Intracerebral Hemorrhage in Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy

Introduction

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), characterized by recurrent ischemic stroke, is the most common monogenic cerebral small vessel disease (SVD).¹ Intracerebral hemorrhage (ICH) was once considered rare in patients with CADASIL and was usually attributed to hypertension or antithrombotic agents,^2–4 but recent accumulating data have shown that the prevalence of ICH in CADASIL might be underestimated.^5–16 However, the role of ICH in CADASIL is yet to be clearly understood because of the lack of adequate relevant studies. Moreover, among individuals with CADASIL, it is crucial to be able to identify those who are prone to developing ICH. Tailored therapeutic strategies, including the antithrombotic agent choice and a goal of blood pressure (BP) control, might help decrease the risks of ICH in this subpopulation.

The present study aims to unravel the role of ICH in CADASIL by investigating the prevalence and clinical characteristics of ICH in a Taiwanese CADASIL cohort and identifying clinical and neuroimaging features associated with the presence of ICH lesions in patients with CADASIL.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Subjects

The participants of this retrospective cross-sectional study were selected from a consecutive series of 255 patients with genetically confirmed CADASIL diagnosed at Taipei Veterans General Hospital between January 2010 and December 2019. Taipei Veterans General Hospital is a 2974-bed tertiary medical center that serves both veterans and regular citizens in Taiwan. It accepts both self-referred patients and referrals of difficult cases from other hospitals. Among the 255 patients with CADASIL, 127 patients had undergone heme sensitive sequences and were enrolled in the present study. A questionnaire was used to collect data from patients or/and their families about their demographic information and medical histories including hypertension, diabetes, hyperlipidemia, and antithrombotic medications as well as their personal histories including smoking and alcohol consumption. Clinical events of ischemic stroke, transient ischemic attack, and ICH were ascertained by patient interviews and medical record reviews. Brain magnetic resonance imaging (MRI) with or without brain computed tomography images was used to identify ICH lesions. The duration of disease was defined as the interval between the onset year of a patient’s initial presentation (stroke, cognitive impairment, or gait disturbance) and the year the brain MRI was performed.

ICH lesions accompanied by a clinical event, newly developed neurological deficit, were defined as symptomatic ICH lesions, while hemosiderin deposits with a diameter >10 mm on susceptibility-weighted imaging (SWI) or T2*-weighted gradient-recalled echo (T2*-GRE) images without corresponding symptoms, signs, and past history were designated as asymptomatic ICH lesions. Individuals with >2 records of systolic and diastolic BPs higher than 140/90 mm Hg in the past medical history were defined as having inadequate BP control. The study was approved by the Institutional Review Board of Taipei Veterans General Hospital, and written informed consent was obtained from each participant.

Genetic Analysis

Genomic DNA was extracted from peripheral blood samples. Mutation analyses of exons 2 to 24 of the NOTCH3 gene and genotyping of ε2/ε3/ε4 variants in the APOE gene were performed by polymerase chain reaction amplification and Sanger sequencing.^17,18

Brain MRI Acquisition and Analysis

Brain MRI of all participants were acquired on the 1.5-Tesla MR system (Signa Excite II HDxt or Signa HDxt, GE Medical Systems, Milwaukee, WI). Among the 255 patients with CADASIL in our cohort, 101 had undergone SWI, 26 had received T2*-GRE imaging, and 128 neither had SWI nor T2*-GRE imaging. Fifteen of the 127 patients receiving heme sensitive sequences and 12 of the 128 patients without heme sensitive sequences ever had symptomatic ICH events (Table 1; Figure 1; Table I and Figure I in the Data Supplement). There was no significant difference in the symptomatic ICH prevalence between the 2 groups (11.8% [15 of 127] versus 9.4% [12 of 128], χ² test P=0.64). Comparing the demographic features and neuroimaging findings between the patients with and without receiving heme sensitive sequences, there was no difference between the 2 groups (Tables II and III in the Data Supplement). We did not include these 128 patients with CADASIL in the present study because SWI or T2*-GRE imaging was indispensable for the detection of cerebral microbleeds (CMBs) and asymptomatic ICH lesions. Dr Hu, who is a board-certified neurologist with 6 years of experience in SVD, reviewed all the images. Lacunes of presumed vascular origin (abbreviated as lacunes in the following), white matter hyperintensity (WMH), CMBs, and perivascular space (PVS) were identified according the Standards for Reporting Vascular Changes on Neuroimaging criteria.¹⁹ The presence of lacunes at the corona radiata, basal ganglia, thalamus, brain stem, and cerebellum was measured using fluid-attenuated inversion recovery images, while PVS at the basal ganglia and centrum semiovale was assessed on axial T2-weighted images with a 4-point rating scale (grade 0=no PVS, grade 1=less than 11 PVSs, grade 2=11–20 PVSs, grade 3=21–40 PVSs, grade 4=more than 40 PVSs).²⁰ The severity of WMH in the periventricular and deep white matter was evaluated by the Fazekas scale ranging from 0 to 3.²¹

