate (ECASS-II criteria) in this review was 27%; among patients treated with intravenous thrombolysis, it was 32%, and in patients without such treatment, it was 20%. Both parenchymal hematoma and type 2 parenchyma hematoma rates were higher in patients who underwent intravenous thrombolysis than in patients who did not receive this reperfusion therapy (12% vs. 5% and 5% vs. 3%, respectively). Atrial fibrillation increased the risk of hemorrhagic transformation by 3-fold, whereas oral anticoagulation did so by two-and-a-half fold. Moreover, patients with hemorrhagic transformation had higher NIHSS scores, a high rate of cerebral infarction, and a poor functional outcome at 3 months as determined by the modified Rankin Scale (75).

Nationwide real-world data on the clinical application and outcomes of endovascular therapy showed that intracranial hemorrhage occurred in 9.4% of 7,674 Chinese patients with acute ischemic stroke; 27% of all patients were treated with prior intravenous thrombolysis, and 73% were treated directly with endovascular therapy because most patients were beyond the therapeutic window for thrombolysis. Compared with patients treated with endovascular therapy, patients treated with prior intravenous thrombolysis had significantly higher rates of intracranial hemorrhage (10.7% vs. 7%) and higher rates of rapid neurologic improvement (24.7% vs. 21.1%) (54).

Clinically significant hemorrhagic transformations most often occur within the first few days following stroke onset. The Cerebral Embolism Study Group reported that 20 (74%) of 27 hemorrhagic infarctions detected on CT were apparent within 4 days of the ischemic event (28). Later transformations also occur. A prospective serial CT study detected hemorrhagic infarction in 28 (43%) of 65 patients with ischemic stroke, of which 11 (39%) were detected in the first week, 15 (54%) between day 7 and 14, and another two (7%) in the third week (77). Clinical deterioration associated with hemorrhagic transformation occurred in only three patients, all within the first week, two of whom suffered parenchymal hematoma. The benign nature of late-occurring hemorrhagic infarction may reflect a different pathogenesis with reperfusion bleeding due to the opening of pial collaterals as infarct edema resolves.

The effect of anticoagulation on the incidence and severity of hemorrhagic transformation remains uncertain. Anticoagulation is frequently used after an embolic stroke to prevent early recurrence. The clinical dilemma rests on the balance of risk between preventing neurologic worsening due to recurrent embolism versus the potential for promoting symptomatic hemorrhagic transformation.

The International Stroke Trial randomized over 19,000 patients to subcutaneous heparin in doses of 10,000 units per day, 25,000 units per day, or placebo (80). The overall incidence of recurrent ischemic stroke within 14 days was 3.8% in the control arm and 2.9% in heparin-treated patients. In patients with atrial fibrillation, the incidence of recurrent stroke within the first 14 days was 4.9% in the control arm and 2.8% in patients treated with heparin. The Trial of ORG 10172 in Acute Stroke Treatment (TOAST) trial evaluated the heparinoid, danaparoid (The Publications Committee for the Trial of ORG 10172 in Acute Ischemic Stroke Investigators 1998). In this trial, 1281 patients with ischemic stroke who presented within 24 hours of symptom onset were randomized to this drug or placebo intravenously for 7 days. Recurrent ischemic strokes were diagnosed during the treatment period in 1.2% of patients; the rate of recurrent stroke was not influenced by stroke type (large-artery atherosclerosis, cardioembolism, small-artery occlusion, other, and undetermined cause). The relatively low risk of recurrent stroke in the International Stroke Trial is at odds with the cumulative data analysis from older reports of untreated patients that showed an average rate of recurrent stroke in the first 21 days of 12% (range 2% to 21%) (104; 26; 65; 147; 182; 133). However, a summary of stroke “megatrials” demonstrated recurrent ischemic stroke rates to range between 0.63 and 2.20/100 patients per week, and most experts now agree that the older estimates for recurrent stroke rates were high (159). The large numbers of evaluated patients in the International Stroke Trial and TOAST trial support the conclusions that the overall risk of early recurrent stroke is low and the absolute benefit of routine heparin is marginal. The risks and benefits in carefully assessed and closely monitored patient subgroups remain uncertain. As such, the American Stroke Association guidelines do not recommend urgent routine anticoagulation to improve neurologic outcome or prevent early recurrent stroke (143).

Clinical deterioration due to major hemorrhage into areas of bland infarction has been reported in patients receiving anticoagulants. In the International Stroke Trial, the reduction in the risk of recurrent strokes in patients with atrial fibrillation treated with heparin was almost offset by the increased incidence of hemorrhagic strokes associated with heparin (1.2% vs. 0.4%) (80). In the TOAST trial, symptomatic hemorrhagic transformation of stroke occurred in nine patients receiving ORG 10172 and in three patients receiving placebo (p=0.14) during the 7-day treatment period; three patients in each group suffered asymptomatic hemorrhagic transformation (The Publications Committee for the Trial of ORG 10172 in Acute Stroke Treatment Investigators 1998). The Cerebral Embolism Study Group reported that 22 of 24 patients with symptomatic hemorrhagic transformation were associated with anticoagulation therapy (27). Some reports suggest that although anticoagulation does not precipitate hemorrhagic transformation, anticoagulation may worsen the severity of spontaneous bleeding (26; 29; 20). A meta-analysis showed that in patients with acute cardioembolic stroke, early anticoagulation is associated with a nonsignificant reduction in ischemic stroke recurrence, no substantial reduction in death and disability, and increased intracranial bleeding (139). The clinical recommendation that anticoagulation be delayed at 5 to 7 days following a large cerebral infarction has been challenged by a report that infarct size and severity did not produce additional bleeding complications in patients treated with heparin when excessive anticoagulation was avoided (31).

The rate of hemorrhagic transformation in patients who received early oral anticoagulation was recently evaluated in a pooled analysis of 509 patients with acute cerebral infarction (313 of whom had atrial fibrillation) in four prospective open-label treatment trials and in two multicenter randomized controlled trials. These patients had a median NIHSS score of 2 points, a median infarct volume of 1.5 ml, and a median time to first dose of oral anticoagulant of 2 days, with 75% of patients receiving the standard full dose (05). Follow-up neuroimaging (non-contrast CT or brain MRI) from 7 to 30 days after ischemic stroke revealed hemorrhagic transformation in 6.8% of all patients, none of whom showed hemorrhagic transformation-related symptoms. Early direct oral anticoagulation (apixaban, rivaroxaban, or dabigatran) was given within 48 hours after stroke onset to 239 patients whose hemorrhagic transformation rate in follow-up neuroimaging was 5%. Direct oral anticoagulation was given 48 hours after stroke onset to 265 patients whose hemorrhagic transformation rate in subsequent neuroimaging was 8.3%. Hemorrhagic transformation was asymptomatic in both of these patient groups, and their only associated variable was infarct volume. This pooled analysis shows that early direct oral anticoagulation within the first 48 hours after index stroke was not associated with a higher hemorrhagic transformation rate on follow-up neuroimaging (05).

