Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Review
Advances in understanding spontaneous intracerebral hemorrhage: insights from neuroimaging Expert Rev. Neurother. 14(6), 661–678 (2014)
Duncan Wilson, Andreas Charidimou and David J Werring* Stroke Research Group, UCL Institute of Neurology and The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK *Author for correspondence: Tel.: +44 020 3448 3541 [email protected]
Spontaneous (non-traumatic) symptomatic intracerebral hemorrhage is a devastating form of stroke, with very high overall mortality and morbidity. Even with the best current medical or surgical treatment, outcomes still remain poor. By contrast with ischemic stroke, the incidence of intracerebral hemorrhage is not decreasing. Indeed, the incidence of intracerebral hemorrhage related to antithrombotic drugs, including oral anticoagulants, has increased in recent decades. Despite the clear unmet research need for both prevention and acute treatment, there has, until recently, been limited progress in understanding the pathogenesis of this disease. New advances, especially related to neuroimaging biomarkers, are rapidly increasing our understanding of the spectrum of mechanisms of brain injury in intracerebral hemorrhage. The aim of this article is to review recent insights from neuroimaging studies into the pathophysiology and causes of intracerebral hemorrhage, focusing on MRI. We also discuss some of the current and future challenges facing clinicians in understanding and treating intracerebral hemorrhage. KEYWORDS: cerebral amyloid angiopathy • cerebral microbleeds • cortical superficial siderosis • hypertensive arteriopathy • intracerebral hemorrhage • leukoaraiosis • MRI
Intracerebral hemorrhage (ICH) is one of the most devastating forms of stroke, with a case fatality approaching 50% at 1 month [1] and high morbidity in survivors [2]. A recent metaanalysis involving 122 studies of longer-term prognosis found a pooled 1-year survival of 46%, and 5-year survival of only 29% [3]. However, not all ICH carries such a poor prognosis: there may be a subgroup that clinically present as a lacunar syndrome (sometimes termed ‘hemorrhagic lacunar stroke’), which may account for up to 7.4% of all cases and have a higher chance of full recovery (25 vs 3.7%; p = 0.012) [4] and lower in-hospital mortality (0 vs 33%; p £ 0.005) [5] than other ICH. Identifying such subgroups may have implications for optimum treatment of ICH. ICH accounts for 10–15% of strokes in Western populations, and up to 40% in some Asian populations. [6,7]. While the incidence of ischemic stroke has fallen [8], the incidence of ICH has remained stable in recent decades [6]. Indeed, there is evidence that the incidence of informahealthcare.com
10.1586/14737175.2014.918506
ICH in the elderly and in association with oral anticoagulant use is increasing [9–11], making it a key research challenge in cerebrovascular disease. ICH describes bleeding into the brain parenchyma, which may result from a wide range of potential causes. Here, we will not discuss other patterns of bleeding (e.g., subarachnoid, subdural, extradural hemorrhage) further. Conventionally, ICH is classified as ‘traumatic’ or ‘spontaneous’ (i.e., ‘nontraumatic’). The spontaneous group is further subdivided into ‘secondary’ (due to identified causes including bleeds into tumors, cavernomas, arterio-venous malformations, CNS infection, cerebral venous sinus thrombosis, bleeding disorders, etc.) or ‘primary’, if there is no obvious underlying cause. Although widely used, the term ‘primary ICH’ has come under criticism since using this label does not enlighten regarding the true underlying cause (s), yet might encourage a spurious diagnostic certainty and a failure to pursue further
2014 Informa UK Ltd
ISSN 1473-7175
661
Review
Wilson, Charidimou & Werring
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Normal brain
• Age • Vascular risk factors, e.g., hypertension • Genetic factors (e.g., ApoE)
Bleeding-prone small vessel diseases (e.g., hypertensive arteriopathy, CAA)
?? Lobar ICH
Acute precipitants? • e.g., severe hypertensive crisis, severe coagulopathy
• Acute precipitants – Hypertension variability – Minor head trauma – Recreational drugs • Chronic risk factors – Persistent hypertension – Antithrombotic drugs
Deep ICH
Figure 1. Aetiology of intracerebral hemorrhage. There is no single cause for ICH but rather an interplay between genetic and environmental factors. Vascular risk factors, age and genetics may all play a part in the development of bleeding-prone arteriopathies. Acute precipitants such as a minor head injury or hypertension then can cause small vessel to rupture. Antithrombotics may further increase potential risk or severity of ICH by impairing haemostasis (for example causing a normally self-limiting microbleed to enlarge into a symptomatic macro haemorrhage). CAA: Cerebral amyloid angiopathy; ICH: Intracerebral hemorrhage.
