PREDICTORS OF HIGHLY PREVALENT BRAIN ISCHEMIA IN INTRACEREBRAL HEMORRHAGE

Abstract

Objective

To determine the prevalence, characteristics, risk factors and temporal profile of concurrent ischemic lesions in patients with acute primary intracerebral hemorrhage (ICH).

Methods

Patients were recruited within a prospective, longitudinal, magnetic resonance imaging (MRI) based study of primary ICH. Clinical, demographic, and MRI data were collected on all subjects at baseline and 1 month.

Results

Of the 138 patients enrolled, mean age was 59 years, 54% were male, 73% black, and 84% had a history of hypertension. At baseline, ischemic lesions on diffusion-weighted imaging (DWI) were found in 35% of patients. At 1 month, lesions were present in 27%, and of these lesions, 83% were new and not present at baseline. ICH volume (p=0.025), intraventricular hemorrhage (p=0.019), presence of microbleeds (p=0.024), and large, early reductions in mean arterial pressure (p=0.003) were independent predictors of baseline DWI lesions. A multivariate logistical model predicting the presence of 1 month DWI lesions included history of any prior stroke (p=0.012), presence of 1 or more microbleeds (p=0.04), black race (p=0.641), and presence of a DWI lesion at baseline (p=0.007)

Interpretation

This study demonstrates that more than 1/3 of patients with primary ICH have active cerebral ischemia at baseline remote from the index hematoma, and 1/4 of patients experience ongoing, acute ischemic events at 1 month. Multivariate analyses implicate blood pressure reductions in the setting of an active vasculopathy as a potential underlying mechanism. Further studies are needed to determine the impact of these lesions on outcome and optimal management strategies to arrest vascular damage.

INTRODUCTION

Intracerebral hemorrhage (ICH) is a devastating disorder associated with substantial morbidity and mortality.¹ The often dismal outcome for patients with ICH highlights the need to better understand the underlying complex pathophysiology of this disease process, which in turn can assist in accurate determination of future stroke risk, long-term outcome assessment, and perhaps most importantly, development of treatment interventions.

Advanced MRI techniques have become increasingly valuable in characterizing and refining our understanding of ICH pathophysiology.² In addition to demonstrating acute hemorrhage, MRI can identify leukoaraiosis, chronic hematomas, and microbleeds, thus providing insight into the etiology and severity of the underlying disease process. Several recent MRI studies have identified a high prevalence of topographically distant ischemic lesions identified on diffusion-weighted imaging (DWI) in the setting of primary ICH.^3–5 However, the underlying mechanism, timecourse and significance of these lesions is poorly understood.

This study seeks to determine the prevalence, characteristics, risk factors, and temporal evolution of silent ischemic lesions in a predominantly black, hypertensive population of patients with primary ICH.

PATIENTS AND METHODS

The current analysis was undertaken as part of an ongoing, prospective, longitudinal MRI study of racial differences in primary ICH (DiffErenCes in the Imaging of Primary Hemorrhage based on Ethnicity or Race: DECIPHER). Since 2007, consecutive patients have been recruited from the inpatient services of 5 Washington, DC metropolitan-area hospitals. Inclusion criteria are: age 18 years or greater, diagnosis of primary ICH (not limited to first ICH), and brain MRI within 1 month of symptom onset. Exclusion criteria are: contraindication to MRI, pregnancy, central nervous system (CNS) tumor/active inflammatory process, CNS arteriovenous malformation/aneurysm, CNS trauma within prior 2 weeks, craniotomy, or international normalized ratio >3. Baseline demographic information, medical history, medications, neurologic assessments and laboratory data are collected on all patients. Blood pressures (BP) measures are collected for the following timepoints: admission, highest BP prior to MRI, lowest BP prior to MRI, and 1 month. Differences in mean arterial pressures (delta MAPs) are calculated for the various time intervals. Results of cardiac evaluations including EKGs and echocardiograms are reviewed in combination with cardiac history such that each patient is categorized as having a high, low or no risk cardioembolic source.

As part of the DECIPHER protocol, MRIs and outcome assessments are performed at admission (baseline), 1 month, 1 year and 3 years. Baseline MRIs are acquired on either 1.5 T or 3.0 T MR scanners. All follow-up MR scans are preferentially acquired on a 3.0 T scanner. A standardized protocol is employed for all timepoints that includes DWI and apparent diffusion coefficient (ADC), fluid attenuated inversion recovery (FLAIR), and gradient recalled echo (GRE) sequences.

The DECIPHER study is being performed with approval of the Institutional Review Boards (IRBs) of the admitting hospitals and Georgetown University which serves as the IRB of record.