The issue of symptomatic hemorrhagic transformation bears directly on the safety of thrombolytic therapy for acute stroke. Severe intracranial hemorrhage resulting in neurologic deterioration or death is the most feared complication of thrombolytic treatment. In several trials, the incidence of parenchymal hematomas or symptomatic intracerebral hemorrhages have ranged from 6.4% to 21.2% (74; 125; 60; 61; 131; 43). These variations may be due to differences in the dose and type of thrombolytic agent used, the use of concomitant antithrombotic agents, the treatment window, the control of hypertension, and the execution of the study in the field. The National Institute of Neurological Diseases and Stroke Trial (131), which reported the lowest incidence of parenchymal hemorrhage, 6.4%, used 0.9 mg/kg of tissue plasminogen activator given within the first 3 hours of symptom onset. Blood pressure was carefully managed by protocol to avoid excessive hypertension during and following treatment. The European Cooperative Acute Stroke Study I reported a 19.8% incidence of parenchymal hematoma or symptomatic bleeding associated with a dose of 1.1 mg/kg of tissue plasminogen activator given within 6 hours of symptom onset (60). The higher dose and longer treatment window might explain the substantially higher incidence of bleeding in this study. Many patients (17.4%) were enrolled in this trial despite major protocol violations. Because of the concerns about European Cooperative Acute Stroke Study I, European Cooperative Acute Stroke Study II (61) used the National Institute of Neurological Diseases and Stroke tPA dosing regimen of 0.9 mg/kg, more rigorous training for interpretation of CT findings, and tighter control of hypertension before, during, and after administration of the drug. The symptomatic parenchymal hemorrhage rate fell to 8.8% in the tPA-treated group in the European Cooperative Acute Stroke Study II. In the ICARO study, which evaluated the efficacy and safety of systemic thrombolysis in patients with internal carotid artery occlusion, there were more cases of intracranial bleeding (17.8% vs. 11.1%) and fatal intracranial bleeding (2.8% vs. 0.4%) among patients treated with rtPA than controls (140). Three large, randomized, double-blind, placebo-controlled trials using cardiac doses of intravenous streptokinase reported rates of symptomatic intracerebral hemorrhage or parenchymal hematomas ranging from 10% to 21.2% (125; 43; 124). All three trials were terminated prematurely because of increased rates of intracerebral hemorrhage and mortality in the streptokinase-treated groups. As a result of these trials, streptokinase has been abandoned as therapy for acute ischemic stroke. Hemorrhagic transformation is also reported with the use of intraarterial thrombolysis for acute ischemic stroke. Historically, rates of hemorrhagic transformation have ranged from 2% to 11% in nonrandomized trials (70). The first and second Prolyse in Acute Cerebral Thromboembolism Trials (PROACT I and II) used intraarterial prourokinase or placebo in patients with acute ischemic stroke who presented within 6 hours of symptom onset (38; 53). The trials differed in dosing of prourokinase and also in dosing regimens for heparin. Symptomatic hemorrhagic transformation occurred in 15.4% of prourokinase-treated patients in the Prolyse in Acute Cerebral Thromboembolism Trial I, with the highest rates occurring in patients concomitantly treated with a high-dose heparin regimen. In Prolyse in Acute Cerebral Thromboembolism Trial II, symptomatic hemorrhagic transformation occurred in 10% of prourokinase-treated patients. Symptomatic hemorrhagic transformation may also differ between high and low doses of intravenous thrombolytic agent within 4.5 hours after ischemia onset. In fact, in a multicenter prospective trial with South Korean patients with ischemic stroke, the rate of symptomatic hemorrhagic transformation was not significantly higher in patients treated with a low dose of alteplase (8.4% at 0.6 mg/kg) compared with patients treated with a standard dose of alteplase (6.4% at 0.9 mg/kg) (96). Endovascular treatment combined with intravenous thrombolysis was more frequent in the low-dose alteplase group, which also had more cardioembolic stroke, atrial fibrillation, arterial hypertension, and a higher NIHSS score than the standard-dose alteplase group.

Trials have examined the effects of intraarterial tPA. The Interventional Management of Stroke Trial (IMS trial) involved patients who had an acute stroke treated with 0.6 mg/kg intravenous rtPA within 3 hours of stroke followed by 22 mg intraarterial rtPA. If a hemorrhage occurred within 36 hours, along with clinical deterioration, this was considered symptomatic. The results showed that symptomatic hemorrhage occurred in 6% and asymptomatic hemorrhage in 43% of patients; this rate is similar to the NINDS trial with just intravenous rtPA (79). Endovascular recanalization therapy is a useful option for select patients with ischemic stroke mainly due to large-artery occlusion. A meta-analysis of eight multicenter prospective randomized endovascular-therapy trials from January 2013 to May 2015, including 2,423 patients with cerebral infarct predominantly due to large-vessel occlusion, compared the results between the intervention arm (endovascular-treated patients) and the control arm (patients with intravenous thrombolysis) (32). Approximately 40% of the intervention arm patients underwent mechanical thrombectomy using retrievable stent devices. Symptomatic hemorrhagic infarction rates were no different between intervention and control arm patients. Moreover, no difference in the rate of symptomatic hemorrhagic transformation was observed between intervention-arm patients treated with retrievable stent devices and those who were not.

In a report of a retrospective observational study comprising 299 patients with acute middle cerebral artery occlusion who had undergone endovascular thrombectomy, an MRI (in 52% of the patients) or brain CT (in 48%) performed 3 days after the endovascular intervention revealed a hemorrhagic infarct or parenchymal hemorrhage as per ECASS criteria in 29% and 4% of patients, respectively (89). Hemorrhagic infarction and parenchymal hemorrhage were both associated with more severe neurologic disability and poor outcome, though factors for both complications were different. At admission, a higher NHISS score and higher glucose levels were associated with hemorrhagic infarct, whereas parenchymal hemorrhage was associated with longer intervals between stroke onset and endovascular treatment, lower pre-interventional ASPECTS score, and wake-up stroke (89).

Furthermore, hemorrhagic transformation seems to occur less commonly in patients with successful reperfusion after thrombolysis. In a systematic review and meta-analysis of 14 trials involving 2,379 patients with acute ischemic stroke and successful recanalization, an estimated reperfusion status of TICI 2c/3 in the grading score was associated with nearly 50% lower risk of any hemorrhagic transformation compared with those patients with a TICI 2b grading score. Moreover, patients with a TICI 2c/3 grading score also had better functional outcomes than those with a TICI 2b grading score (90).

In a post-hoc analysis of 478 patients who had acute anterior circulation ischemic stroke due to proximal artery occlusion and were included in a large trial of endovascular therapy (intra-arterial thrombolysis or mechanical thrombectomy, or both) administered within 6 hours of stroke onset versus usual care alone, hemorrhagic transformation occurred in 222 patients (46%) who had a CT scan at 5 days (169). Based on ECASS 2 classification, hemorrhagic infarction type 1 and type 2 occurred in 76 and 71 patients, respectively. Parenchymal hematomas, on the other hand, were less common: type 2 occurred in 38 patients and type 1 in 36 patients. Hemorrhagic transformation was symptomatic in 35 patients (46%), most of whom had parenchymal hemorrhage type 2. Compared with patients with no hemorrhagic transformation, all those with hemorrhagic transformation showed a trend towards poor functional outcome at 90 days, as estimated by the modified Rankin scale. In addition, the worst outcome was observed in patients with parenchymal hematoma type 2 (46% deceased), hemorrhagic infarction type 2 (28% deceased), and symptomatic hemorrhagic transformation (no figure available) (169).

Clinical, radiological, and procedural hemorrhagic transformation predictors were assessed in a prospective multicenter endovascular trial of patients with large vessel occlusion in the anterior circulation. All of these patients had ASPECTS scores of at least 6 points and underwent mechanical thrombectomy within 6 hours after stroke onset, with successful recanalization in the first pass (24). Hemorrhagic transformation (ECASS II criteria) occurred in 18% of patients at 24 hours, and 3% showed symptoms. Patients with ischemic stroke and radiological hemorrhagic transformation were mean aged 73.4 years and had a mean NISS score of 18 and a mean ASPECTS score of 9. Half of these patients previously had intravenous thrombolysis, and their onset-to-groin time was 255 minutes. In this trial, a higher NHISS score, a lower ASPECTS score, and a longer onset-to-groin puncture were independent hemorrhagic transformation predictors (24).

The aim of intravenous thrombolysis and endovascular stroke treatment is rapid restoration of brain perfusion to minimize cerebral tissue loss during an acute ischemic stroke. In a metaanalysis of 25 clinical trials from 2012 to 2018 enrolling 7,178 patients, there were no differences in the incidence of symptomatic intracranial hemorrhage between patients with ischemic stroke and large vessel occlusion who were treated with intravenous thrombolysis followed by mechanical thrombectomy and those who were treated with mechanical thrombectomy alone (47).

Endovascular therapy is the standard of care in patients with large vessel occlusion stroke in the anterior circulation. It has been suggested that hemorrhagic transformation may be associated with multiple attempts to retrieve the occluding thrombus. In support of this hypothesis, more than three passes significantly increased the risk of symptomatic hemorrhagic transformation by 3.6-fold in a large multicenter cohort of 593 patients with large vessel occlusion stroke (118). These patients had a mean age of 72 years, a median baseline NIHSS score of 15 points, a median ASPECTS score of 9 at admission, and two thirds of them underwent intravenous thrombolysis. Endovascular therapy (50%, stent retriever devices; 20%, aspiration catheters; and 30%, both aspiration and stent retriever devices) achieved successful reperfusion (TICI score 2b/3) in 92% of patients. The median time from stroke onset to groin puncture was 193 minutes and the median time to flow restoration was 93 minutes. Symptomatic intracranial hemorrhage occurred in 4.4% of patients and 14% of patients had intracerebral hemorrhage on neuroimaging within the first 24 hours. Neither the device type nor the rate of successful or unsuccessful recanalization were associated with hemorrhagic transformation risk (118).