investigations. With cumulative advances in imaging and histopathological correlates, it is now generally accepted that the main processes underlying so-called ‘primary ICH’ are intrinsic diseases affecting cerebral small vessels (generally a few hundred microns and up to about 1–2 mm), usually collectively termed small vessel disease (SVD). A recent large international consensus group standardized definitions for neuroimaging markers of SVD and suggested the term ‘spontaneous ICH presumed to be due to SVD’ as preferable to primary ICH [12]. Challenges in assessing the ‘cause’ of spontaneous ICH
A discussion of the nature of causation is beyond the scope of this review, but a ‘cause’ can most simply be defined as something that affects the prevalence, likelihood or clinical effect of a disease. For ICH, contributory causes include the underlying SVD processes described above, but also ‘risk factors’ (e.g., hypertension, diabetes, lipid profile, smoking, alcohol use, etc.), which may influence the clinical expression of these SVD. The challenge with many studies of ICH (particularly cross-sectional) is that they are able to show associations, but cannot provide proof (or direction) of causality. Whether an association reflects causation can be considered according to the strength of association, consistency, 662
specificity, dose–response relationship, biological plausibility and consistency with natural history of the disease [13]. SVD is highly prevalent in older populations [14], yet ICH is much less common. Thus, as in other types of stroke, spontaneous ICH is likely to result from an interplay between environmental and individual patient (e.g., genetic) factors relating to the expression of SVD. Indeed, recent data suggest that genetic variation plays a significant role in ICH risk and outcome [15]. It was estimated that 44% of ICH risk variance was accounted for by genetic risk factors, with a greater contribution of genetic factors (especially apoE alleles) to lobar ICH than deep ICH [15]. One model of ICH causation is that multiple acute or chronic risk factors (e.g., age, sustained hypertension or shortterm blood pressure fluctuations, antithrombotics, serum cholesterol levels or statin use, minor head trauma, etc.) interact with vulnerable damaged small vessels (subject to the influence of genetic or other individual patient factors), which, when a certain threshold is exceeded, ultimately rupture to culminate in ICH as represented in FIGURE 1. Indeed, a recent populationbased study suggested that ICH may result from short-term increases in blood pressure prior to the event (over weeks to months), by contrast, with ischemic stroke where blood
Expert Rev. Neurother. 14(6), (2014)
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Advances in understanding spontaneous ICH
Vascular amyloid-β deposition in CAA
Review
Hypertensive arteriopathy
Figure 2. Bleeding prone small vessel vasculopathies. Schematic of the typical locations of the bleeding prone vasculopathies. Hypertensive arteriopathy: affecting mainly the deep arterial perforators by causing arteriolosclerosis, lipohyalinosis, and fibrinoid necrosis. CAA: accumulation of amyloid-b peptide in the walls of small-to-medium-sized arteries and arterioles predominantly located in the leptomeninges and the cerebral cortex. CAA: Cerebral amyloid angiopathy. Reproduced with permission from [181,182].
