Management of Thrombolysis-Associated Symptomatic Intracerebral Hemorrhage

Abstract

Background

Symptomatic intracerebral hemorrhage (sICH) is the most devastating complication of thrombolytic therapy for acute stroke. It is not clear whether patients with sICH continue to bleed after diagnosis, nor has the most appropriate treatment been determined.

Methods

We performed a retrospective analysis of our prospectively collected Get With the Guidelines–Stroke database between April 1, 2003, and December 31, 2007. Radiologic images and all procoagulant agents used were reviewed. Multivariable logistic regression was performed to identify factors associated with in-hospital mortality.

Results

Of 2362 patients with acute ischemic stroke during the study period, sICH occurred in 19 of the 311 patients (6.1%) who received intravenous tissue plasminogen activator and 2 of the 72 (2.8%) who received intra-arterial thrombolysis. In-hospital mortality was significantly higher in patients with sICH than in those without (15 of 20 [75.0]% vs 56 of 332 [16.9%], P<.001). Eleven of 20 patients (55.0%) received therapy for co-agulopathy: 7 received fresh frozen plasma; 5, cryoprecipitate; 4, phytonadione (vitamin K₁); 3, platelets; and 1, aminocaproic acid. Independent predictors of inhospital mortality included sICH (odds ratio, 32.6; 95% confidence interval, 8.8–120.2), increasing National Institutes of Health Stroke Scale score (1.2; 1.1–1.2), older age (1.3; 1.0–1.7), and intra-arterial thrombolysis (2.9; 1.4–6.0). Treatment for coagulopathy was not associated with outcome. Continued bleeding (>33% increase in intracerebral hemorrhage volume) occurred in 4 of 10 patients with follow-up scans available (40.0%).

Conclusions

In many patients with sICH after thrombolysis, coagulopathy goes untreated. Our finding of continued bleeding after diagnosis in 40.0% of patients suggests a powerful opportunity for intervention. A multicenter registry to analyze management of thrombolysis-associated intracerebral hemorrhage and outcomes is warranted.

Acute ischemic stroke causes substantial morbidity and is one of the leading causes of death worldwide.¹ The only current pharmacotherapy approved by the US Food and Drug Administration is intravenous recombinant tissue plasminogen activator (tPA). Thrombolytic therapies can also be delivered intra-arterially,² and this method of therapy is commonly used off-label.³ Several clinical trials have found that thrombolytic therapy decreases morbidity and improves long-term outcome.^4–7 The American Heart Association has recommended that systems of care be established to maximize the ability of all patients with stroke to receive the best available therapy, including thrombolytic agents.⁸ In fact, it appears that the use of thrombolytic agents is increasing⁹ and may continue to increase because recent evidence suggests an expanded time window for intravenous thrombolysis.⁴

The most feared complication of this therapy is symptomatic intracerebral hemorrhage (sICH).¹⁰ Although the overall benefits of tPA for stroke appear to outweigh the risks,^4,5,11 up to 7% of patients treated for stroke will develop this complication. The development of sICH is associated with worse outcome,^12–14 and clinicians caring for these patients are faced with the difficult decision of how best to treat them.

There are no evidence-based guidelines that address management of thrombolysis-associated sICH. This is likely because of the lack of published original research examining management. The American Heart Association suggests empirical therapies to replace clotting factors and platelets but acknowledges the lack of evidence to support any specific therapy.¹⁵ Even in the absence of supporting evidence, some institutions (including ours) have developed care pathways for thrombolysis-associated sICH for current use.¹⁶ At the moment, the most appropriate management of this complication is not clear.

To determine best practice, it is first necessary to examine current practice and to ascertain whether any specific factors or therapies are associated with outcome. We performed a retrospective review of current sICH management to determine the frequency of ongoing bleeding and patterns of emergency therapy.