Hemorrhagic transformation has been associated with a prominent, angiographically blush brain vascularity observed immediately after recanalization of proximal large-vessel occlusion with mechanical thrombectomy (148). This angiographic finding may result from hyperemia in the early stages of infarction and was defined as a capillary blush with or without arteriovenous shunting and early venous drainage. Hemorrhagic transformation on brain CT or MRI was found in 54% of 48 patients with ischemic stroke and proximal large-vessel occlusion in the anterior circulation who underwent mechanical thrombectomy with stent retriever devices within 6 hours of stroke onset. Hemorrhagic transformation was associated with younger age, higher NIHSS scores, lower pretreatment ASPECTS scores, intravenous rtPA before endovascular therapy, and a longer delay between stroke onset and mechanical thrombectomy (148). Angiographic blush was observed in 31 (64%) of 48 patients; 18 patients had a capillary blush alone, and 13 had a capillary blush with arteriovenous shunt and early venous drainage. Angiographic blush occurred more frequently in patients with hemorrhagic transformation (69% of cases) than those without (59%). In patients who had angiographic blush, hemorrhagic transformation may result from endothelial damage or loss of autoregulation; both mechanisms disrupt blood-brain barrier permeability (148). An early venous filling (contrast opacification of any cerebral vein before the late arterial phase) was identified on post-reperfusion digital subtraction angiography in 22% of 147 patients with large vessel ischemic stroke in the anterior circulation successfully treated by mechanical thrombectomy (45). The presence of this angiographic finding significantly increased by 6-fold both the risk of reperfusion hemorrhage (ECASS II classification) and the risk of symptomatic hemorrhage (increase in NIHSS score > 2). Compared with NIHSS at admission, patients with early venous filling on angiography and no reperfusion hemorrhage on neuroimaging had lower NIHSS scores 24 hours after endovascular mechanical revascularization. Early venous filling reflects a state of focal hyperperfusion in the ischemic area and may serve as a neuroimaging marker to predict hemorrhagic reperfusion and early outcome. Thus, this focal hyperperfusion may either preserve the ischemic penumbra, resulting in an early neurologic improvement, or may facilitate an excess of blood flow in the territory of brain infarction, resulting in hemorrhagic reperfusion. At 90 days, neither the early venous filling nor reperfusion hemorrhage significantly influenced the functional outcome (45).

In a small single-center trial, a higher risk of hemorrhagic transformation was associated with poor collateral circulation as determined by a 4-dimensional CT angiography (4D-CTA), a relatively new technique for assessing intracerebral artery occlusion (23). In this trial, 30 out of 71 (42%) patients on endovascular treatment for acute cerebral infarct due to large vessel occlusion showed hemorrhagic transformation. The mean time from stroke onset to groin puncture was 355 minutes, and from puncture to recanalization it was 89 minutes; the mean number of passes was 1.69. Furthermore, a history of atrial fibrillation, a high baseline NIHSS score, and hyperglycemia were also independent risk factors associated with hemorrhagic transformation (23).

Coronavirus disease 2019 (COVID-2019), caused by the novel SARS-coronavirus-2 (SARS-CoV-2), emerged in China in late 2019 and was declared a pandemic by the World Health Organization in March 2020. Ischemic stroke has been reported in up to 5% of patients hospitalized with SARS-CoV-2 infection (167). In a retrospective case-control study from a single hospital system during the COVID-19 pandemic, the proportion of large vessel occlusion stroke in patients with COVID-19 infection (31.7%) was twice that of non-COVID-19 patients with stroke (15.3%) (95). A systematic review of eight studies of mechanical thrombectomy in patients with large vessel occlusion and SARS-CoV-2 infection found a postprocedural cerebral hemorrhage in 4% of 73 patients (04). In a multicenter cohort study, symptomatic hemorrhage occurred in 5.4% of 93 patients with large vessel occlusion and SARS-COV2 infection who underwent mechanical thrombectomy (19). Therefore, the rate of symptomatic hemorrhagic transformation was similar between patients with large vessel occlusion stroke treated with endovascular therapy during the pandemic and those so treated in prepandemic times (56; 19). In another single-center retrospective study, the rate of hemorrhagic transformation was twice as high in prepandemic patients with large vessel occlusion treated with mechanical thrombectomy (20.6%) as in pandemic-era patients (9.5%), though this was not statistically significant (183).

Prognosis and complications

Although the spectrum of hemorrhagic transformation ranges from minor petechial bleeding to major mass-producing hemorrhage, several reports suggest that most hemorrhagic changes are minor and not associated with clinical deterioration (77; 76; 109). Prognosis is usually determined by infarct size and location and by systemic complications rather than by secondary bleeding, unless massive. Hornig and colleagues reported clinical worsening in only three of 28 patients with hemorrhagic transformation detected by serial CT scan (77). Late-appearing hemorrhagic transformation was typically petechial in nature and rarely associated with neurologic worsening. Ott and colleagues reported deterioration in seven of 44 patients with hemorrhagic infarcts (136). All seven patients were on anticoagulants, and six of seven had moderate or large infarcts. In one MRI study, no clinical worsening occurred among 24 patients with hemorrhagic infarction (76). The first and second European Cooperative Acute Stroke Studies (ECASS I and II) found that hemorrhagic infarction was not associated with an increased risk of early or late neurologic deterioration or 3-month death or disability (50; 11). In contrast to the benign course of hemorrhagic infarction, parenchymal hematomas are typically accompanied by neurologic decline or death, except for small hemorrhages (77; 135; 13; 37).

A study found that patients who experienced asymptomatic hemorrhagic transformation were more likely to have an increment of mRS at 3 months (adjusted OR 1.90, 95% CI 1.27-2.82) (142). Also, Lei and colleagues found that asymptomatic hemorrhagic transformation and symptomatic hemorrhagic transformation after acute ischemic stroke worsened long-term clinical outcomes, although it did not affect the risk of stroke recurrence (111). In fact, after adjusting for other confounding factors, the risk of poor outcome at 3 months and 1 year was significantly higher among those with asymptomatic hemorrhagic transformation and symptomatic hemorrhagic transformation than among those without hemorrhagic transformation.

In thrombolytic trials, approximately half of all parenchymal hematomas associated with thrombolytic therapy were fatal. European Cooperative Acute Stroke Studies I and II data confirm these findings, with large parenchymal hematomas far more likely to result in early neurologic deterioration and death at 3 months (50; 11).

Ninety of 128 patients with ischemic stroke received intravenous thrombolysis and a brain MRI (3 Tesla), including susceptibility-weighted imaging to increase the detection of small hemorrhages, within 4.5 hours of symptom onset and at 7 days after thrombolysis; 52 patients (58%) showed hemorrhagic transformation (ECASS II definition) of infarct on SWI at 7 days, and 44 patients were asymptomatic. Patients with post-thrombolytic hemorrhagic transformation had a significantly higher NIHSS score and glucose levels at baseline than those with no hemorrhagic transformation. Hemorrhagic transformation was associated with an unfavorable functional outcome at 3 months (mRS 3-6) in both symptomatic and asymptomatic patients with hemorrhagic transformation; in this small cohort, nevertheless, after adjustment for initial stroke severity, no significant impact of hemorrhagic transformation on outcome at 3 months was observed (08).

Short-term outcome in patients with asymptomatic hemorrhagic transformation was assessed in the Thrombolysis Implementation and Monitoring of Acute Ischemic Stroke in China (TIMS-China) registry. In this study, 89 (9.5%) of 904 patients with ischemic stroke treated with intravenous thrombolysis within 4.5 hours showed asymptomatic hemorrhagic transformation on brain CT performed 24 to 36 hours after thrombolysis (87). The HT2 and HT1 subtypes (ECASS II classification) were the first and second most common patterns, respectively, and the least common subtype was PH2. Patients with asymptomatic hemorrhagic transformation were older and had more atrial fibrillation and cardioembolic stroke, higher NIHSS scores, and fewer previous transient ischemic attacks than patients without asymptomatic hemorrhagic transformation. There was no difference in mortality rates at 7 and 90 days after thrombolysis and in functional outcomes at 3 months between patients with and without hemorrhagic transformation. A more favorable outcome at 90 days was observed in patients with HT2 and HT1 subtypes of hemorrhagic transformation than in those with the PH1 and PH2 subtypes.

Clinical vignette

The following clinical vignette comes from a case report by Dr. Jose Maria Calvo-Romero from The Department of Internal Medicine, Hospital de Zafra, in Spain (21).

A 61-year-old man presented 8 hours after developing dysarthria and weakness in his right arm and leg. Past medical history included smoking, hypertension, dilated cardiomyopathy, and chronic atrial fibrillation. His medication list involved digoxin and enalapril. At presentation, blood pressure was 160/80 mm Hg. Neurologic examination showed right hemiplegia and motor aphasia. CT was normal. Lab results revealed normal glucose, creatinine, hemoglobin, and coagulation studies. Total cholesterol was 181 mg/dL, HDL 26 mg/dL, triglycerides 215 mg/dL, and fibrinogen 690 mg/dL.

