White Matter Lesion Severity is Associated with Intraventricular Hemorrhage in Spontaneous ICH

Abstract

Background:

Intraventricular hemorrhage (IVH) and white matter lesion (WML) severity are associated with higher rates of death and disability in intracerebral hemorrhage (ICH). A prior report identified an increased risk of IVH with greater WML burden but did not control for location of ICH. We sought to determine whether a higher degree of WML is associated with a higher risk of IVH after controlling for ICH location.

Methods:

Utilizing the patient population from two large ICH studies; the Genetic and Environmental Risk Factors for Hemorrhagic Stroke (GERFHS III) Study and the Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH) Study, we graded WML using the Van Swieten Scale (0-1 for mild, 2 for moderate, and 3-4 for severe WML) and presence or absence of IVH in baseline CT scans. We used multivariable regression models to adjust for relevant covariates.

Results:

Among 3023 ICH patients, 1260 (41.7%) had presence of IVH. In patients with IVH, the proportion of severe WML (28.6%) was higher compared with patients without IVH (21.8%) (p<0.0001). Multivariable analysis demonstrated that moderate-severe WML, deep ICH, and increasing ICH volume were independently associated with presence of IVH. We found an increased risk of IVH with moderate-severe WML (OR=1.38; 95% Cl 1.03-1.86, p = 0.0328) in the subset of lobar hemorrhages.

Conclusion:

Moderate to severe WML is a risk for IVH. Even in lobar ICH hemorrhages, severe WML leads to an independent increased risk for ventricular rupture.

Introduction:

Hemorrhagic stroke affects approximately 100,000 persons annually in the United States, with a 30 day mortality rate of 40-50%.¹ An estimated 66% of all hemorrhagic strokes are classified as intracerebral hemorrhages (ICH) which is the most devastating form of stroke resulting in high mortality and disability.² Despite substantial improvement in acute ischemic stroke treatment in recent years, progress in ICH therapeutics has remained limited.

Multiple prognostic factors contributing to outcome after ICH including age, ICH location, hematoma size, hemorrhage expansion, anticoagulant use, baseline GCS score, and presence and extent of intraventricular hemorrhage have been identified and are included in current prediction models such as ICH and FUNC scores.^3–5 Intraventricular hemorrhage (IVH) is caused by the hemorrhage rupturing into the ventricles and is an important predictor of outcome, increasing mortality by up to 80%.^6–8 The burden of white matter lesion (WML) or leukoariaosis has been identified as a predictor of worse ICH outcomes.⁹

Although both IVH and WML severity are associated with higher rates of death and disability among patients with ICH,^{6, 10} there is uncertainty regarding the association between the two.¹⁰ Prior studies have suggested an increased risk of IVH with higher WML burden in patients with ICH.¹¹ However, hypertension is a major risk factor for both IVH and WML and the prior analyses were not stratified by location of ICH.¹¹ If WML is weakening the ventricular wall or permits ICH to track more easily through the brain tissue into the ventricles, then the association should also be true for lobar ICH. Our objective was to test the hypothesis that a higher degree of WML is associated with a higher risk of IVH after controlling for ICH location.

Materials & Methods:

Study Population

We utilized the patient population from two large ICH studies; the Genetic and Environmental Risk Factors for Hemorrhagic Stroke (GERFHS III) study and the Ethnic/Racial Variations of Intracerebral Hemorrhage (ERICH) study.^{12, 13} GERFHS III is a population-based study of ICH patients residing in the Greater Cincinnati/Northern Kentucky (GCNK) region and ERICH is a prospective, multi-center, case-control study of ICH epidemiology among 1,000 whites, 1,000 blacks, and 1,000 Hispanics matched to 3,000 controls.^{12, 13} The study was carried out in accordance with International Conference on Harmonization Good Clinical Practice (ICH GCP). Investigators responsible for the conduct of this study have completed Human Subjects Protection and ICH GCP Training. Institutional Board Review approval was required and obtained for the GERFHS and ERICH studies from each participating hospital system and study site prior to the enrollment of participants.^{12, 13}

Detailed methods for the GERFHS III and ERICH studies have been published previously.^{12, 13} Out of the entire GERFHS III and ERICH population, we analyzed patients from the interviewed cohort comprised of enrolled participants that underwent an interview and had baseline demographics, clinical, discharge status, and outcome data available per study protocol (Figure 1).