METHODS

PATIENT SELECTION AND DATA COLLECTION

We performed a retrospective analysis of a prospectively collected cohort of consecutive patients with acute ischemic stroke presenting to a single urban tertiary care hospital between April 1, 2003, and December 31, 2007. Data from all patients at our institution who are diagnosed as having ischemic stroke are entered into our local Get With the Guidelines—Stroke database. The details of this database have been described in previous publications.^17,18

Demographic data including medical history of atrial fibrillation, diabetes mellitus, hypertension, and coronary artery disease; laboratory values; initial National Institutes of Health Stroke Scale score (NIHSS); whether the patient presented primarily or was transferred in; any therapy (including intravenous tPA, intra-arterial thrombolysis [IAT], and mechanical thrombectomy); any development of intracerebral hemorrhage (ICH); and in-hospital mortality were prospectively ascertained. Patients receiving thrombolytic therapy were treated in our neurosciences intensive care unit for the first 24 hours. (Our standard protocols are available at http://www.stopstroke.org.) Platelet function tests were not routinely performed. Clinical course and radiographic images for all patients were reviewed by our hospital’s Acute Stroke Quality Task Force as part of a required ongoing quality-assurance review (members of the task force included J.N.G., S.M., M.P., A.A., E.E.S., and L.H.S.). Symptomatic ICH was defined using National Institute of Neurological Disorders and Stroke (NINDS) criteria as the presence of intracranial blood in association with neurologic deterioration⁵; although this definition may include patients for whom deterioration was not due to hemorrhage, it has a high inter-rater reliability and is the most inclusive. The presence or absence of ICH and the clinical assessment of symptomatic worsening by NINDS criteria were determined through consensus review by this group of local experts.

A separate retrospective review of electronic and hospital records was performed for all patients with sICH. Serial laboratory values, including coagulation measures, fibrinogen levels, and all procoagulant therapies, were recorded. Our hospital uses a computerized order entry and laboratory reporting system, and all computer and laboratory orders were retrieved electronically and reviewed; nursing notes in the medical records were then reviewed to confirm interventions that the patients received. Radiology images were electronically transferred in Digital Imaging and Communications in Medicine (DICOM) standard format to a dedicated workstation for analysis with the use of commercial image-analysis software (Alice; PAREXEL International, Waltham, Massachusetts) and reviewed by study staff blinded to the patient’s clinical outcomes.¹⁹ For the purposes of this analysis, intracerebral hemorrhage volumes were measured as the sum total of all intracranial blood, irrespective of type or location, by means of computerized planimetry.

DATA ANALYSIS

Continuous variables were analyzed with the Kruskal-Wallis test because none demonstrated a normal distribution as determined by the Shapiro-Wilk test. Dichotomous variables were analyzed with the Fisher exact test. Multivariable analysis for the outcomes of sICH and in-hospital mortality was performed with a backward-selection logistic regression model, including variables associated (P<.20) in univariate analysis and then removing variables in a stepwise fashion for P>.10. Given the small number of outcomes, no interaction terms were included. All analyses were performed in Stata software, version 10 (StataCorp, College Station, Texas).

RESULTS

During the study period, 2362 patients presented with acute ischemic stroke, of whom 352 (14.9%) received a thrombolytic agent: intravenous tPA only (n=282), IAT only (n=41), or both (n=29). Of the 352 patients, sICH developed in 20 (5.7%). Their demographic and clinical features are shown in Table 1.

Table 1.

Characteristics of 352 Patients Who Received a Thrombolytic Intervention

Characteristic	No sICH (n = 332)	sICH (n = 20)	P Value
Age, median (IQR), y	74 (62–82)	80 (71–88)	.06
Female sex, No. (%)	176 (53.0)	9 (45.0)	.50
White race, No. (%)	306 (92.2)	18 (90.0)	.67
Medical history, No. (%)
Atrial fibrillation	92 (27.7)	7 (35.0)	.45
Diabetes mellitus	65 (19.6)	4 (20.0)	>.99
Hypertension	221 (66.6)	14 (70.0)	>.99
Coronary artery disease	80 (24.1)	8 (40.0)	.12
Smoker	58 (17.5)	1 (5.0)	.22
Initial NIHSS score, median (IQR)	14 (9–20)	16 (9–20)	.41
In-hospital mortality, No. (%)	55 (16.6)	15 (75.0)	<.001

Abbreviations: IQR, interquartile range; NIHSS, National Institutes of Health Stroke Scale; sICH, symptomatic intracerebral hemorrhage.