An ECG confirmed atrial fibrillation. A transthoracic echocardiogram showed left ventricular dilatation, normal ejection fraction, and left atrial dilatation without thrombus formation. The most likely cause of the stroke was thromboembolic; thus, the patient was placed on enoxaparin 1 mg/kg twice daily and 300 mg/day of aspirin. Neurologic examination did not change, and blood pressure lowered from the initial reading. CT two days after admission revealed a left cerebral infarct with hemorrhagic transformation. Platelet count and coagulation studies were rechecked and remained normal. Enoxaparin was discontinued, but aspirin was continued. Carotid and vertebral duplex ultrasonography and brain CT were normal 8 days after admission.

According to this article by Dr. Calvo-Romero, concomitant therapy with enoxaparin and aspirin within the first 24 hours of the stroke contributed to the hemorrhagic transformation, which in this case was asymptomatic.

Although not illustrated in this case, hemorrhagic transformation has been noted to contribute to the pathophysiology of headache during an acute stroke, so headaches after a stroke must be monitored carefully (40).

Biological basis

Etiology and pathogenesis

Hemorrhagic transformation after an ischemic insult is multifactorial, involving multiple pathological processes from peripheral blood cells to neurovascular units.

Hemorrhage in an area of ischemic infarction occurs when blood extravasates through vessel walls damaged by ischemia. The occurrence of bleeding, therefore, requires an ischemic insult of sufficient severity and duration to alter vessel wall permeability and integrity, plus the restoration of adequate reperfusion, direct or collateral, to the site of injury (117; 63; 07). Secondary bleeding may occur with most stroke mechanisms, but several studies have demonstrated the special predilection of embolic infarction to undergo hemorrhagic transformation (51; 182; 77; 14).

Systemic inflammation can also increase the risk of hemorrhagic transformation with a 5-fold increase in MMP-9, which is involved in disrupting the cerebrovascular tight junction; inhibition of MMP-9 in mice has resulted in a reduced incidence of hemorrhagic transformation (120). MMP-9 variation is independently associated with symptomatic intracerebral hemorrhage or death during ischemic stroke treated with thrombolysis (81; 88).

A meta-analysis of blood biomarkers predicting hemorrhagic transformation of ischemic infarct before reperfusion therapy found that MMP-9 levels above 140 ng/ml increased the risk of symptomatic hemorrhagic transformation 30-fold (105). Fluid-attenuated Inversion Recovery (FLAIR) hyperintensity correlates with both MMP-9 level and risk of hemorrhagic transformation (86). In a retrospective study, an early fluid-attenuated inversion recovery (FLAIR) hyperintensity within acute ischemic lesions was associated with an increased risk of hemorrhagic transformation (02). This study included 134 patients with acute ischemic stroke of the anterior circulation who underwent thrombolytic therapy (intravenous, intra-arterial, or combined) within 6 hours of ischemia and a neuroimaging follow-up (diffusion-weighted imaging, FLAIR, and T2-weighted gradient echo sequences) within 5 days after the pretreatment scan. A FLAIR positivity (hyperintense lesion within the diffusion-weighted image lesion) was found in 56 patients (42%), and hemorrhagic transformation--mainly small petechiae and hematoma with a mass effect occupying 30% or less of the infarcted area--was observed in 51 patients (38%). A FLAIR lesion was geographically matched with the hemorrhagic transformation, and a FLAIR positivity was independently associated with a 4.3-fold higher risk of hemorrhagic transformation (02).

Leucocyte infiltration is important in triggering a blood-brain barrier disruption and hemorrhagic transformation. Leukocytes are critical to the neuro-inflammatory response (171). Xing and colleagues reported that white blood cell count was significantly higher in patients who experienced hemorrhagic transformation than in those who did not (180). The subsequent enhanced leukocyte infiltration might damage microvascular endothelial cells, mediate the opening of the blood-brain barrier, and lead to hemorrhagic transformation.

An acute disruption of the blood-brain barrier is a key player in hemorrhagic transformation of ischemic stroke. Accordingly, several biomarkers indicating blood-brain barrier disruption or activation of systemic inflammatory processes have been proposed as potentially useful tools for predicting a high risk of hemorrhagic transformation following cerebral ischemia. In 270 patients with ischemic stroke, high neutrophil-to-platelet ratio at admission increased by 2-fold the risk of hemorrhagic transformation based on ECASS II criteria, mainly parenchymal hematoma, compared to 270 matched controls (67). In 1005 patients with acute ischemic stroke, 74 of whom underwent reperfusion therapy, lower baseline lymphocyte-to-monocyte ratio was found to be associated with hemorrhagic infarction according to ECASS II criteria (155). Expression of circulating microRNAs (small, non-coding RNA molecules of 20 to 24 nucleotides) that regulate MMP-9 expression during stroke may be higher in patients with hemorrhagic infarct due to cardioembolic source based on TOAST criteria. In fact, plasma levels of miR-21-5p (highly expressed in vascular smooth muscle cells and endothelial cells), miR-206, and miR-3123 measured by quantitative real-time polymerase chain reaction (qRTPCR) were higher in 14 patients with hemorrhagic transformation in the first 7 days following cardioembolic stroke (185). In addition, low non-high-density lipoprotein cholesterol levels (triglyceride-rich lipoproteins, low-density lipoproteins, and very low-density lipoproteins) measured within 24 hours after admission increased the risk of hemorrhagic transformation by almost 2-fold (ECASS II criteria) after adjusting for confounding factors in 2043 patients with acute ischemic stroke. The mechanism of this association remains to be elucidated (172).

An acute disruption of the blood-brain barrier is a key player in the hemorrhagic transformation of cerebral ischemia. Several imaging modalities have been proposed for the identification of blood-brain barrier disruption. In this setting, arterial spin labeling MRI enables cerebral blood flow measurements and is sensitive for detecting hyperemic lesions, which are probably a predictor of blood-brain barrier breakdown. In a retrospective study of 25 patients with acute brain ischemia, hemorrhagic transformation was associated with hyperemic lesions (areas with relative CBF 1.4 or greater) on arterial spin labeling MRI maps within 6 hours after stroke onset and before treatment (intravenous thrombolysis, intra-arterial thrombolysis, or mechanical endovascular therapy, either as single procedures or in combination) (130). Hemorrhagic transformation was found in 15 patients (60%), 11 asymptomatic, mainly on brain MRI performed at a median of 7 days after stroke. Hemorrhagic infarct occurred in all nine patients with hyperemic lesions and in 6 of 16 patients (37.5%) without hyperemic lesions. Hemorrhagic transformation was also more frequent in patients with blood-brain barrier disruption (14 of 15; 93%) compared to patients with blood-brain barrier integrity (one of 10; 10%). Recanalization, estimated by digital subtraction angiography, was not associated with hemorrhagic transformation (130).

Moreover, modification of average blood-brain barrier permeability of the ischemic region has been estimated in patients with acute ischemic stroke before and 2 hours after intravenous thrombolysis, through a dynamic susceptibility contrast (T2*) sequence used for perfusion-weighted imaging (153). Half of the patients (18 of 36) showed increased permeability 2 hours after intravenous thrombolysis. The reperfusion rate was lower in patients with no reversible blood-brain barrier permeability than patients with reversibility of blood-brain barrier disruption. Hemorrhagic transformation (ECASS classification of parenchymal hematoma), evaluated by echo gradient MRI 24 hours after intravenous thrombolysis, was associated with focal regions of maximal blood-brain barrier permeability (153).

Early impairments in dynamic cerebral autoregulation during acute ischemic stroke were associated with hemorrhagic transformation (25). In 46 patients with acute infarct in the middle cerebral artery territory, dynamic cerebral autoregulation was noninvasively assessed by transfer function analysis between spontaneous oscillations in blood pressure and cerebral blood flow velocity in the middle cerebral artery within 6 hours after stroke onset. Hemorrhagic transformation was found on brain CT at 24 hours in 10 patients (22%). At admission, the cerebral dynamic phase--reflecting impaired cerebral autoregulation--was significantly lower in the ipsilateral hemisphere in patients who subsequently developed hemorrhagic transformation; gain measure, conversely, was not associated with this finding (25).

Fisher and Adams proposed the theory of "migratory embolism," with secondary reperfusion bleeding from ischemically injured capillaries as the pathophysiologic basis responsible for the evolution of hemorrhagic infarction (51). This concept, derived from their pathological observations, proposes that an embolic occlusion results in a distal ischemic injury that is initially pale. The embolic material is subject to fragmentation, dissolution, or lysis leading to distal migration and reperfusion of the ischemic tissue bed. Hemorrhage results from extravasation and diapedesis of blood through ischemically damaged vessels. Loss of cerebral microvascular integrity secondary to plasmin-generated laminin degradation, matrix metalloprotease activation, vascular adhesion protein-1 activity (69), high serum ferritin level (33), and transmigration of inflammatory leukocytes through the vessel wall have also been implicated (63).