• Download figure
• Download PowerPoint

CMBs were defined as small, rounded hypointense areas that typically had a diameter of 2 to 5 mm but may have had a diameter up to 10 mm on SWI or T2*-GRE images.²² CMB presence was evaluated in the lobar regions (including frontal, parietal, temporal, and occipital lobes), the basal ganglia, the thalamus, the brain stem, and the cerebellum separately. The total CMB count in the whole brain was calculated manually, and a cutoff value of 10 was used to define a high CMB burden (total CMB count >10).¹⁶ The total SVD score was used to estimate the global burden of SVD with 1 point awarded for each of the following: presence of one or more lacunes, presence of one or more CMBs, presence of ≥11 PVSs (grade 2–4) in the basal ganglia, and presence of moderate to severe WMH (Fazekas scale 2–3 in the deep white matter or Fazekas scale 3 in the periventricular white matter).²³

Statistical Analysis

Statistical analysis was performed using SPSS version 19.0. All data are presented as the mean±SD or a number (%).We first compared the clinical and imaging features between patients with CADASIL with ICH lesions and those without ICH lesions. Later, similar analyses were conducted by dividing the patients with CADASIL with ICH lesions into 2 groups: individuals with symptomatic ICH lesions and those with asymptomatic ICH lesions only. We first used univariate logistic regression to identify vascular risk factors or image parameters that were significantly related to the presence of ICH lesions in CADASIL. Odds ratio and corresponding 95% CI of each variable was estimated. Generally, a 2-sided P<0.05 was considered statistically significant. For the image features tested in different brain regions, we used Bonferroni correction for the concern of multiple testing. Multivariate regression analysis with a stepwise forward variable selection method was then performed to determine factors independently associated with ICH presence in CADASIL. To avoid overfitting the model, we included the only significant clinical feature (ie, hypertension) and the most significant parameter in each image feature (ie, lacunes in the basal ganglia, CMBs in the brain stem, total CMB count >10, and total SVD score) in the multivariate regression analysis.

Results

Thirty-seven ICH lesions, including 15 symptomatic lesions and 22 asymptomatic lesions, were identified in 27 (21.3%) of the 127 patients with CADASIL (Table 1). The thalamus and lobar regions were the most common sites of these ICH lesions, each accounting for approximately one-third of the ICH lesions (12 thalamic hemorrhages and 11 lobar hemorrhages). Eight of the 11 (72.7%) lobar ICH lesions were asymptomatic, suggesting that lobar hemorrhages in CADASIL often occurred silently. Frontal and temporal lobes were the most common sites of lobar hemorrhages, with 4 lesions each. Other ICH lesions were located in the cerebellum (N=5), corona radiata (N=5), basal ganglia (N=2), and pons (N=2).

Among the 27 patients with CADASIL with ICH lesions on MRI, 15 individuals exhibited symptomatic ICH lesions, including 4 patients with additional asymptomatic ICH lesions and 11 patients only with symptomatic ICH lesions (Table 1, Figure 1). The other 12 patients only had asymptomatic ICH lesions (Figure 2).

• Download figure
• Download PowerPoint

Comparison Between Patients With CADASIL With and Without ICH Lesions

We then compared the demographic data, vascular risk factors, genetic features, and percentage of patients using antithrombotic agents between the 27 patients with CADASIL with ICH lesions and the other 100 subjects without ICH lesions (Table 2). The only feature differently distributed between the 2 groups was hypertension. NOTCH3 p.R544C is the most common mutation in Taiwanese patients with CADASIL²⁴ and was present in 77.2% (98/127) of the patients in this study. The other 19 NOTCH3 mutations of the remaining 29 patients with CADASIL are detailed in Table IV of the Data Supplement. We did not find any association between the NOTCH3 p.R544C mutation and ICH presence, and the distributions of APOE genotypes were similar between the patients with CADASIL with and without ICH lesions.