The risk of developing sICH in this cohort was stratified by the intervention received, and the results are shown in Table 2. None of the interventions tested were significantly associated with sICH in univariate analysis. Multivariable analysis did not identify any independent predictors of sICH. Multivariable analysis of in-hospital mortality is shown in Table 3. The use of do-not-resuscitate orders was not included in the model because it was so highly covariate with death; all deaths but 1 were preceded by a do-not-resuscitate order, and only 1 patient had such an order but did not die in the hospital. (That patient was discharged to hospice care at home.)

Table 2.

Frequency of sICH Stratified by Therapy Received

Therapy	No. (%)
	No. of Patients (N = 352)	Frequency of sICH (n = 20/352 [5.7%])a
Any use of IV tPA	311 (88.4)	19/311 (6.1)
Any use of IAT	70 (19.9)	2/70 (2.9)
Any mechanical device	76 (21.6)	4/76 (5.3)
IV tPA given at outside hospital	177 (50.3)	11/177 (6.2)
IV but no IAT	282 (80.1)	18/282 (6.4)
IAT but no IV	41 (11.6)	1/41 (2.4)
Combined IV and IAT	29 (8.2)	1/29 (3.4)

Abbreviations: IAT, intra-arterial thrombolysis; IV, intravenous; sICH, symptomatic intracerebral hemorrhage; tPA, tissue plasminogen activator.

^aThe frequency of sICH is shown as a proportion of patients receiving each therapy.

Table 3.

Multivariable Analysis of Predictors of In-Hospital Mortality

Variable	OR (95% CI)	P Value
sICH	32.6 (8.8–120.2)	<.001
Any IAT	2.9 (1.4–6.0)	.004
Age, per 10-y increase	1.3 (1.0–1.7)	.02
NIHSS score, per unit increase	1.2 (1.1–1.2)	<.001

Abbreviations: CI, confidence interval; IAT, intra-arterial thrombolysis; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; sICH, symptomatic intracerebral hemorrhage.

We next examined how patients with sICH were treated. Table 4 shows the 20 patients with sICH and their treatment. A fibrinogen level was measured in 14 patients (mean [SD] level, 264 [86] mg/dL) (to convert fibrinogen to micromoles per liter, multiply by 0.0294), and no patients had a level less than 100 mg/dL. D-dimer was measured in 12 patients (mean level, 5.1 [2.7] μg/mL) (to convert to nanomoles per liter, multiply by 5.476). Eleven patients (55.0%) received some type of procoagulant therapy: 7 (35.0%) received fresh frozen plasma, 5 (25.0%) received cryoprecipitate, 4 (20.0%) received phytonadione (vitamin K₁), 3 (15.0%) received platelets, and 1 (5.0%) received aminocaproic acid. Five patients received more than 1 type of procoagulant therapy. No patient received recombinant factor VIIa. The use of procoagulant therapy was not significantly associated with any patient characteristic or outcome.

Table 4.