The duration and the severity of ischemia are important determinants of hemorrhagic transformation. Angiographic studies show that partial or complete spontaneous recanalization occurs in up to 90% of embolic occlusions (182). The hypothesis that early recanalization may be protective against reperfusion bleeding remains unproven (117). Loh and colleagues found that among patients with MRI patterns of advanced basal ganglionic injury, successful recanalization predicts a higher risk of hemorrhagic transformation but better outcome (114). Also, reperfusion after stroke sonothrombolysis with microbubbles may predict intracerebral bleeding but does not seem to increase the risk of symptomatic intracranial hemorrhage (41). Experience to date shows that the incidence of secondary bleeding is acceptably low if thrombolysis is given within 180 minutes of stroke onset, whereas the frequency of bleeding complications increases when treatment is delayed (16; 60; 61; 131; 35). Using transcranial Doppler, Molina demonstrated that delayed (longer than 6 hours) spontaneous arterial recanalization following cardioembolic stroke was an independent predictor of hemorrhagic transformation (odds ratio 8.9, 95% confidence interval 2.1 to 33.3) (121).

The amount of residual cerebral blood flow within the territory of the occluded vessel determines the severity of ischemia. Hypodense changes on CT scans performed within 5 hours of symptom onset, presumably reflecting severe ischemia, predicts hemorrhagic transformation (166; 121). In a canine stroke model, reducing cerebral blood flow to less than 50% was critical for developing hemorrhagic infarction (150). SPECT has been utilized to evaluate pretreatment cerebral blood flow in patients undergoing superselective intraarterial thrombolytic therapy (168). Cerebral blood flow values were significantly lower in the five patients who developed hemorrhagic infarction compared to 15 patients who did not develop hemorrhage.

Hemorrhagic infarction may also occur in thrombotic strokes and other nonembolic stroke mechanisms but is uncommon compared to embolic infarction (182). Hemorrhage transformation has been recorded following early carotid endarterectomy for recent strokes. Hemorrhage is also commonly noted in posterior cerebral artery territory infarction associated with herniation of the temporal lobe, which compresses the artery perhaps incompletely or intermittently against the tentorial edge. Infarction due to vasospasm may become hemorrhagic. Terada reported hemorrhagic transformation in 35% (13 of 37) of infarcts due to vasospasm induced by aneurysmal subarachnoid bleeding (162). Presumably, the common theme of vessel occlusion leading to capillary ischemia and altered permeability followed by reperfusion and subsequent bleeding is operative in many of these examples.

Hemorrhagic transformation can occur distal to persisting arterial occlusion due to blood flow provided by collateral channels, although the frequency of hemorrhages secondary to this mechanism remains uncertain (77; 134; 13). Bang and associates found that angiographic grade of collateral flow strongly influences the rate of hemorrhagic transformation after therapeutic recanalization for acute ischemic stroke (09). Ogata and colleagues reported clinical and autopsy data on seven patients with hemorrhagic transformation distal to persistent embolic occlusion (134). The authors proposed that transient arterial pressure surges and the presence of efficient blood flow through leptomeningeal collateral vessels may lead to reperfusion bleeding. Persisting arterial occlusion was detected in four of 10 pathologically verified hemorrhagic infarction cases reported by Yamaguchi and colleagues (182). Evidence from several animal stroke models provides experimental support for the role of collateral circulation in the genesis of secondary bleeding (62; 48; 117). The late development of hemorrhagic transformation, occurring after the first week, may be due to the development of collateral circulation and the reperfusion of injured capillaries that reopen as infarct edema subsides (48; 66; 77; 14).

Acute ischemic stroke treatment involves thrombolysis, which increases the risk of hemorrhage by up to 10 times compared to controls (63). Cerebral ischemia can result in a loss of basal lamina of cerebral microvasculature through mechanisms such as plasmin-generated laminin degradation, activation of matrix metalloproteinases, or transmigration of leukocytes through vessel walls (63). tPA can cause hemorrhagic transformation in ischemic stroke through various mechanisms that have been proposed. It has neurotoxic side effects; it lyses clots, is an extracellular protease, and may increase excitotoxic calcium currents by being an NMDA-type glutamate receptor (173). Wang and colleagues mention how tPA may degrade extracellular matrix integrity by increasing matrix metalloproteinase dysregulation; this can increase the risk of hemorrhage (173).

The pathogenesis of parenchymal hematoma following ischemic injury has not been thoroughly studied. Hart and Easton proposed that only quantitative differences exist between parenchymal hematomas and hemorrhagic infarction (66). In some instances, however, parenchymal hematomas must result from the rupture of small penetrating arterioles analogous to mechanisms of hypertensive hemorrhages or the rupture of larger arteries rather than by extravasation of blood through disrupted capillary endothelial junctions.

Epidemiology

Hemorrhagic transformation is a common and natural consequence of embolic infarction. The true incidence of hemorrhagic transformation remains uncertain, and reported frequencies vary depending on the methodology used and the underlying pathogenesis of the ischemic insult (123).

Autopsy series have reported secondary bleeding in 51% to 71% of recent embolic infarctions as compared to a 2% to 21% incidence in nonembolic strokes (51; 113). The high incidence of hemorrhagic transformation reported in these studies reflects, in part, the ascertainment bias toward large infarct size and brain herniation in autopsy series. In a retrospective cohort of consecutive brain autopsies in 100 patients, a clinical diagnosis of ischemic stroke was given to 64 of the patients at admission (160). A brain CT performed 24 hours after intravenous thrombolysis or clinical worsening found hemorrhagic transformation in 10 patients with ischemic stroke. A neuropathological study showed 59 territorial infarcts and five lacunar infarcts. Thirty-four of these infarcts showed hemorrhagic transformation; diagnosis was established in 24 cases only at autopsy. Eighteen hemorrhagic infarcts occurred in nonthrombolysed patients, and intravenous thrombolysis was administered in the remaining 16 cases. No patient was anticoagulated during hospitalization; 31 patients were put on prophylactic isocoagulation and 17 on antiplatelet therapy. Hemorrhagic transformation distribution according to ECASS classification showed a predominance of HI2 and PH1 subtypes, each with frequencies of 30%; one fourth of hemorrhagic infarcts were of HI1 subtype. PH2 subtype (30% of the infarcted area with significant space-occupying effect or clot remote from infarcted area) accounted for one of six hemorrhagic infarcts and occurred exclusively in patients treated with intravenous thrombolysis (160).

CT studies have reported hemorrhagic infarction in 26% to 43% of nonanticoagulated patients with predominantly embolic infarcts (77; 135). Hemorrhagic infarction was detected in 28 (43%) of 65 patients with predominantly cardioembolic stroke studied prospectively with serial CTs over 4 weeks (77). In most CT series involving consecutive nonanticoagulated patients with infarcts of all types, the overall incidence reflects the inclusion of all stroke types, including small vessel infarcts as well as a nonserial neuroimaging methodology.

Spontaneous hemorrhagic transformation rates were 38% to 71% in autopsy studies and 13% to 43% in CT studies (83). Thrombolysis increases the risk of hemorrhagic stroke by about 7% as compared to groups that did not receive thrombolysis (83) (See also Table 2).

MRI is highly sensitive for the detection of blood degradation products. Hemorrhagic transformation was reported in 24 (68.6%) of 35 consecutive cardioembolic strokes studied by MRI at 3 weeks after the event, a percentage comparable to autopsy data (76).

The natural incidence of parenchymal hematoma in nonanticoagulated patients has not been clearly established. CT studies have reported a 2.0% to 8.6% incidence of parenchymal hematoma in nonanticoagulated patients with acute embolic stroke (52). The incidence of hemorrhagic stroke in untreated patients in the International Stroke Trial was only 0.4%; however, serial CTs were not routinely obtained (80). The incidence of spontaneous parenchymal hematoma or secondary bleeding severe enough to be symptomatic in control or placebo-treated patients in thrombolytic trials ranged from 0.6% to 6.5% (Table 1). Although the inclusion and exclusion criteria for these trials are rigid and, therefore, do not represent consecutive patients, the definitions for hemorrhagic transformation were clearly stated and carefully looked for by protocol.

Hemorrhagic transformation occurred in 30% of children with arterial ischemic stroke within 30 days. Most hemorrhages were petechial and asymptomatic (12).

Prevention

Reliable clinical and radiologic predictors are needed to identify those patients at highest risk for hemorrhagic transformation to guide the safe use of anticoagulants or thrombolytic therapy (07; 177).

Cardioembolic stroke mechanisms, large infarcts, occlusion of the middle cerebral artery stem, absence of collateral flow, hyperglycemia, and detection of early hypodense changes on CT may help predict hemorrhagic transformation (15; 60; 03; 55; 121).

A neuropathologic lab study evaluated 245 autopsies and found that age older than 75 years is a risk for hemorrhagic transformation in embolic infarcts. It also found that diabetes mellitus (but not serum glucose levels), along with infarct size greater than 10 cm3, are independent predictors of hemorrhagic transformation of ischemic stroke (92).