When comparing the neuroimaging features between the ICH and non-ICH groups, lacunes in the basal ganglia and brain stem; CMBs in the lobar regions, basal ganglia, brain stem, and cerebellum; a total CMB count >10 and total SVD scores were all significantly associated with ICH presence after Bonferroni correction (Table 3). There was a much higher percentage of individuals with CMBs in the brain stem in the ICH group than in the non-ICH group (81.5% versus 29.0%). Up to 85.2% of the patients with CADASIL with ICH lesions and 39.0% of those without ICH lesions had >10 CMBs, suggesting a significantly greater CMB burden in the ICH group than in the non-ICH group. When the 4 neuroimaging features of SVD were jointly evaluated, the patients with CADASIL with ICH lesions tended to have a higher total SVD score than those without ICH lesions (3.81±0.40 versus 3.01±1.24). However, the severity of PVS and WMH was similar between the 2 groups.

We then analyzed clinical or neuroimaging features that were independently associated with ICH presence in patients with CADASIL. Only CMBs in the brain stem and a total CMB count >10 were independently associated with the presence of ICH lesions in patients with CADASIL, with ORs of 5.82 (95% CI, 1.80–18.80) and 3.83 (95% CI, 1.08–13.67), respectively (Table 4).

Comparison Between Patients With CADASIL With Symptomatic ICH Lesions and Those With Asymptomatic ICH Lesions

For the 15 individuals with symptomatic ICH lesions, 8 (53.3%) suffered from ischemic stroke/transient ischemic attack before the hemorrhagic event, and 7 (46.7%) had ICH as their first presentation of CADASIL. Interestingly, the majority of them (11 of 15, 73.3%) had multiple stroke episodes with ischemic and hemorrhagic events occurring alternatively (Table 1).

When comparing patients with symptomatic ICH and those with only asymptomatic ICH lesions, the distributions of age, sex, disease duration, vascular risk factors, and percentage of the patients using antithrombotic agents were all similar between the 2 groups (Table V in the Data Supplement). In addition, the CMB burden and the MRI features of SVD were comparable between the 2 groups (Table VI in the Data Supplement). There was no difference in the average total SVD scores between the symptomatic ICH group and asymptomatic ICH group (3.73±0.46 and 3.92±0.29, respectively).

Discussion

The present study demonstrates 35 ICH lesions in 27 out of 127 Taiwanese CADASIL patients and identifies clinical and neuroimaging features associated with the occurrence of ICH lesions in CADASIL. There are 5 major findings and implications. First, approximately one-fifth of the patients with CADASIL had ICH, and most of them suffered from recurrent stroke with both ischemic and hemorrhagic events. Accordingly, ICH is an important clinical presentation of CADASIL, similar to lacunar infarction, cognitive decline, psychiatric symptoms, and migraine. Second, some of the ICH lesions are asymptomatic, suggesting that ICH lesions in patients with CADASIL could be easily overlooked if SWI or T2*GRE images are not routinely examined. Third, patients with CADASIL with ICH lesions had significantly higher total SVD scores than those without ICH lesions, implying a greater burden of vascular brain injury in the ICH group. On the other hand, patients with asymptomatic ICH lesions and those with symptomatic ICH lesions may share similar clinical and neuroimaging features. Fourth, hypertension is the most important vascular risk factor for ICH in CADASIL. Age, sex, NOTCH3 p.R544C mutation, APOE ε2/ε3/ε4 alleles, and common vascular factors other than hypertension are not associated with ICH in the patients with CADASIL. Fifth, multivariate regression analysis revealed that the presence of CMBs in the brain stem and a total CMB count > 10 were independently associated with the presence of ICH lesions in patients with CADASIL. Therefore, these factors may be the candidates to serve as biomarkers to identify patients with CADASIL at risk of ICH and may facilitate the prevention of ICH in patients with CADASIL.

Approximately one-fifth of the patients with CADASIL in our cohort had ICH lesions on brain MRI. Similarly, ICH lesions were present in 17% of the Korean patients with CADASIL.¹⁶ However, the ICH prevalence in White patients with CADASIL ranged from 0.5% to 2%.^25,26 The prevalence of ICH reported in Chinese and Korean populations seems to be higher than the prevalence of ICH expected in Whites. Ethnicity and underestimation of ICH in patients with CADASIL might be 2 factors contributing to the difference in ICH prevalence. In general, the incidence of spontaneous ICH in Asians is >2 times that in White populations.²⁷ Underestimation of ICH in patients with CADASIL may come from a lack of clinical awareness of the role of ICH in CADASIL. The present study identified ICH lesions by utilizing SWI or T2*-GRE imaging, which are exquisitely sensitive for detecting hemorrhage.