Imaging analyses

For the current analysis, three investigators (R.M., R.B., C.K.) performed the imaging evaluations. An initial series of 15 cases were analyzed as consensus reads to establish interrater reliability. All imaging analyses were performed using Mango (Multi-Image Analysis GUI, http://ric.uthscsa.edu/mango/). GRE sequences were evaluated for the following: hematoma volume (using a semi-automated segmentation tool) and location; number and location of microbleeds and chronic hematomas; and presence of intraventricular hemorrhage (IVH). Microbleeds were defined as rounded foci of hypointense signal on GRE sequences within the brain parenchyma.⁶

DWI/ADC sequences were evaluated to characterize the presence, number, location, volume and acuity of ischemic lesions topographically remote from the hematoma. DWI hyperintensities immediately adjacent to the hematoma were not included in this analysis. A comparison of equivalent axial slices from the two timepoints was performed for each DWI lesion visualized on the 1 month scan to determine 1) if it was new (or persistent from the baseline timepoint), and 2) if new, whether it was acute (low ADC value), or subacute (normal or elevated ADC value). Lateralized assessment of the topography and severity of leukoaraiosis was assessed using the Fazekas Scale⁷ for periventricular (PV) and deep white matter (DWM) regions. A sum score of white matter disease (WMD) was calculated as the sum of the PV and DWM scores.

Statistical methods

Differences in dichotomous variables were analyzed using chi-square analysis or Fisher’s exact test. The student’s T test or the Wilcoxon rank sum test was used to analyze differences in the mean or median of continuous variables between groups. Variables that indicated a univariate relationship with lesion presence (p<0.1) and were not strongly correlated with one another were considered for multivariable logistic modeling. All statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, N.C.).

RESULTS

During the study timeframe, 138 of 452 (31%) patients meeting study eligibility criteria were enrolled. Of these, 119 subjects had a baseline MRI, 113 subjects had a 1 month MRI, and 94 subjects had imaging at both baseline and 1 month time-points. For the overall cohort, mean age was 59 years, 73% of patients were black, 54% male, 84% had a history of hypertension, and median National Institutes of Health Stroke Scale⁸ (NIHSS) score was 5 (mean 9). On imaging, mean hematoma volume was 25 mL (median 14 mL, range 0.4–136.4 mL), and 74% of hematomas originated in deep structures. The underlying etiology of the hemorrhage was determined to be hypertension in 62% of patients, CAA in 1.5%, unknown (no risk factors) in 3.6% with the remainder due to hypertension plus 1 or more additional risk factors. Median time to baseline MRI was 2 days, and to 1 month MRI was 35 days.

At the baseline timepoint, 35% of patients had one or more DWI lesions (median 2, range 1–67), for a total of 227 lesions in 42/119 patients. At the 1 month timepoint, 27% of patients had one or more DWI lesions (median 2, range 1–7) for a total of 67 lesions in 30/113 patients. Lesions were generally small at both timepoints: median volume for baseline lesions was 0.5 mL (range, 0.04–16 mL), and for 1 month lesions was 0.4 mL (range, 0.01–14 mL). Figure 1 provides an example of a patient with multiple baseline DWI lesions and Figure 2 shows an example of a patient with a new acute lesion visualized at 1 month and not present at baseline.

Figure 1.

Figure 2.

To clarify the evolution, acuity, and timecourse of the ischemic lesions, an analysis was performed on the subset of patients with MRIs at both baseline and 1 month timepoints. In this subset, there were 158 lesions at baseline, and of these, 78% were identified as acute (low ADC), and 22% subacute (iso- or hyperintense ADC). At 1 month, there were a total of 40 lesions, and of these 58% were acute and 42% were subacute. In total, 87% of patients with lesions at 1 month had 1 or more new DWI lesions compared to baseline, and 83% of all lesions at 1 month were new compared to baseline. Of these new lesions, 64% were acute and 36% subacute.

Additional analyses were performed to further characterize the topography and laterality of the DWI lesions. At baseline, 44% of the DWI lesions were located in lobar locations, whereas at 1 month, only 13% were lobar. At both the baseline and 1 month timepoints, approximately 1/3 of patients had DWI lesions ipsilateral to the primary hematoma, 1/3 contralateral, and 1/3 had bilateral lesions (Figure 3).

Figure 3.

Of the 138 patients, 49 (35.5%) patients had either a transthoracic or transesophageal echocardiogram performed. Only 9 patients were deemed to have a high risk cardioembolic source, and this did not differ significantly between those with DWI lesions versus without (p=0.179). Thirty-two patients (23%) had diagnostic cerebral angiograms performed, with no difference in the frequency of angiography between those with DWI lesions versus without (baseline p=0.134, 1 month p=0.325).