Figure 1:

Flowchart of Included Patients.

Imaging Analysis

We analyzed baseline head CT scans of the interviewed cohort of both GERFHS III and ERICH. The baseline CT scans were non-contrast standard of care imaging and obtained at minimum, 5mm sections. These CT scans were analyzed by trained investigators who graded WML using the Van Swieten scale.¹⁴ This scale ranges from 0-4; a score of 0 indicates no WML and a score of 4 indicates severe WML.¹⁴ For our study we defined mild WML having a score of 0-1, moderate WML having a score of 2, and severe WML having a score of 3-4. We also analyzed the CT scans for presence or absence of IVH. Presence of IVH was defined as Graeb score >0 or IVH volume > 0 for ERICH dataset and Graeb score >0 or IVH score > 0 for GERFHS dataset.¹⁵ The Graeb score is a semiquantitative scaling system for IVH extension of ICH, ranges from 0 to 12 points and is based on blood filling and expanding the third, fourth and lateral ventricles.¹⁶ The IVH score grades each lateral ventricle with a score of 0-3, third and fourth ventricles with score of 0-1 and hydrocephalus is scored as present (1) or absent (0).¹⁷

Statistical Analysis:

The data were managed and analyzed using SAS version 9.4 (SAS Institute). Descriptive statistics are presented as means, standard deviations and medians for continuous variables and as percentages for categorical variables. Those with presence of IVH were compared with those without on the various risk factors using two-sample t-test, Wilcoxon rank-sum test or chi-square test as appropriate. Multivariable logistic regression model for presence of IVH was constructed by initially including all factors with p<0.2 and backward eliminated to include only those that were p<0.05. A similar multivariable logistic regression model for presence of IVH was constructed using only those with lobar ICH. Initial analysis revealed a non-linear distribution of hematoma volume; thus, hematoma volumes were fitted in the multivariable model as the natural logarithm of volume plus one. We also tested for WML interactions between location, hypertension, and ICH volume.

Results:

Baseline characteristics

A total of 3023 patients were included using both GERFHS III and ERICH cohorts. Among these ICH patients (mean± SD age 63.1± 14.8; 41.3% females), 1260 (41.7%) had presence of IVH. Table 1 presents the demographic variables for the combined cohorts divided by presence of IVH.

Table 1:

Demographic and clinical characteristics of GERFHS III and ERICH

Variables	Overall N=3023	No IVH N=1763	Presence of IVH N=1260	p-Value
Age: Mean (SD)	63.1 (14.8)	63.0 (14.8)	63.1 (14.7)	0.8643
Gender: Females	1249 (41.3)	730 (41.4)	519 (41.2)	0.9053
Race:
Black	914 (30.2)	504 (28.6)	410 (32.5)
Hispanic	877 (29.0)	527 (29.9)	350 (27.8)
White	1232 (40.8)	732 (41.5)	500 (39.7)	0.0627
Hypertension	2541 (84.3)	1464 (83.1)	1077 (86.0)	0.0290
Diabetes	866 (28.7)	517 (29.3)	349 (27.7)	0.3430
High Cholesterol	1422 (47.8)	828 (47.4)	594 (48.3)	0.6146
History of Ischemic stroke	375 (12.4)	213 (12.1)	162 (12.9)	0.5237
History of Hemorrhagic stroke	193 (6.4)	122 (6.9)	71 (5.6)	0.1542
Smoking:
Ever Smoker	1476 (49.1)	846 (48.2)	630 (50.5)
Never Smoker	1528 (50.9)	910 (51.8)	618 (49.5)	0.2134
Heavy Alcohol Use	289 (9.9)	156 (9.1)	133 (11.0)	0.1046
Location of ICH:
Lobar	930 (30.8)	643 (36.5)	287 (22.8)
Deep/Pure IVH	1685 (55.7)	870 (49.4)	815 (64.7)
Brainstem/Cerebellum	408 (13.5)	250 (14.2)	158 (12.5)	<0.0001
WML Score:
Mild (0-1)	1728 (57.2)	1069 (60.6)	659 (52.3)
Moderate (2)	550 (18.2)	310 (17.6)	240 (19.0)
Severe (3-4)	745 (24.6)	384 (21.8)	361 (28.6)	<0.0001
ICH Volume: Median (IQR)	10.3 (3.7, 25.2)	8.6 (2.9, 20.7)	12.7 (5.4, 31.6)	<0.0001