pressure is more stable [16]. This finding suggests that consistent long-term blood pressure control may be especially important in reducing the risk of ICH. Types of SVD
Most age-related small vessel damage is secondary to one of two main SVD processes (FIGURE 2): an arteriolar process often related to aging and other common vascular risk factors (e.g., hypertension and diabetes), characterized pathologically by lipohyalinosis, arteriolosclerosis or fibrinoid necrosis and typically affecting the small perforating end-arteries of the deep grey nuclei and deep white matter (often termed ‘ hypertensive arteriopathy’ ); and sporadic cerebral amyloid angiopathy (CAA), a disease process affecting superficial cortical and leptomeningeal vessels through the deposition of amyloid b (Ab). Less commonly, ICH occurs in the context of much rarer genetic diseases (familial cerebral amyloid angiopathies, collagen 4A1 mutations, Fabry’s disease, CADASIL, CARASIL, etc.) or cerebral and systemic vasculitides, which will not be discussed in this review. A classification of SVDs with their relevance to ICH is shown in TABLE 1. Sporadic cerebral amyloid angiopathy
Sporadic CAA is characterized by the progressive accumulation of Ab within the walls of small- to medium-sized arteries and arterioles – and, to a lesser extent, the capillaries and veins – in the leptomeninges and cerebral cortex [17]. In the most severe form, CAA-affected small vessels become thickened and disrupted, with focal wall fragmentation and blood extravasation, with or without microaneurysmal dilatation, and sometimes informahealthcare.com
show luminal occlusion [17]. The pathophysiology of sporadic CAA is poorly understood, but transgenic mouse models suggest an increased ratio of Ab40: Ab42 in the brain results in a shift of Ab from brain parenchyma to the vasculature [18] (perhaps by increasing the solubility of Ab and thus its diffusion into the vessel wall); and that vascular Ab deposition largely results from impaired clearance of Ab (rather than overproduction) [19]. The prevalence of CAA increases with age, especially over 60 years [20,21]. Autopsy studies have shown CAA in more than 70% of a healthy population above the age of 90 [22,23]. CAA may vary according to ethnicity: some studies suggest a higher predilection for amyloid deposition in the frontal lobe arteries in Eastern populations [24]; it has also been suggested that Eastern populations have a lower prevalence of CAA, but this may in part reflect the higher relative prevalence of hypertension and hypertensive arteriopathy in this population [24]. By contrast with other cerebrovascular diseases, CAA does not appear to be related to conventional vascular risk factors. Although hypertension may aggravate bleeding risk in CAA [25], most patients with CAA-related ICH (up to 68% in one study) are not hypertensive [26,27]. Anticoagulation or antiplatelet treatments may also increase the likelihood of CAArelated ICH [28–31], but, apart from genetic variants, there are no other known strong risk factors that increase the presence of CAA or its bleeding risk. apoE alleles are associated with CAA. A recent meta-analysis showed a dose-dependent association between apoE e4 alleles and the presence of sporadic CAA [32]. A more recent study, taking into account CAA severity, found a possible association of severe CAA with apoE e4 but not 663
Review
Wilson, Charidimou & Werring
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Table 1. Types, histopathology and prevalence of cerebral small vessel disease. Name
Histology
Prevalence
Strength of relationship to intracerebral hemorrhage
Arteriosclerosis Common sporadic SVD Hypertensive arteriopathy
Arteriolosclerosis Lipohyalinosis Fibrinoid necrosis Microatheroma Microaneurysms
Very common
Common cause of deep or lobar ICH
Cerebral amyloid angiopathy
Ab amyloid protein Dilated and disrupted walls Double-barreled arterial wall
Age related 7% those aged 65–74 18% those aged 75–84 [20] 70% those aged over 85 [23]
Common cause of lobar ICH
Inherited or genetic SVD CADASIL COL4A1 Fabry’s disease CARASIL
Depends on underlying disease
CADASIL >2 per 100,000 COL4A1 very rare prevalence unknown Fabry’s disease 1 per 120,000 CARASIL very rare, unknown
CADASIL – ICH rare (CMBs common) COL4A1 – ICH common Fabry’s disease – ICH rare CARASIL – ICH rare (CMBs common)
Inflammatory and Immunologically mediated SVD. (primary angiitis CNS and systemic vasculitis)
Typical of vasculitis
Very rare
ICH common
Other (radiation, etc.)
Depends on etiology
Rare
Rare
ICH: Intracerebral hemorrhage; SVD: Small vessel disease. Modified with permission from [17].