Univariate analyses were performed for baseline and 1 month, respectively, to identify differences in characteristics for patients with 1 or more ischemic DWI lesions versus those without. Table 1 shows the results for the baseline timepoint. In contrast to patients without baseline lesions, patients with lesions were significantly more likely to have a lobar ICH (p=0.027), larger ICH volume (p=0.033), microbleeds (p=0.015), IVH (p=0.002), greater mean delta MAP (p<0.001), and were more likely to be treated with antihypertensive agents during the acute hospital stay (p=0.031). In multivariate logistical modeling, hematoma volume, IVH, microbleeds, and delta MAP remained independent predictors of the presence of baseline DWI lesions (Table 2).

Table 1.

Patient characteristics stratified by presence of baseline DWI lesions

	All Subjects (n=119)	DWI Positive (n=42)	DWI Negative (n=77)	p Value
Age, mean (SD)	59.8 (13.1)	60.9 (10.6)	59.1 (14.3)	0.445
Black race, n (%)	87 (73.1)	32 (76.2)	54 (71.1)	0.575
Male gender, n (%)	65 (54.6)	26 (61.9)	38 (50.0)	0.239
Hypertension (y/n), n (%)	100 (84.0)	36 (85.7)	63 (82.9)	0.573
Delta MAP (mm Hg), mean (SD)	40.2 (24.1)	50.8 (25.0)	34.4 (21.7)	<0.001
Acute in-hospital antihypertensive treatment, n (%)†	101 (88.6%)	39 (97.5%)	62 (83.8%)	0.031
NIHSS, median (range)	5 (0–40)	7.5 (0–40)	4 (0–35)	0.136
Prior stroke (any type), n (%)	33 (27.7)	15 (35.7)	18 (23.7)	0.151
High risk CES, n (%)	9 (7.6)	4 (9.5)	5 (6.5)	0.179
Lobar ICH, n (%)	31 (26.1)	16 (38.1)	15 (19.7)	0.027
Pre-ICH antithrombotic use, n (%)*	31 (26.1)	11 (26.2%)	20 (27%)	0.922
ICH Volume, median (IQR) mL	13.9 (0.4–136.4)	25.1 (0.6–109.9)	12.6 (0.4–136.4)	0.033
IVH, n (%)	48 (40.3)	25 (59.5)	23 (30.3)	0.002
Microbleeds (y/n), n (%)	69 (59.0)	31 (73.8)	38 (51.4)	0.015
Chronic hematomas (y/n), n (%)	36 (30.5)	15 (35.7)	21 (28.0)	0.361
WMD score, median (range)	6 (0–12)	6 (2–12)	6 (0–12)	0.105

SD = standard deviation, CES = cardioembolic source, IQR = interquartile range

^*Baseline antithrombotic use unavailable in 3 patients;

^†In-hospital antihypertensive use unavailable in 5 patients

Table 2.

Multivariate logistic model for presence of baseline DWI lesions

	Odds Ratio	95% CI		p Value
Hematoma Volume (mL)	1.02	1.00	1.04	0.025
Baseline IVH (y/n)	3.00	1.20	7.48	0.019
Baseline Microbleeds (y/n)	2.94	1.16	7.49	0.024
Delta MAP (mm Hg)	1.03	1.01	1.05	0.003

Univariate analysis for the 1 month timepoint (Table 3) demonstrated that patients with DWI lesions had a greater likelihood of more severe leukoaraiosis (p=0.004), microbleeds (p=0.004), chronic hematomas (p=0.019), DWI lesions at baseline (p=0.001), and greater mean delta MAP (p< 0.001). There was a trend for DWI lesions to occur with greater frequency in black patients (p=0.056), and in patients with prior stroke history (p=0.056). A multivariate logistic regression model predicting the presence or absence of 1 month DWI lesions was developed that included black race, history of any prior stroke, presence of 1 or more microbleeds, and presence of a DWI lesion at baseline (Table 4). As mean delta MAP was significantly correlated with baseline DWI lesions, it was not incorporated into the 1 month model. In this study with a primary focus on racial disparities in a sample of predominantly black ICH patients, black race was integrated into the model given the significant associations with microbleeds and vascular risk factors.

Table 3.