In patients with IVH, the proportion of severe WML (28.6%) was significantly higher compared with patients without IVH (21.8%) (p<0.0001). The median volume of ICH was 12.7 mL (IQR, 5.4-31.6) in patients with IVH as compared with 8.6 ml (IQR, 2.9-20.7) in patients without IVH (p <0.0001). Compared with patients without IVH, those with IVH had greater percentage of hypertensive patients and included less brainstem/cerebellar hemorrhages (Table 1).

Multivariate analysis

Multivariable analysis (Table 2) demonstrated that moderate WML (OR=1.39; 95% Cl 1.14-1.71, p = 0.0015), severe WML (OR=1.68; 95% Cl 1.40-2.02, p <0.0001), deep ICH (OR=3.36; 95% Cl 2.77-4.07, p <0.0001), infratentorial location (OR=2.47; 95% Cl 1.89-3.23, p<0.0001), and increasing ICH volume (OR=1.61; 95% Cl 1.50-1.74, p<0.0001) were independently associated with presence of IVH. There was no significant interaction between WML and ICH location (p=0.3186), WML and hypertension (p=0.3188), and WML and ICH Volume (p=0.3686).

Table 2:

Multivariable Logistic Regression Model for Presence of IVH in GERFHS and ERICH Patients

	Model 1		Model 2		Model 3 (Full model)
Variable	OR (Cl)	P-value	OR (Cl)	P-value	OR (Cl)	P-value
WML score (Van Swieten):
Mild/None (0-1)	Reference		Reference		Reference
Moderate (2)	1.30 (1.06,1.58)	0.0114	1.30 (1.06,1.59)	0.0105	1.39 (1.14,1.71)	0.0015
Severe (3-4)	1.61 (1.33,1.95)	<0.0001	1.61 (1.33,1.95)	<0.0001	1.68 (1.40,2.02)	<0.0001
Age	1.00 (0.99,1.00)	0.195	1.00 (0.99,1.00)	0.132
Race:
White	Reference		Reference
Black	1.16 (0.97,1.40)	0.1078	1.12 (0.92,1.35)	0.2577
Hispanic	1.00 (0.83,1.20)	0.9598	0.97 (0.81,1.18)	0.7854
Gender
Female	Reference		Reference
Male	1.02 (0.88,1.18)	0.8014	1.01 (0.87,1.18)	0.8806
Hypertension			1.20 (0.97,1.47)	0.0923
Location of ICH:
Lobar					Reference
Deep					3.36 (2.77,4.07)	<0.0001
Infratentorial					2.47 (1.89,3.23)	<0.0001
ICH Volume					1.61 (1.50,1.74)	<0.0001

Abbreviations: OR= Odds Ratio, 95% CI 95% Confidence Interval, ICH = Intracerebral Hemorrhage, IVH = Intraventricular Hemorrhage, WML=White matter lesions

Additional variables included in the full model (model 3) were prior hemorrhage, and heavy alcohol use.

Subset Analysis of Lobar Hemorrhages

Since most deep ICH have little WML between the source of hemorrhage and ventricle, we sought to validate our hypothesis that greater WML weakens the ventricle wall in lobar ICH. Multivariate analysis for lobar only ICH cases was performed using both GERFHS III and ERICH data. After controlling for important risk factors, WML was independently associated with IVH in lobar ICH and moderate to severe WML increased the probability of IVH (adjusted OR 1.38; 95% Cl 1.03-1.86; p=0.0328). (Table 3)

Table 3:

Multivariate Model for Presence of IVH in GERFHS and ERICH Lobar ICH Patients

Variable	OR (95% Cl)	P value
WML Score (Van Swieten):
None-Mild (0-1)	Reference
Moderate – Severe (2-4)	1.38 (1.03, 1.86)	0.0328
In (ICH volume)	2.12(1.80, 2.49)	<0.0001

Abbreviations: OR= Odds Ratio 95% CI = 95% Confidence Interval, ICH = Intracerebral Hemorrhage, IVH = Intraventricular Hemorrhage, WML=White matter lesions, In = natural logarithm. Additional variables included in the model were race, hypertension, prior hemorrhage, and heavy alcohol use.