apoE e2, but was limited by the number of participants with e2 genotypes [33]. In addition, there have been multiple studies linking both apoE alleles (e2 and e4) to the risk of recurrent hemorrhage [34–37]. apoE e2 has been shown to confer greater vessel fragility [38]: a large genetic association study has shown carriers of apoE e2 had increased ICH expansion, mortality (odds ratio [OR]: 1.50; 95% CI: 1.23–1.82) and poorer functional outcomes (modified Rankin scale score 3–6; OR: 1.52; CI: 1.25–1.85) compared with non-carriers after lobar ICH. By contrast, apoE e4 was not associated with lobar ICH volume, functional outcome or mortality [39]. A number of other genetic variants are also associated with CAA-related ICH, for example, complement component receptor, TGF-b1 and TOMM40, but further work is needed to confirm these [35]. New genetic associations with CAA are sure to emerge from large international collaborative efforts in the coming years. There is substantial evidence supporting the hypothesis that CAA is an important contributory factor in causing ICH. This association between CAA and ICH was noted as far back as the 1970s and 1980s in a series of pathologically confirmed case reports of ICH [40–44]; more recently a large systematic review and meta-analysis of published histopathological studies confirmed an association between CAA and lobar ICH (OR: 2.21; 95% CI: 1.09–4.45) [45]. Given the very high pathological prevalence of CAA in population-based studies, and the much lower incidence of CAA-related ICH, the vast majority of patients with CAA pathology do not suffer ICH. Elucidating which patients with CAA will develop CAA-related ICH is a key question for preventing CAA-related ICH. A pathological study comparing brains with CAA and ICH with those with CAA without ICH 664
found CAA was more severe in the brains with cerebral hemorrhage than in those without, and that fibrinoid necrosis was seen only in the brains with cerebral hemorrhage [46]. Microaneurysms occurred only in the presence of severe, rather than moderate or mild, CAA [47]. This suggests that mild CAA may not confer such a high risk of ICH, and that preventing progression of CAA may thus reduce the incidence of ICH. Serum & CSF markers in CAA
Circulating serum and CSF biomarkers may be another way to detect and monitor CAA, but results to date are inconsistent. Decreased levels of Ab40 and 42 in the CSF in patients with CAA have been reported compared with both controls and patients with Alzheimer’s disease (AD) [48]. Another study [46] found that in CAA cases CSF total and phosphorylated tau were significantly lower in CAA than AD, but total tau was higher than healthy controls. Ab42 levels were significantly decreased in CAA when compared with controls and marginally lower than patients with AD, while Ab40 was significantly lower in CAA when compared with both AD and controls. Ab(1–42) and truncated fragments Ab(N–42) were higher in probable CAA patients than in controls (p < 0.001 and p = 0.046, respectively) [49]. Ab(N–42) was higher in those patients with multiple cerebral microbleeds (CMBs) when compared with those with only one CMB, suggesting that circulating Ab fragments in the peripheral circulation may only be found with more severe CAA pathology. Further studies of CSF markers of CAA are needed to confirm and clarify associations of Ab levels, and to explore other potential markers relevant for amyloid or neurodegenerative Expert Rev. Neurother. 14(6), (2014)
Advances in understanding spontaneous ICH
pathways (e.g., soluble amyloid precursor protein a and b, total tau, phospho-tau, neurofilament light and ferritin). Promising biomarkers should be correlated with well-established imaging markers of CAA, cognitive status and clinical outcome over time to determine their sensitivity to disease progression.
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Treatment of CAA
With no effective disease-modifying treatment for CAA, trials to date have focused on reducing the risk of recurrent ICH by influencing modifiable risk factors. Results from the PROGESS trial showed a striking reduction in presumed CAA-related ICH by 77% (95% CI: 19–93%) for those patients treated with antihypertensives [25], but this study did not use MRI to classify the likely cause of ICH and requires confirmation in other trial populations of ICH survivors. Reducing blood pressure is now the mainstay of secondary prevention treatment, but the optimum treatment target is not known. There is an urgent need for further trials to improve outcome after CAArelated ICH, for example, intensive long-term blood pressure lowering, use of statins and antithrombotic drugs. A Phase II randomized disease modification trial of the safety and efficacy of the monoclonal antibody ponezumab in CAA has been conducted. Ponezumab