Embolic strokes, particularly those due to cardiogenic embolism, are particularly prone to undergoing hemorrhagic transformation (51; 66; 03). Large infarct size, the presence of mass effect, cerebral edema, and brain herniation all significantly increase the risk of secondary bleeding (27; 77; 136; 109; 31). Reports conflict on the contribution of hypertension, advanced age, and diabetes or hyperglycemia in promoting hemorrhagic transformation (48; 26; 15; 17). The detection of early hypodense changes on CT performed within hours of stroke onset may predict subsequent bleeding (182; 14; 166). Lower apparent diffusion coefficient values and persistent perfusion deficits demonstrated on diffusion/perfusion-weighted MRI imaging have been linked to increased risk for hemorrhagic transformation (165). Regional very low cerebral blood volume predicts hemorrhagic transformation after thrombolysis better than diffusion-weighted imaging volume and thresholded apparent diffusion coefficient (22; 101). Additionally, blood-brain barrier permeability assessed by perfusion CT scan may predict symptomatic hemorrhagic transformation (73; 110). Nevertheless, this finding awaits definite confirmation based on the results of a prospective trial in which hemorrhagic transformation (mainly hemorrhagic infarct type 2 as per ECASS criteria) was noted on neuroimaging follow-up in 10% of 545 patients with acute ischemic stroke treated with intravenous tPA or given intra-arterial treatment (thrombolysis or mechanical thrombectomy) (78). Increased blood-brain permeability assessed by extended perfusion CT scan on admission was associated with a 2-fold higher risk of hemorrhagic transformation on univariate analysis, although this finding was not corroborated in the multivariate analysis; only age and the NIHSS admission score were independent predictors of hemorrhagic transformation (78).

In a meta-analysis of 15 studies involving 1134 patients, perfusion CT showed moderate diagnostic performance in predicting hemorrhagic transformation in acute ischemic stroke (156). Five of the 15 studies had a prospective design, six used traditional perfusion CT parameters, and nine used permeability-related parameters. Most patients in this meta-analysis underwent intravenous thrombolysis alone; the second largest number underwent intraarterial thrombolysis, and a small proportion of patients received endovascular therapy. The reference standard at follow-up was noncontrast CT scan and MRI (gradient-echo T2-weighted sequences) between one and 14 days after the index stroke. This meta-analysis demonstrated that high blood-brain barrier permeability and hypoperfusion status derived from perfusion CT were associated with hemorrhagic transformation, with a pooled sensitivity of 84% and a pooled specificity of 74% for predicting hemorrhagic transformation in acute ischemic stroke (157).

Another meta-analysis also assessed the diagnostic accuracy of CT perfusion at admission compared to reference standard (non-contrast brain CT or MRI) in predicting hemorrhagic transformation of acute ischemic stroke (01). The overall hemorrhagic transformation rate was 30.2% among the 808 patients with acute ischemic stroke. Patients who underwent intravenous thrombolysis or endovascular therapy had a higher hemorrhagic transformation rate (37%) compared to patients who did not receive reperfusion therapies (24.6%), though the difference was not statistically significant. The mean time from stroke onset to CT perfusion was 2.5 hours, and from stroke onset to imaging reference standard it was 2.9 days. CT perfusion showed a pooled sensitivity of 85.9% and a specificity of 73.9% to predict hemorrhagic transformation (01).

The early use of anticoagulants has been associated with hemorrhagic transformation (29; 80). In contrast, several reports have not demonstrated clinical worsening associated with early anticoagulation, and widespread clinical usage persists. Proposed treatment guidelines for embolic stroke range from immediate anticoagulation to a delay of 2 or more weeks for large infarcts (27; 29; 30). These recommendations are largely empirical, and anticoagulation is generally delayed or avoided in patients with severe neurologic deficits, impaired level of consciousness, large infarcts with mass effect on CT, or uncontrolled severe hypertension (161). A study of 171 patients anticoagulated following acute stroke, including 83 patients with embolic hemispheric infarction treated within 72 hours after symptom onset, found that excessive prolongation of the activated partial thromboplastin time (greater than two times control) was the only significant factor associated with hemorrhagic worsening (31). The presence of large infarcts, severe clinical deficits, and age, although related to the frequency of hemorrhagic conversions, did not predict a higher risk of symptomatic bleeding. Patients with small or moderate cardioembolic infarcts who are at high risk for recurrence may be safely anticoagulated early or immediately if no hemorrhage is present on the initial CT (26; 30).

Predictors of hemorrhagic transformation associated with thrombolytic therapy have also been identified. The National Institute of Neurological Diseases and Stroke Trial reported an increased risk of hemorrhage associated with the presence of a severe initial neurologic deficit or the presence of mass effect on the initial CT scan (131; 132). Another study also indicated that elevated baseline serum glucose greater than 300 mg/dL was an independent predictor of hemorrhagic transformation in tPA-treated patients (39). An analysis of the European Cooperative Acute Stroke Study data also showed an association between the severity of initial neurologic deficit and the presence of early ischemic changes on CT with subsequent hemorrhagic infarction (107). Furthermore, blood pressure variability within the first 24 hours of an acute stroke can increase the risk of hemorrhagic transformation (184). In fact, post-thrombolysis blood pressure elevation and blood pressure variability more than the absolute level of blood pressure are associated with hemorrhagic transformation in acute ischemic stroke (18; 103). Analysis of data from the European Cooperative Acute Stroke Study II confirmed these findings and also suggested that increasing patient age, baseline systolic blood pressure, congestive heart failure, and treatment with aspirin before thrombolysis were predictors of hemorrhagic transformation (108; 44). Investigators have also been evaluating various neuroimaging techniques, including MR diffusion, perfusion, and CT perfusion scanning, in an attempt to assess hemorrhagic transformation risk after thrombolysis (151). Early infarct FLAIR hyperintensity seems associated with increased hemorrhagic transformation after thrombolysis (106). A study analyzing thrombolytic therapy with intravenous tPA and risks of hemorrhagic transformation discovered treatment time after 3 to 6 hours, a larger lesion volume, and a high National Institutes of Health Stroke Scale score on admission were significant independent predictors of hemorrhagic transformation (164; 84). Also, very low cerebral blood volume predicts parenchymal hematoma in acute ischemic stroke (68). Old age, however, was the main significant predictor of parenchymal hemorrhage after intravenous tPA (164). One study, however, emphasized that old age (older than 80 years old) does not increase the risk of symptomatic hemorrhagic transformation after administration of thrombolytic therapy; thus, old age should not deter one from administering thrombolytic therapy (145). Parenchymal hematoma has been known to have a more adverse outcome; it is predicted by larger lesions caused by cardioembolic disease, high blood glucose, or thrombolysis (137). A meta-analysis found that individual baseline variables were modestly associated with post-rtPA intracranial hemorrhages (178). Post-rtPA intracerebral hemorrhage was associated with higher age (odds ratio, 1.03 per year; 95% confidence interval, 1.01-1.04), higher stroke severity (odds ratio, 1.08 per National Institutes of Health Stroke Scale point; 95% confidence interval, 1.06-1.11), and higher glucose (odds ratio, 1.10 per mmol/L; 95% confidence interval, 1.05-1.14). There was approximately a doubling of the odds of intracerebral hemorrhage with the presence of atrial fibrillation, congestive heart failure, renal impairment, previous antiplatelet agents, leukoaraiosis, and a visible acute cerebral ischemic lesion on pretreatment brain imaging.

In the Third International Stroke Trial (IST-3), a multicenter randomized controlled trial of intravenous thrombolysis administered within 6 hours of stroke onset plus best medical care (1,507 patients) versus best medical care alone (1,510 patients), the presence of early ischemic changes and pre-existing structural lesions on pre-randomization brain scans (98% brain CT scans) were assessed to determine the increased risk of symptomatic intracerebral hemorrhage within 7 days after stroke (82). In this trial, 41% of patients had early ischemic changes, 51% had pre-existing structural lesions with no early ischemic changes, and 9% had normal brain imaging. Radiological findings increasing the risk of symptomatic intracranial hemorrhage were an old infarct (odds ratio 1.72) and two early ischemic signs: hyperattenuation of the cerebral artery (odds ratio 1.54) and tissue hypoattenuation (odds ratio 1.54). Moreover, when hyperattenuation of the cerebral artery and evidence of an old infarct were found on brain imaging in patients who underwent intravenous thrombolysis, the risk of symptomatic intracranial hemorrhage was nearly 3-fold that of patients with neither of these characteristics.

Because hemorrhagic transformation is related to endothelial damage after an ischemic stroke, albuminuria, a marker of chronic endothelial damage, can be used to predict hemorrhagic transformation (146). A study found that albuminuria is a significant independent predictor of hemorrhagic transformation, especially in the most severe hemorrhagic transformations, especially the parenchymal hemorrhage type 1 and 2 in patients who have had an acute ischemic stroke (146).