In our CADASIL cohort, ICH lesions most frequently occurred in the thalamus and lobar regions, which were also the most common locations of CMBs (Table 3). The positional proximity between the ICH lesions and CMB sites implies that CMBs reflect underlying hemorrhage-prone microangiopathy, which may lead to hemorrhagic stroke. A previous retrospective MRI study of general ICH patients also demonstrated a close spatial relationship between the locations of recurrent ICH and prior CMBs.²⁸ Concordant with our findings, the thalamus and lobar regions were also the most common sites of CMBs in the Italian-British and Korean CADASIL cohorts.^16,25,29

We found a higher total SVD score in patients with CADASIL with ICH lesions than in those without ICH lesions, which might imply a greater burden of CADASIL vasculopathy in individuals presenting with ICH. A recent study showed that patients with CADASIL with ICH carried a 2.17-fold risk of recurrent stroke and had a much higher mortality rate than those with only ischemic stroke (28.6% versus 8.7%).³⁰ In the population-based Rotterdam study, general participants with a higher SVD sum score had a much greater risk for incident stroke, dementia, and mortality during the 10-year follow-up.³¹ More studies are warranted to investigate whether patients with CADASIL with ICH have a greater disease burden and poorer prognosis than those without ICH. Since the total SVD score was similar between patients with CADASIL with asymptomatic ICH lesions and those with symptomatic ICH lesions, the 2 groups might share a similar disease course with recurrent hemorrhagic and ischemic stroke.

Identifying patients with CADASIL who are at risk of hemorrhagic stroke is crucial in clinical management and prognosis prediction. In the present study, the existence of hypertension, the presence of CMBs in the brain stem, a CMB count >10, and a higher total SVD score were significantly associated with ICH presence in patients with CADASIL. Similarly, one previous study showed that the CMB burden was independently associated with the occurrence of ICH in patients with CADASIL.^6,16 There is still no consensus for the management of patients with CADASIL with ICH or those at risk of ICH. According to the American Heart Association/American Stroke Association guidelines, the long-term goal of BP control is systolic BP and diastolic BP <130/80 mm Hg in general patients with ICH.³² A more intensive target of systolic BP/and diastolic BP <120/80 mm Hg may be considered in patients with CADASIL with ICH. This is because that the latest evidence showed that intensive BP control significantly reduced the recurrence rate of ischemic and hemorrhagic stroke in general populations, whereas the PRESERVE trial (Prevention of Serious Adverse Events Following Angiography) confirmed that intensive BP control did not reduce cerebral perfusion in patients with SVD.³³ Regarding antiplatelet therapy, specific evidence supporting its benefit to patients with CADASIL is still lacking. Since CMBs have been reported as a potential risk factor for antiplatelet-related ICH,^34,35 their use in patient with CADASIL with ICH or those at risk of ICH should be cautious. Further studies are urgently required to ascertain the role of antiplatelet therapy or other therapeutic strategies for patients with CADASIL with or without ICH.

Our study has limitations that should be considered when interpreting the findings. For instance, the estimation of ICH prevalence in CADASIL might be suboptimal using a retrospective study design. The present study only included patients with CADASIL with heme sensitive sequences into analysis, which might introduce bias in the estimation of ICH prevalence in our cohort. However, it would not change the main finding that ICH is an important and common manifestation of CADASIL, especially for patients from Asian populations. Besides, both SWI and T2*-GRE images were used to detect ICH/CMBs in the present study. The influence of different sensitivities between SWI and T2*-GRE^36,37 might be trivial in the current study design since we focused on qualitative analyses rather than measuring the exact CMB numbers.

Conclusions

This study identified ICH lesions in approximately one-fifth of the 127 patients with CADASIL, and the lesions were most commonly located in the thalamus and lobar regions. Hypertension was the most important vascular risk factor for ICH in the patients with CADASIL. Patients with CADASIL with ICH lesions had a greater burden of SVD than those without ICH lesions. The presence of CMBs in the brain stem and a total CMB count >10 were independently associated with the presence of ICH in CADASIL. These findings suggest that ICH is an underestimated but important manifestation of CADASIL.

Acknowledgments

We are grateful to the patients who participated in the study.

Supplemental Materials

Figures I

Tables I–VI

Footnotes