is a humanized monoclonal antibody originally used in AD trials. Unlike other monoclonal antibodies used in AD trials, ponezumab did not cause antibodyinduced microhemorrhage or other amyloid-related imaging abnormalities. Although this agent lacked efficacy in AD, it may be more effective in CAA as the amyloid burden is vascular in CAA (thus not protected by the blood–brain barrier, potentially allowing easier binding of the antibody to antigen) than in AD. The study is a Phase II, randomized, double-blind placebo-controlled trial to evaluate the safety, tolerability, pharmacokinetics and efficacy of ponezumab, with saline being used as the placebo. The primary end point of this study is an increase in cerebrovascular reactivity measured by visual BOLD fMRI, which is a potentially reversible marker of amyloid-related vascular functional compromise [50,51]. Hypertensive arteriopathy
Hypertensive arteriopathy is a term often used to describe multiple different (non-CAA) pathologies affecting mainly the deep arterial perforators (including arteriolosclerosis, lipohyalinosis and fibrinoid necrosis) [52–54], some of which are not clearly directly related to hypertension [55,56]. Sporadic SVD is a more encompassing term [55], but may also include CAA depending on how a small vessel is defined. The most direct histopathological correlate of hypertensive injury is fibrinoid necrosis, which is much more commonly found in brains from hypertensive patients than in those without hypertension [57–59] as well as in arterioles adjacent to deep ICH [58–61]. Furthermore, a study from the late 1970s showed that marked elevation in blood pressure can produce fibrinoid necrosis acutely [62]. Historically, hypertension has been considered the cause of the majority of non-lobar (i.e., deep) ICH. This observation first arose from autopsy series from the late 50s and early 70s of patients with deep ICH, in which classical pathological vessel informahealthcare.com
Review
changes suggested hypertensive arteriopathy [58,60]. These presumed hypertensive-related changes, including microaneurysms (‘Charcot-Bouchard’ or ‘miliary’ aneurysms) occurred in small penetrating arteries and arterioles subjected to high pressures such as the lenticulostriate arteries emanating from the middle cerebral artery, basilar and posterior cerebral artery perforators. These small arteries feed deep central areas of the brain such as the basal ganglia, thalami and brainstem [63] corresponding to the common locations of deep ICH. Thus, deep ICH has for many years been commonly attributed to hypertension and termed ‘hypertensive hemorrhage’. This theory has been challenged, primarily by the finding of a low prevalence of hypertension in some ICH cohorts [64,65] as well as pathological studies in which findings previously attributed to hypertension have been observed in the absence of hypertension [66] and associations of deep perforator SVD with other factors (e.g., reduced cerebral perfusion) [55,67–73]. Nevertheless, for simplicity and consistency, we will use the term ‘hypertensive arteriopathy’ to include nonCAA SVD affecting the small deep perforating arteries. Genetic studies have shown associations with the occurrence and outcome of deep ICH presumed due to hypertensive arteriopathy. A recent study showed a strong association between size of hemorrhage, clinical outcome and known hypertension risk alleles in those with deep ICH [74]. This study demonstrated an increase in hematoma volume by 28% and an increased risk of poor clinical outcome by 71% for every standard deviation rise in a blood pressure-based genetic risk score. Attributing lobar ICH to CAA and non-lobar ICH to hypertensive arteriopathy, while attractive for clinicians, is clearly an oversimplification. There is evidence from a recent large systematic review and meta-analysis showing lobar hemorrhages are significantly associated with CAA [45], but there was no statistically significant negative association between CAA and deep ICH. Furthermore, hypertensive arteriopathy can affect the white matter perforators and thus cause a proportion of lobar ICH in addition to deep ICH [75]; importantly, older individuals may in fact have a mixture of both CAA and hypertensive arteriopathy, as both are age-dependent [17]. Treatment
Treatment of hypertensive arteriopathy has focused on treating hypertension. The original PROGRESS trial showed the addition of perindopril-based antihypertensive medication regimen reduced the risk of all types of ICH by 50% (95% CI: 26–67%) in those with a previous stroke or transient ischemic attack [76]. Further analysis of this trial has shown that blood pressure lowering can reduce risk of ICH attributed to hypertensive arteriopathy (i.e., non-CAA-related ICH) by 46% (95% CI: 4–69%) [25], which, interestingly is a smaller treatment effect than seen in presumed CAA-related ICH. The reduced apparent effectiveness of lowering blood pressure in hypertensive-related ICH compared with CAA-related ICH likely reflects the notion that not all ‘hypertensive arteriopathy’ is related to hypertension, and not all ICH attributed to possible or possible CAA in this study was actually solely due to CAA, since advanced MRI was not used to 665