Clinical Management of sICH

Patient Age, y/Sex	Intervention	Laboratory Values		Therapy	ICH Volume, mL	Change in Volume, %	Outcome
		D-Dimer, μg/mL	Fibrinogen, mg/dL
89/M	IV	8.6	417	Phytonadionea	31.3	36.9	Death
88/M	IV	NM	NM	None	153.8	ND	Death
80/F	IV	NM	NM	None	105.8	ND	Death
56/M	IV	6.6	288	None	15.7	64.1	Death
78/M	IAT	NM	286	FFPb	140.3	1.4	Death
91/M	IV	2.8	132	None	103.6	ND	Death
77/M	IV	NM	377	Cryo	81.1	ND	Rehab
53/M	IV	3.0	249	FFP, cryo, and platelets	9.0	−1.0	Rehab
71/M	IV	7.3	291	FFP, cryo, and platelets	53.7	39.7	Death
92/F	IV	NM	NM	Phytonadionea	9.3	6.3	Death
72/M	IV	4.5	284	FFP and phytonadionea,b	43.7	ND	Death
86/M	IV	NM	194	None	8.8	ND	Death
89/F	IV	2.8	301	Cryo	9.8	45.7	Death
75/F	IV	8.2	282	FFP, cryo, aminocaproic acid, and platelets	131.6	ND	Death
81/F	IV	8.6	315	None	169.0	6.8	Death
79/F	IV	NM	NM	Noneb	9.5	ND	Rehab
84/F	IV	5.2	164	None	41.1	−2.4	Home
89/F	IV+IAT	NM	NM	FFP and phytonadionea	132.4	ND	Death
24/F	IV	1.7		None	8.8	−1.1	Death
58/M	IV	1.7	115	FFPb	49.3	ND	Rehab

Abbreviations: cryo, cryoprecipitate; FFP, fresh frozen plasma; IAT, intra-arterial thrombolysis; ICH, intracerebral hemorrhage; IV, intravenous thrombolysis; ND, follow-up computed tomography not done; NM, not measured; Rehab, rehabilitation; sICH, symptomatic ICH.

SI conversion factors: To convert D-dimer to nanomoles per liter, multiply by 5.476; fibrinogen to micromoles per liter, multiply by 0.0294.

^aVitamin K₁.

^bThese patients were receiving warfarin sodium before their stroke but were still considered candidates for thrombolysis depending on the initial international normalized ratio as per institutional protocol.

To evaluate the frequency of hematoma expansion or further intracranial bleeding after diagnosis of sICH, serial computed tomographic (CT) scans were evaluated for total ICH volume (Figure). The initial ICH volume was a mean (SD) of 65.3 (56.2) mL. Of 10 patients with follow-up scans available for analysis (performed a median of 8.8 [IQR, 7–14] hours after the initial scan), 4 patients (40.0%) had an increase in ICH volume of greater than 33% (Table 4). Of these patients, the mean (SD) percentage of change in ICH volume was 47% (12%). There were too few events to statistically analyze any effect of hemostatic therapy on ICH expansion.

Figure.

Ongoing bleeding after diagnosis in a patient with thrombolysis-associated symptomatic intracerebral hemorrhage. A, A 56-year-old man with acute left-sided weakness and dysarthria underwent computed tomography (CT) 1.5 hours after symptom onset and received intravenous tissue plasminogen activator 35 minutes later. B, Follow-up CT 11 hours later showed hemorrhagic conversion of a large right frontoparietal ischemic infarct, along with small foci of subarachnoid hemorrhage along the right cerebral convexity. C, Follow-up CT 8 hours later showed increased hemorrhage with increasing edema and mass effect.

COMMENT

To our knowledge, this is the first published analysis of current therapies used for thrombolytic agent reversal following sICH. Overall, we found that different strategies were used in clinical practice to treat thrombolysis-associated sICH. For many patients, no procoagulant therapy was provided, whereas others received a single therapy or a combination of multiple therapies. Very few patients were treated the same way. This heterogeneity likely reflects the absence of evidence supporting any specific intervention. The low frequency of sICH probably contributes to this dearth of data.