A study assessed 279 patients via MRI and evaluated whether cerebral microbleeds are related to early hemorrhagic transformation after thrombolytic therapy for hyperacute ischemic stroke. The results proved that microbleeds, whether few or many, were not independent risk factors for early hemorrhagic transformation of ischemic stroke or for any symptomatic hemorrhagic after thrombolytic therapy for hyperacute ischemic stroke (100). It was found that patients with cerebral microbleeds and acute cerebral infarct who underwent intravenous thrombolysis were two times more prone to early hemorrhagic transformation than those who had no microbleeds (128). Thus, 24% of the 366 patients with ischemic stroke in a retrospective cohort with a mean age of 67 years and a median NIHSS score of 6 showed hemorrhagic transformation on brain neuroimaging (MRI or CT scan) performed 18 to 36 hours after alteplase administration. Cerebral microbleeds defined as homogeneous round or oval foci demonstrating hypointensity on GRE sequences were observed in approximately a quarter of all patients. Hemorrhagic transformation occurred in a third of the patients with cerebral microbleeds. Lobar (cortical and subcortical white matter) microbleeds based on the BOMBS criteria were more frequently associated with hemorrhagic transformation than were deep microbleeds. Interestingly, deep and periventricular white matter hyperintensities on FLAIR sequences estimated by the Fazekas and Schmidt score were not associated with hemorrhagic transformation (128).

Predictors for intraarterial thrombolysis have also been evaluated. A study demonstrated that higher National Institutes of Health Stroke Scale scores, longer times to recanalize, lower platelet counts, lower coated-platelet levels, and higher glucose levels were independent predictors of hemorrhagic transformation after intraarterial thrombolysis (94; 144). It has also been found that after recanalization therapy for ischemic stroke, low LDL levels (independent of statin usage), smoking, leukoaraiosis, and an increase in stroke severity increased the risk of symptomatic hemorrhagic transformation (10; 152). Low level of LDL cholesterol increases hemorrhagic transformation in large artery atherothrombosis stroke but not in cardioembolic stroke (97; 138). A review of hemorrhagic transformation in the major clinical trials of acute ischemic stroke intervention demonstrates observable trends toward increasing hemorrhage rates with endovascular interventions and increased time lapses between stroke onset and vessel recanalization (158).

The IMS study published in 2006 found that atrial fibrillation and internal carotid artery occlusion, as opposed to middle cerebral artery occlusion, were significantly associated with hemorrhagic transformation of stroke.

A score predicting the risk of hemorrhagic transformation in patients with anticoagulated acute ischemic stroke has been validated. In fact, out of 241 acute ischemic stroke patients with a mean age of 67 years who were treated with anticoagulation (83% initiated during hospitalization), brain neuroimaging (ECASS II criteria) in nearly one third (75 patients) showed hemorrhagic transformation in the first week (mean of 4.6 days after admission), and 19 of these patients had symptomatic hemorrhagic transformation (119). The useful predictors of hemorrhagic transformation in this cohort of patients were older age, large infarct volume, renal impairment, higher serum glucose, leukocytosis, and warfarin use before admission (119).

The Hemorrhagic Transformation Index score has been proposed to predict hemorrhagic transformation within the first 14 days after an infarct involving the middle cerebral artery territory, regardless of tPA administration (91). This composite score, ranging from 0 to 8 points, comprises four items: ASPECTS score, NIHSS score, hyperdense MCA sign, and atrial fibrillation on ECG at admission. Hemorrhagic transformations (half of them symptomatic) occurred in 126 (23.6%) of 783 patients on CT follow-up within 14 days after stroke. The optimal cut-point to differentiate hemorrhagic transformation from no- hemorrhagic transformation was 2. In the validation cohort (248 patients), the ROC analysis showed an AUC of 0.83 with a sensitivity of 0.8 and specificity of 0.87 (91).

The spontaneous hemorrhagic acute ischemic stroke (SHAIS) score has been proposed as a useful tool for predicting spontaneous hemorrhagic transformation of cerebral infarct (176). In the 2021 study by Wei and colleagues, among a derivation cohort comprising 539 Chinese patients with a mean age of 68.1 years and a mean NIHSS score of six, 94% of cerebral infarcts occurred in the anterior circulation. During hospitalization, 17% (91) of these patients developed spontaneous hemorrhagic transformation (49 diagnosed by MRI and 42 by brain CT), and 25% of the hemorrhagic transformations were symptomatic. The score (ranging from 0 to 11 points) consists of five factors: atrial fibrillation, NIHSS score, hyperdense artery sign, hypodensity of MCA territory, and anterior circulation infarction (ACI). A score of 3 points as cut-off showed 81.3% and 83.3% of sensitivity and specificity, respectively, in predicting spontaneous hemorrhagic transformation. Compared with other preexisting models of hemorrhagic transformation after reperfusion therapy, the SHAIS score showed a good ability to predict hemorrhagic transformation (176). Results replication and validation in a large cohort of patients are required to confirm its value as a tool for predicting spontaneous hemorrhagic transformation.

The rate and predictors of hemorrhagic transformation occurring in the setting of acute ischemic stroke associated with tandem large-vessel occlusive lesions in the anterior circulation was estimated in 289 patients (186). At neuroimaging follow-up 24 hours after thrombectomy, hemorrhagic transformation was found in approximately one third of the patients (84 out of 289); hemorrhagic infarction type 2 (35 patients) was the most common pattern according to ECASS II criteria; and hemorrhagic infarction type 1 and parenchymal hematoma type 1 were observed in 31 and 24 patients, respectively, whereas parenchymal hematoma type 2 was the least common pattern (14 patients). Intracranial carotid occlusion, diabetes mellitus, no prior intravenous thrombolysis, and complete extracranial carotid occlusion were independent predictors of hemorrhagic infarction. The same above-mentioned predictors were associated with parenchymal hemorrhage transformation, with the addition of a higher baseline NIHSS score and an ASPECTS score lower than 7. Patients with parenchymal hematoma had a 2.5-fold higher mortality risk at 90 days than those with hemorrhagic infarction and those with no type of hemorrhagic transformation (186).

Differential diagnosis

Confusing conditions

Hemorrhagic transformation of cerebral infarction should be distinguished clinically from several other clinical entities: lobar hemorrhage as in amyloid angiopathy, subarachnoid hemorrhage from an arteriovenous malformation or aneurysm, or hypertensive hemorrhage. The radiographic findings in combination with the clinical history are useful in making these distinctions. Usually, hemorrhagic transformation of an ischemic infarct will present initially with radiographic findings consistent with ischemia rather than hemorrhage. The hemorrhage typically develops in a delayed fashion over the subsequent hours to days. The other entities mentioned above will present with hemorrhage on the initial head CT. However, when a stroke patient presents in a delayed fashion and has concomitant radiographic findings consistent with hemorrhage, it may be more difficult to discern the exact etiology of the hemorrhage. In fact, there is substantial observer variability in the differentiation between primary intracerebral hemorrhage and hemorrhagic transformation of infarction on CT brain imaging (115).

The detection of hemorrhagic transformation of an ischemic infarct raises the suspicion of an underlying embolic cause. Secondary bleeding, however, can occur with other stroke mechanisms, and causes unrelated to ischemia must also be considered. Cerebral venous thrombosis frequently results in hemorrhagic transformation ranging from petechial bleeding to massive hemorrhage. Strokes related to infective endocarditis are often accompanied by bleeding complications. Hematologic disorders, including thrombocytopenia, disseminated intravascular coagulation, and other coagulopathies, may all promote hemorrhagic infarction. Hemorrhagic infarction has been observed in strokes due to vasculitis and noninflammatory vasculopathies such as migraine and amyloid angiopathy. The realization that early hemorrhagic transformation may produce a CT pattern similar to a parenchymatous hematoma has led to the suggestion that primary intracerebral hemorrhage may be erroneously overdiagnosed if the initial CT is delayed or not performed (13).

Associated or underlying disorders

Detecting hemorrhagic transformation of an ischemic infarct raises the suspicion of an underlying embolic cause. Secondary bleeding, however, can occur with other stroke mechanisms and causes unrelated to ischemia must also be considered. Cerebral venous thrombosis frequently results in hemorrhagic transformation ranging from petechial bleeding to massive hemorrhage. Strokes related to infective endocarditis are often accompanied by bleeding complications. Hematologic disorders, including thrombocytopenia, disseminated intravascular coagulation, and other coagulopathies, may all promote hemorrhagic infarction. Hemorrhagic infarction has been observed in strokes due to vasculitis and noninflammatory vasculopathies such as migraine and amyloid angiopathy. The realization that early hemorrhagic transformation may produce a CT pattern similar to a parenchymatous hematoma has led to the suggestion that primary intracerebral hemorrhage may be erroneously overdiagnosed if the initial CT is delayed or not performed (13).