Review
Wilson, Charidimou & Werring
has increased fivefold [11], which closely corresponds to the fourfold increase in warfarin use (per capita) over the same period [11]. Restricted Anticoagulants should not ‘cause’ ICH diffusion per se but it is hypothesized that in vulnerable individuals (e.g., those with SVD) they Cortical lead to larger ICH volumes due to impaired superficial hemostasis. If this is the case then patients Leukoaraiosis siderosis with ICH on anticoagulants should have Deep CMBs larger hematoma volumes, and more hematoma expansion, than those not taking oral anticoagulants. The available studies are summarized in the online SUPPLEMENTARY TABLE 1 (supplementary material can be found online at www.informahealthcare.com/ suppl/14737175.2014.918506). Indeed, ICH anticoagulation is generally associated with larger baseline hemorrhages [63,79–84] and Figure 3. Small vessel disease. Schematic of small vessel disease markers seen on MR hematoma growth [83–85]; all but one of the modalities. CMB: Cerebral microbleed. studies with over 100 patients [86] show a Reproduced with permission from [183]. significant relationship between anticoagulation and hematoma size. One study revealed establish ICH etiology. Further large clinical trials are needed to an association between international normalized ratio intensity of confirm the effect of intensive blood pressure lowering on the anticoagulation and the volume of deep bleeds, but not for lobar risk of recurrent ICH. bleeds [79], suggesting differing underlying pathology (e.g., amyloid angiopathy and hypertensive arteriopathy) in different brain regions may respond differently to anticoagulation. However, this finding Changing epidemiology of spontaneous ICH & the role has not been replicated, and further studies are warranted. Three of anticoagulation The epidemiology of ICH has changed over the years with an studies to date show that hematoma expansion is increased by antiaging population, more aggressive treatment for hypertension coagulation; two of these studies were observational hospital based and increasing use of anticoagulation. It may well still be studies [84,85] without standardized CT timing scanning. changing: with the advent of the novel oral anticoagulants, we may expect to see an even higher use of anticoagulants in Neuroimaging markers in ICH elderly patients. Although these drugs have lower ICH risks Neuroimaging, particularly with MRI, is the most useful way than warfarin, the increased use may still overall lead to a to visualize the consequences of SVD. Although the small arterhigher prevalence of ICH in the elderly population. This is of ies are generally beyond the resolution of MRI, their complex particular importance as higher age is associated with worse out- effects on the brain (from both ischemia and hemorrhagic procomes [77,78]. One paper comparing ICH in the very elderly cesses) can be clearly seen. Here, we consider each neuroimagfound those aged greater than 85 years had increased incidence of ing marker in terms of its diagnostic and prognostic utility in neurological deficit at hospital discharge (89 vs 58%; p < 0.005) regard to future ICH. A schematic illustration of the main and in-hospital mortality (50 vs 27%; p < 0.01) than those youn- MRI markers of the SVDs relevant to ICH are shown ger than 85 years [77]. However, the cutoff age could be consid- in FIGURE 3 and example images are shown in FIGURE 4. ered arbitrary, and there was no significant difference in either neurological deficit or mortality when comparing those aged White matter hyperintensity (leukoaraiosis) greater than 85 to those aged between 65 and 85. Leukoaraiosis is a broad term originally used to describe confluWhile global ICH rates have remained stable [6], the improve- ent areas of low density on CT scans, but subsequently applied ments seen from improved hypertension control seem to be offset to high signal areas on T2-weighted MRI scans. More recently, by the increasing number of anticoagulation and CAA-related these changes on T2 and FLAIR MRI have been termed hemorrhages [9–11]. Large cohorts, each over a 20-year period from ‘White matter hyperintensities of presumed vascular origin’ Oxfordshire and Dijon have shown a decrease in the incidence of (WMH) [12]. WMH has a heterogeneous pathological substrate ICH in the younger age group (
Review