Our finding that 40.0% of patients with follow-up CT scans showed evidence of ICH expansion and ongoing bleeding suggests a potential window of opportunity for therapy. Reducing early hematoma expansion after sICH might improve clinical outcomes. A similar frequency of hematoma expansion has been noted following spontaneous, nonthrombolysis-associated ICH,²⁰ and clinical trials for this entity have demonstrated that hemostatic therapy reduces the risk of hematoma expansion.^21,22 Of note, patients with any coagulopathy were excluded from these trials. Patients who are coagulopathic because of receipt of a thrombolytic agent may be more likely to receive clinical benefit from hemostatic therapy.

The choice of hemostatic therapy remains controversial. Thrombolytic agents convert plasminogen to plasmin, which degrades fibrin at the site of thrombus formation.²³ Antifibrinolytics such as aminocaproic acid (Amicar) seem to be a logical antidote to fibrinolytic therapy and have been proposed as an aggressive measure for limiting intracranial bleeding.^24–27 Only 1 of our patients received this; it is not specifically recommended in the American Heart Association guidelines or our own internal guidelines because of the risk of inadvertent prothrombotic activity.¹⁵ In addition, if tPA induces systemic fibrinogenolysis,²⁸ fibrinogen replacement in the form of fresh frozen plasma or cryoprecipitate would be a logical choice for patients with low fibrinogen levels.²⁹ However, we found no cases in which patients developed fibrinogen levels of less than 100 mg/dL, suggesting that currently used tPA dosing regimens are unlikely to induce hypofibrinogenemia, and any benefit of such therapy may be limited. Because clot lysis releases D-dimers, which can exert an antiplatelet effect by binding to the platelet fibrinogen receptor, there is a theoretical basis for infusion of platelets as a rescue therapy.²⁹ To our knowledge, this hypothesis has not yet been tested. Finally, thrombolytic agents may affect a number of processes independent of their effects on clot disruption, including extracellular matrix degradation and cell signaling.²³ Interventions aimed at reversing these noncoagulopathic effects may offer promising avenues of research.

It is also possible that no currently available intervention provides benefit. Many of the therapies used in practice or that are recommended by guidelines are not clearly specific for reversal of fibrinolytic activity. Given the short half-life of thrombolytic agents, by the time the diagnosis is made the biological effect of the drug may have abated. In addition, not all intracranial bleeding after stroke can be ascribed to tPA, and it may be that only a subset of those with sICH will benefit from procoagulant therapy. Nevertheless, our finding that ongoing bleeding is common suggests that a therapeutic opportunity exists. Given the low frequency of sICH after thrombolysis, large multicenter prospective trials will be required to determine which therapy can best arrest or prevent ongoing bleeding.

Our analysis has several limitations. First, only a small number of individuals developed sICH after thrombolytic therapy. Second, the wide variation in therapy made it impossible to perform robust statistical analysis examining the effect of any specific therapy. Similarly, the wide variation in workup performed limited our ability to examine whether laboratory values such as D-dimer or fibrinogen level could predict development of sICH or hemorrhage expansion following sICH. Third, our single-center experience may not reflect practice at other institutions nationwide. Fourth, measurement of hematoma volume was difficult owing to the variation in types of hemorrhage and the frequent use of contrast medium, which can mimic blood, for CT-angiography and interventional procedures. Fifth, not all patients received follow-up CT, and the timing of follow-up CT was not standardized; this could reflect withdrawal of care or patients whose clinical course was so benign that follow-up studies were not obtained. As a result, our estimates of the frequency of hematoma expansion were based on a small sample and may be biased. Future studies in which follow-up CT is systematically performed would provide more accurate estimates. Finally, we were not able to systematically obtain long-term outcome information on patients in this cohort, limiting our outcome evaluation to the time of hospital discharge.

In conclusion, we found that for many patients with sICH after thrombolysis, no acute reversal of coagulopathy is attempted. Even among those treated, no particular therapy is associated with improved survival. The finding that continued bleeding occurs after diagnosis suggests an opportunity for intervention; it remains to be determined whether any currently available therapy can meet this need or whether novel treatments should be developed. As a next step, a multicenter registry of current management of thrombolysis-associated ICH and outcomes appears warranted.