Discussion:

In this combined dataset of two large ICH cohorts, we found that higher proportion of those with IVH have moderate to severe WML compared to those without IVH. Further supporting this finding is that in lobar ICH, where the physical distance from the hemorrhage to the ventricle is greater than for deep caudate or thalamic ICH, greater WML burden was associated with presence of IVH. Our study had the advantages of sufficient power to control for risk factors for IVH such as hypertension to identify an independent relationship. This finding suggests that in addition to its association with risk factors for ICH, white matter lesions or leukoariaosis may lead to less resistance from the hemorrhage.

Although it is well known that IVH is an important variable for poor outcomes in ICH, our knowledge about the vessel and ventricular wall injury in an ICH event is very limited.¹⁸ The exact mechanisms to predict which hemorrhages will and will not dissect into the ventricular system are poorly understood. Hematoma extension into the ventricles is not only dependent on the deep locations and volume of ICH but may also be related to blood pressure fluctuations and abnormalities in coagulation cascade.¹⁹

Preexisting white matter disease could be an additional important link in understanding the hematoma rupture into the ventricular system. Studies have demonstrated that white matter disease increases risk of brain damage in the presence of acute injury, for example, severe WML are associated with infarct growth and worse outcome after ischemic stroke.²⁰

Greater white matter disease may weaken the ventricular wall. There are a multitude of histopathological changes in white matter disease including demyelination, gliosis, ependymal discontinuity, vacuolation, and tissue loss.²¹ Furthermore, different mechanisms of white matter disease including endothelial injury, abnormalities in blood brain barrier permeability and neurovascular unit disruption have been proposed.²² It is well recognized that WML are associated with changes at a microstructural level resulting in reduced white matter integrity, which can also be demonstrated by imaging using Diffusion tensor imaging (DTI)²³. In addition, histopathological correlates of MRI white matter changes in Alzheimer’s disease versus controls have shown denudation in the ventricular lining correlating with loss of myelinated axons in the deep white matter.²⁴ These wider pathophysiologic alterations in white matter could possibly lead to ventricular wall alterations and weakening.²⁵

Only a few studies have investigated the association of WML and IVH in ICH patients. A post hoc analysis of a nationwide cohort in South Korea found similar associations like our study demonstrating that severity of WML on CT scans is related to the occurrence and amount of IVH in spontaneous ICH cases.¹¹ However, this study did not assess this relationship by hematoma location. In contrast to our findings, a study using the placebo group of the Factor Seven for Acute Hemorrhagic Stroke (FAST) trial found an association of WML with poor outcomes in ICH but not with intraventricular extension; however, the sample size (n=262) was smaller than our study.²⁶

Our study has important future clinical implications. As IVH is the most significant predictor of outcomes, preventing extension of the hematoma into the ventricles is a key target to prevent worse outcomes. If white matter disease makes the brain vulnerable to pathological insults such as ICH, then moderate to severe WML may be a new target for preventive and prognostication strategies.

Our study has several important limitations. We used CT scans rather than MRI scans for grading of WMH. While CTs were uniformly available at stroke onset, MRI is more accurate for visualizing the extent of WMH. However, a recent study showed substantial agreement between CT and MRI for visual rating scales of white matter disease.²⁷ Additionally, advanced imaging such as DTI to study the microstructural alterations in white matter were not included. Another limitation is that we did not perform volumetric analysis of WML. Volumetric assessments, although more accurate, are not practical in real life settings and visual rating of WML is a close substitute.

Summary & Conclusion

In conclusion, moderate to severe burden of WML is a risk for intraventricular hemorrhage in patients with ICH. Even in lobar ICH where hemorrhages are more remote from the ventricles, more severe WML leads to an independent increased risk for ventricular rupture. Further studies are needed to understand the mechanistic effects of white matter disease and associated microstructural damage resulting in intraventricular hemorrhage.

Figure 2.

Baseline CT head demonstrating lobar ICH with IVH and moderate to severe WML.