Diagnostic workup

CT and MRI neuroimaging have largely supplanted the role of autopsy and lumbar puncture for diagnosing hemorrhagic infarction (77). The recognition of hemorrhagic transformation can be made with a good degree of reliability by neuroradiologists and neurologists with CT training (122). Hemorrhagic transformation evolves in a dynamic fashion over time; therefore, serial CT examination will yield a higher incidence of bleeding than single or random radiologic studies. The accuracy of CT scanning for detecting small areas of hemorrhage remains uncertain (136). MRI is more sensitive than CT for detecting bleeding, especially when newer MRI techniques such as diffusion- and perfusion-weighted imaging are used (129; 165; 127). Recognizing hemorrhagic infarction may prompt diagnostic evaluation for an embolic stroke mechanism or other cause. Parenchymal enhancement on an MRI 2 hours after thrombolytic therapy can predict hemorrhagic transformation with high specificity (72).

Formerly, it had not been demonstrated whether MRI or CT was more accurate in detecting acute intracerebral hemorrhage in patients presenting with acute focal stroke symptoms. Kidwell and colleagues performed a study comparing the accuracy of MRI versus CT and discovered that gradient-recalled echo MRI was more sensitive in detecting hemorrhages than CT. Both CT and MRI could detect acute hemorrhages equally, but MRI was superior in detecting chronic hemorrhages (microbleeds) along with hemorrhagic transformation of ischemic stroke (93). CT could still, however, detect subarachnoid hemorrhage better than MRI (93).

Although it has been found that early parenchymal enhancement on MRI is specific for hemorrhagic transformation (170), diffusion-weighted imaging lesion volumes and apparent diffusion coefficient values are not strongly related to hemorrhagic transformation (99). It has been discussed, however, that diffusion-weighted imaging and susceptibility-weighted imaging can be used to reliably predict intracranial hemorrhage in times when CT imaging can be equivocal in the detection of bleeding after intraarterial thrombolysis (57).

In the last few months, a meta-analysis of nine trials covering 665 patients with acute stroke (most of whom received intravenous thrombolysis treatment) showed that MRI between 1 and 6 days after stroke had a moderate diagnostic performance (a pooled sensitivity and specificity of 82% and 79%, respectively) in predicting hemorrhagic transformation (ECASS criteria) of acute ischemic stroke. Sensitivity improved to 92% when a subset of six studies using advanced MRI techniques (perfusion and diffusion-weighted parameters) were considered, whereas specificity showed no change. Hemorrhagic transformation was associated with high blood-brain barrier permeability, low cerebral blood volume, low apparent diffusion coefficient, and FLAIR hyperintensity (157).

It has been discovered that hemorrhage after tPA treatment occurs in regions with a low apparent cerebral blood volume on bolus contrast MRIs; apparent cerebral blood volume provides better prediction of hemorrhagic transformation after tPA than apparent diffusion coefficient (06). A hyperintense MCA sign on post-Gd T1W1 MRI has been associated with a greater probability of hemorrhagic transformation (59).

Other tools, such as transcranial sonographic monitoring, can detect hemorrhagic transformation of stroke. A study noted that in 18 of 20 patients, transcranial sonographic monitoring confirmed hemorrhagic transformation, which had been detected by cranial CT. The sensitivity was 90%, and specificity was 97.5% (149).

In those patients treated with intraarterial thrombolysis after an acute ischemic episode, sulcal hyperintensity on FLAIR imaging is caused by iodinated contrast medium, not a subarachnoid hemorrhage. This hyperintensity, however, is significantly associated with subsequent hemorrhagic transformation (98).

A systematic January 1990 to November 2019 review of possible radiological markers for predicting hemorrhagic transformation emphasizes the value of different CT perfusion parameters (Evidence Level 1, systematic review or meta-analysis); nevertheless, there is no consensus to date on which CT parameter is the most sensitive and predictive of hemorrhagic transformation. Moreover, the current level of evidence suggests that any parameter of brain MRI, digital subtraction angiography, or transcranial Doppler sonography is superior to other parameters in predicting an increased risk of hemorrhagic transformation (46).

As the risk of hemorrhagic transformation increases with the advent of therapies such as rtPA to reperfused ischemic brain, the risk of reperfusion injury increases. Studies have been designed to measure this injury by measuring the blood-brain barrier disruption evidenced as delayed gadolinium enhancement of CSF on FLAIR images. This has been termed as HARM or hyperintense reperfusion marker. HARM has been found in 33% of patients with ischemic stroke as reperfusion is a strong independent predictor of early blood-brain barrier disruption (86). HARM has also been associated with hemorrhagic transformation and worsening clinical outcomes along with more severe strokes at onset and with older age. HARM is timed early enough (approximately 3.8 hours from the onset of stroke) so that it can help reduce complications of acute thrombolytic therapy (174).

Management

Managing hemorrhagic infarction is guided by the same principles of general and supportive care as nonhemorrhagic strokes. The most recent American Heart Association/American Stroke Association guidelines for acute ischemic stroke management recommend carefully lowering blood pressure to less than 185/105 mm Hg during tPA and in the first hours after intravenous rt-PA administration and mechanical thrombectomy to reduce the risk of hemorrhagic transformation (143; 64). Cerebral edema may be managed in the standard manner, with osmotic diuretics, hypertonic saline, increasing the head of the bed, hyperventilation, and limitation of excessive fluid resuscitation. Anticoagulation must be reversed or discontinued in patients with symptomatic hemorrhagic transformation. The safety of continued anticoagulation to prevent recurrent embolism in clinically stable patients after detecting hemorrhagic transformation on neuroimaging remains uncertain. The latest U.S. guidelines for secondary prevention recommend starting oral anticoagulation at least 14 days after index stroke in patients with atrial fibrillation considered at high risk of hemorrhagic conversion (102). The timeframe to initiate or restart oral anticoagulation may depend on the type of hemorrhagic infarct and the patient’s clinical stability. In patients with small or confluent petechial hemorrhages, a delay of at least 6 days may be reasonable before initiating oral anticoagulation; for patients with homogeneous hemorrhage occupying less than 30% of infarcted tissue, the suggested delay for anticoagulation treatment is at least 12 days, and for patients with hemorrhage occupying more than 30% of infarcted tissue, a delay between 12 and 28 days is recommended (154).

The Stroke Council of the American Heart Association has published guidelines for managing hemorrhagic complications of intravenous thrombolytic therapy (181).

One study suggests that the use of minocycline with TPA can extend TPA time windows in treating ischemic stroke (126). The study used rats to discover that late TPA use (within 6 hours) increases hemorrhagic transformation, but with the use of minocycline, this hemorrhagic transformation could be ameliorated. Protease-activated receptor-1 (PAR-1) agents given along with thrombolysis or thrombectomy are promising therapeutic targets to reduce the rate of hemorrhagic transformation according to results observed in a phase 2 trial (116).

Special considerations

Pregnancy

Pregnancy increases the risk of stroke up to 13-fold over that of nonpregnant age-matched controls. Although no data specifically address the issue of hemorrhagic infarction during pregnancy, it is reasonable to expect that stroke mechanisms unique to pregnancy may be complicated by hemorrhagic transformation (179). Cardioembolic strokes during pregnancy may result from peripartum cardiomyopathy. Paradoxical emboli occur more frequently during pregnancy, particularly in the third trimester. The early postpartum period is associated with a hypercoagulable state that may lead to stroke. Amniotic fluid or air embolism, though infrequent, also occurs during pregnancy and the puerperium. Hemorrhagic infarction secondary to intracranial venous thrombosis is a well-recognized cause of pregnancy-related stroke and typically occurs in the first few weeks postpartum. Metastatic choriocarcinoma may be present with a hemorrhagic infarction.

Consideration must be given to other causes of pregnancy-related intracerebral hemorrhage presenting in a stroke-like fashion. Eclampsia and preeclampsia may produce cortical petechiae, subcortical hemorrhage, and intracerebral and subarachnoid hemorrhage (42). Pregnancy is also associated with an increased risk of aneurysmal subarachnoid hemorrhage and bleeding secondary to an arteriovenous malformation.

Anesthesia

In patients who undergo carotid endarterectomy, either due symptomatic carotid disease or due to severe internal carotid artery lesions, two types of anesthesia are commonly used: general and cervical. Fiorani and colleagues found that general anesthesia can result in a higher stroke rate during the operation than cervical anesthesia (49). The cardiac complication rate was similar in both types of anesthesia (49). It was also found that cervical block anesthesia has better results as it can involve more reliable cerebral monitoring and reduces the need for internal carotid artery shunting.