Patient characteristics stratified by presence of baseline 1 month DWI lesions

	All Subjects (n=113)	DWI Positive (n=30)	DWI Negative (n=83)	p Value
Age, mean (SD)	58.3 (12.4)	58.8 (10.0)	58.0 (13.3)	0.751
Black race, n (%)	83 (73.5)	26 (86.7)	57 (68.7)	0.056
Male gender, n (%)	62 (54.9)	19 (63.3)	43 (51.8)	0.277
Hypertension (y/n), n (%)	91 (80.5)	27 (90.0)	64 (77.1)	0.28
Delta MAP (mm Hg), mean (SD)	40.3 (26.0)	53.6 (25.6)	35.6 (24.5)	< 0.001
NIHSS, median (range)	4 (0–40)	7 (0–40)	4 (0–40)	0.315
Prior stroke (any type), n (%)	27 (23.9)	11 (36.7)	16 (19.3)	0.056
High risk CES, n (%)	7 (6.2)	2 (6.7)	5 (6.0)	0.999
Lobar ICH, n (%)	22 (23.4)	6 (26.1)	16 (33.5)	0.727
ICH volume*, median (IQR) mL	12 (0.4–97.6)	12.7 (0.8–97.6)	11.2 (0.4–87.4)	0.667
IVH*, n (%)	36 (38.3)	9 (39.1)	27 (38.0)	0.925
Microbleeds* (y/n), n (%)	51 (55.4)	18 (81.8)	33 (47.1)	0.004
Chronic hematomas (y/n), n (%)	30 (32.3)	12 (52.2)	18 (25.7)	0.019
WMD score, median (range)	6 (0–12)	8 (3–12)	6 (0–12)	0.004
Baseline DWI lesions (y/n)*, n (%)	30 (31.9)	14 (60.9)	16 (22.5)	0.001

^*Imaging data based on baseline imaging timepoint; SD = standard deviation, CES = cardioembolic source, IQR = interquartile range

Table 4.

Multivariate logistic model for presence of 1 month DWI lesions

	Odds Ratio	95% CI		p Value
Any prior stroke (y/n)	4.58	1.40	14.93	0.012
Baseline microbleeds (y/n)	3.92	1.07	14.38	0.040
Baseline DWI lesions (y/n)	4.72	1.53	14.63	0.007
Black vs. non-black race	1.37	0.36	5.18	0.641

DISCUSSION

Summary of Findings

This prospective, longitudinal MR imaging-based study of primary ICH not only confirms prior studies reporting a high frequency of ischemic lesions (topographically remote from the hematoma) in the acute phase of primary ICH, but also extends these findings to provide important new insights into the dynamic nature and underlying etiology of these lesions. Our analysis identifies active regions of remote ischemia at baseline in more than 1/3 of patients, while follow-up imaging at 1 month demonstrates ischemic lesions in over 1/4. Notably, 83% of 1 month lesions were new and not present on baseline imaging, a finding indicative of an ongoing and active pathologic process. Independent predictors of DWI lesions at baseline included larger hematoma volume, intraventricular hemorrhage, microbleeds, and importantly, large reductions in mean arterial blood pressure. A multivariate model predicting ischemic lesions at 1 month included presence of DWI lesions at the baseline timepoint, prior stroke, microbleeds, and black race.

Prior Literature

One prior study with a similar demographic profile to our cohort, evaluated 129 patients with MRI in the first 28 days after acute primary ICH.³ DWI abnormalities were identified in 22.9% of patients and were associated with prior stroke, mean arterial blood pressure lowering and craniotomy. An analysis of microbleeds was not reported in this study. In a separate study of patients with cerebral amyloid angiopathy, 15% of patients demonstrated subacute cerebral infarction on DWI studies.⁴ In this cohort, DWI lesions were not associated with recent symptomatic ICH but were associated with higher microbleed burden.

Mechanisms and Pathophysiology

The etiology and underlying pathophysiology leading to ischemic lesions remote from the index hematoma is not yet well understood. Putative mechanisms include 1) an active cerebral vasculopathy leading to contemporaneous vessel rupture and occlusion, and/or 2) overly aggressive blood pressure lowering in the acute (and possibly subacute) phase in the setting of a diffuse angiopathy, and/or 3) an acute prothrombotic/proinflammatory cascade triggered by the initial ICH or the underlying vasculopathic process itself. Less likely causes include misidentification of hemorrhagic conversion of embolic ischemic lesions (i.e. misdiagnosis of primary ICH), iatrogenically-induced lesions related to diagnostic angiography, or active vasculitis.