Advances in understanding spontaneous intracerebral hemorrhage: insights from neuroimaging Expert Rev. Neurother. 14(6), 661–678 (2014)
Duncan Wilson, Andreas Charidimou and David J Werring* Stroke Research Group, UCL Institute of Neurology and The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK *Author for correspondence: Tel.: +44 020 3448 3541 [email protected]
Spontaneous (non-traumatic) symptomatic intracerebral hemorrhage is a devastating form of stroke, with very high overall mortality and morbidity. Even with the best current medical or surgical treatment, outcomes still remain poor. By contrast with ischemic stroke, the incidence of intracerebral hemorrhage is not decreasing. Indeed, the incidence of intracerebral hemorrhage related to antithrombotic drugs, including oral anticoagulants, has increased in recent decades. Despite the clear unmet research need for both prevention and acute treatment, there has, until recently, been limited progress in understanding the pathogenesis of this disease. New advances, especially related to neuroimaging biomarkers, are rapidly increasing our understanding of the spectrum of mechanisms of brain injury in intracerebral hemorrhage. The aim of this article is to review recent insights from neuroimaging studies into the pathophysiology and causes of intracerebral hemorrhage, focusing on MRI. We also discuss some of the current and future challenges facing clinicians in understanding and treating intracerebral hemorrhage. KEYWORDS: cerebral amyloid angiopathy • cerebral microbleeds • cortical superficial siderosis • hypertensive arteriopathy • intracerebral hemorrhage • leukoaraiosis • MRI
Intracerebral hemorrhage (ICH) is one of the most devastating forms of stroke, with a case fatality approaching 50% at 1 month [1] and high morbidity in survivors [2]. A recent metaanalysis involving 122 studies of longer-term prognosis found a pooled 1-year survival of 46%, and 5-year survival of only 29% [3]. However, not all ICH carries such a poor prognosis: there may be a subgroup that clinically present as a lacunar syndrome (sometimes termed ‘hemorrhagic lacunar stroke’), which may account for up to 7.4% of all cases and have a higher chance of full recovery (25 vs 3.7%; p = 0.012) [4] and lower in-hospital mortality (0 vs 33%; p £ 0.005) [5] than other ICH. Identifying such subgroups may have implications for optimum treatment of ICH. ICH accounts for 10–15% of strokes in Western populations, and up to 40% in some Asian populations. [6,7]. While the incidence of ischemic stroke has fallen [8], the incidence of ICH has remained stable in recent decades [6]. Indeed, there is evidence that the incidence of informahealthcare.com
10.1586/14737175.2014.918506
ICH in the elderly and in association with oral anticoagulant use is increasing [9–11], making it a key research challenge in cerebrovascular disease. ICH describes bleeding into the brain parenchyma, which may result from a wide range of potential causes. Here, we will not discuss other patterns of bleeding (e.g., subarachnoid, subdural, extradural hemorrhage) further. Conventionally, ICH is classified as ‘traumatic’ or ‘spontaneous’ (i.e., ‘nontraumatic’). The spontaneous group is further subdivided into ‘secondary’ (due to identified causes including bleeds into tumors, cavernomas, arterio-venous malformations, CNS infection, cerebral venous sinus thrombosis, bleeding disorders, etc.) or ‘primary’, if there is no obvious underlying cause. Although widely used, the term ‘primary ICH’ has come under criticism since using this label does not enlighten regarding the true underlying cause (s), yet might encourage a spurious diagnostic certainty and a failure to pursue further
2014 Informa UK Ltd
ISSN 1473-7175
661
Review
Wilson, Charidimou & Werring
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Normal brain
• Age • Vascular risk factors, e.g., hypertension • Genetic factors (e.g., ApoE)
Bleeding-prone small vessel diseases (e.g., hypertensive arteriopathy, CAA)
?? Lobar ICH
Acute precipitants? • e.g., severe hypertensive crisis, severe coagulopathy
• Acute precipitants – Hypertension variability – Minor head trauma – Recreational drugs • Chronic risk factors – Persistent hypertension – Antithrombotic drugs
Deep ICH
Figure 1. Aetiology of intracerebral hemorrhage. There is no single cause for ICH but rather an interplay between genetic and environmental factors. Vascular risk factors, age and genetics may all play a part in the development of bleeding-prone arteriopathies. Acute precipitants such as a minor head injury or hypertension then can cause small vessel to rupture. Antithrombotics may further increase potential risk or severity of ICH by impairing haemostasis (for example causing a normally self-limiting microbleed to enlarge into a symptomatic macro haemorrhage). CAA: Cerebral amyloid angiopathy; ICH: Intracerebral hemorrhage.