Our results strongly suggest an association between DWI lesions in ICH and blood pressure lowering during the acute phase of ICH hospital stay. Historically, the foci of prior studies of blood pressure management in acute ICH have varied based on the understanding of the underlying pathophysiology of ICH at the time. Initial theories regarding a penumbra of perihematomal ischemia fostered a concern for blood pressure lowering and hypoperfusion in the acute phase. This topic has been controversial with some studies suggesting a transient zone of true ischemia in a subset of patients, while others have not.^9,10,11 Some advanced imaging studies have additionally identified a perihematomal penumbra of nonischemic, metabolically compromised tissue.¹² Most recent studies have focused upon the association between early hematoma expansion, elevated blood pressure, and worsened clinical outcome. Two clinical trials have demonstrated the relative safety of early BP lowering with avoidance of substantial neurological deterioration and adverse events, along with modest reductions in hematomal expansion.^{13, 14} However, neither of these studies integrated MRI assessments that might identify clinically silent ischemia, or microvascular damage.

Blood pressure lowering alone, however, may not be sufficient to explain remote ischemia in the setting of acute ICH. We postulate an interaction may exist between blood pressure lowering and a severe underlying vasculopathy in the setting of ICH. Results from our study and others, suggest that imaging and clinical markers of an underlying severe vasculopathy (e.g. prior stroke and microbleeds) are independent predictors of ischemic lesions.^3,4 Disease of the microvasculature, whether due to hypertension or cerebral amyloid angiopathy, has the potential to simultaneously lead to both ischemic and hemorrhagic insults sharing a common underlying microangiopathic process. We hypothesize that the development of ischemic lesions in acute and/or subacute hypertensive ICH is caused by hemodynamic insufficiency due to BP lowering in the setting of a diseased microvasculature that is unable to compensate due to autoregulatory failure. However, it is important to note that our data only demonstrates an association between blood pressure lowering and DWI lesion presence, and further prospective studies are needed not only to confirm this hypothesis, but also to demonstrate that less aggressive blood pressure treatment decreases the rate of lesion development.

Significance

An understanding of the underlying mechanism(s) and pathophysiology of ischemic lesions in the setting of ICH has important implications for both prognosis and treatment. Our study uniquely characterizes the frequency of DWI lesions at two discrete timepoints (baseline and 1 month). The presence of new ischemic lesions as late as 1 month from ictus demonstrates the active nature of this process. While the delta MAP did not enter into the 1 month predictive model, this was likely due to colinearity with baseline DWI lesions which in turn were associated with the delta MAP. This finding suggests that overaggressive blood pressure reductions in the subacute phase of ICH may have important implications. This population of hypertensive patients, often with severe microvascular disease, may have persistent loss of autoregulation. Therefore, blood pressure control within the normal range may not be tolerated, and instead could induce hemodynamic ischemia at the microvascular level.

Moreover, if blood pressure lowering is the predominant factor at either timepoint, it would be important to understand the clinical significance of the DWI lesions – e.g. whether they are clinically silent or have a negative impact on long-term functional outcome or cognitive status. If associated with poorer outcome, future studies exploring optimal blood pressure control are needed and should include MR imaging as an outcome marker. However, if ischemic lesions are thrombotic in nature and triggered by the acute ICH and/or caused by an ongoing, active vasculopathy at 1 month independent of blood pressure, future studies would be needed to explore the clinical significance of the lesions, as well as optimal approaches to secondary prevention (e.g. role of antiplatelet agents). Our planned analysis of this population with a 1 year MRI will allow us to characterize the frequency of DWI ischemic lesions at a timepoint far removed from the acute hemorrhage (e.g. chronic rate of lesion development in this hypertensive ICH population).

Limitations

This study had some limitations. There is likely a selection bias toward milder hemorrhages with smaller volumes as patients with more severe hemorrhages may have been too unstable for MRI. The sensitivity for detection of microbleeds and ischemic DWI lesions may depend on magnet strength. One prior report suggests that 1.5T magnets may be more sensitive for detecting hyperacute ischemia on diffusion-weighted imaging.¹⁵ Since the 1 month scans were preferentially acquired on a 3.0T scanner, this may have decreased the overall detection rate at this timepoint. We did not formally analyze ischemic lesions based upon classical definitions of watershed territories. However, it is unclear that diffuse microvascular disease would have a similar watershed pattern as large vessel disease. Finally, we did not routinely collect MR perfusion studies as part of our protocol. Future studies are needed to examine these issues.

Summary and Conclusions

In conclusion, we found that more than one-third of patients with ICH have active cerebral ischemia topographically remote from the acute hematoma at baseline, and, importantly, more than one-quarter of patients have imaging confirmation of active ischemia in the subacute phase of ICH. Multivariate analyses implicate blood pressure reductions in the setting of an active vasculopathy as an underlying mechanism. Further studies are needed to determine the impact of these lesions on outcome and optimal management strategies to arrest vascular damage.