investigations. With cumulative advances in imaging and histopathological correlates, it is now generally accepted that the main processes underlying so-called ‘primary ICH’ are intrinsic diseases affecting cerebral small vessels (generally a few hundred microns and up to about 1–2 mm), usually collectively termed small vessel disease (SVD). A recent large international consensus group standardized definitions for neuroimaging markers of SVD and suggested the term ‘spontaneous ICH presumed to be due to SVD’ as preferable to primary ICH [12]. Challenges in assessing the ‘cause’ of spontaneous ICH
A discussion of the nature of causation is beyond the scope of this review, but a ‘cause’ can most simply be defined as something that affects the prevalence, likelihood or clinical effect of a disease. For ICH, contributory causes include the underlying SVD processes described above, but also ‘risk factors’ (e.g., hypertension, diabetes, lipid profile, smoking, alcohol use, etc.), which may influence the clinical expression of these SVD. The challenge with many studies of ICH (particularly cross-sectional) is that they are able to show associations, but cannot provide proof (or direction) of causality. Whether an association reflects causation can be considered according to the strength of association, consistency, 662
specificity, dose–response relationship, biological plausibility and consistency with natural history of the disease [13]. SVD is highly prevalent in older populations [14], yet ICH is much less common. Thus, as in other types of stroke, spontaneous ICH is likely to result from an interplay between environmental and individual patient (e.g., genetic) factors relating to the expression of SVD. Indeed, recent data suggest that genetic variation plays a significant role in ICH risk and outcome [15]. It was estimated that 44% of ICH risk variance was accounted for by genetic risk factors, with a greater contribution of genetic factors (especially apoE alleles) to lobar ICH than deep ICH [15]. One model of ICH causation is that multiple acute or chronic risk factors (e.g., age, sustained hypertension or shortterm blood pressure fluctuations, antithrombotics, serum cholesterol levels or statin use, minor head trauma, etc.) interact with vulnerable damaged small vessels (subject to the influence of genetic or other individual patient factors), which, when a certain threshold is exceeded, ultimately rupture to culminate in ICH as represented in FIGURE 1. Indeed, a recent populationbased study suggested that ICH may result from short-term increases in blood pressure prior to the event (over weeks to months), by contrast, with ischemic stroke where blood
Expert Rev. Neurother. 14(6), (2014)
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Advances in understanding spontaneous ICH
Vascular amyloid-β deposition in CAA
Review
Hypertensive arteriopathy
Figure 2. Bleeding prone small vessel vasculopathies. Schematic of the typical locations of the bleeding prone vasculopathies. Hypertensive arteriopathy: affecting mainly the deep arterial perforators by causing arteriolosclerosis, lipohyalinosis, and fibrinoid necrosis. CAA: accumulation of amyloid-b peptide in the walls of small-to-medium-sized arteries and arterioles predominantly located in the leptomeninges and the cerebral cortex. CAA: Cerebral amyloid angiopathy. Reproduced with permission from [181,182].
pressure is more stable [16]. This finding suggests that consistent long-term blood pressure control may be especially important in reducing the risk of ICH. Types of SVD
Most age-related small vessel damage is secondary to one of two main SVD processes (FIGURE 2): an arteriolar process often related to aging and other common vascular risk factors (e.g., hypertension and diabetes), characterized pathologically by lipohyalinosis, arteriolosclerosis or fibrinoid necrosis and typically affecting the small perforating end-arteries of the deep grey nuclei and deep white matter (often termed ‘ hypertensive arteriopathy’ ); and sporadic cerebral amyloid angiopathy (CAA), a disease process affecting superficial cortical and leptomeningeal vessels through the deposition of amyloid b (Ab). Less commonly, ICH occurs in the context of much rarer genetic diseases (familial cerebral amyloid angiopathies, collagen 4A1 mutations, Fabry’s disease, CADASIL, CARASIL, etc.) or cerebral and systemic vasculitides, which will not be discussed in this review. A classification of SVDs with their relevance to ICH is shown in TABLE 1. Sporadic cerebral amyloid angiopathy
Sporadic CAA is characterized by the progressive accumulation of Ab within the walls of small- to medium-sized arteries and arterioles – and, to a lesser extent, the capillaries and veins – in the leptomeninges and cerebral cortex [17]. In the most severe form, CAA-affected small vessels become thickened and disrupted, with focal wall fragmentation and blood extravasation, with or without microaneurysmal dilatation, and sometimes informahealthcare.com
show luminal occlusion [17]. The pathophysiology of sporadic CAA is poorly understood, but transgenic mouse models suggest an increased ratio of Ab40: Ab42 in the brain results in a shift of Ab from brain parenchyma to the vasculature [18] (perhaps by increasing the solubility of Ab and thus its diffusion into the vessel wall); and that vascular Ab deposition largely results from impaired clearance of Ab (rather than overproduction) [19]. The prevalence of CAA increases with age, especially over 60 years [20,21]. Autopsy studies have shown CAA in more than 70% of a healthy population above the age of 90 [22,23]. CAA may vary according to ethnicity: some studies suggest a higher predilection for amyloid deposition in the frontal lobe arteries in Eastern populations [24]; it has also been suggested that Eastern populations have a lower prevalence of CAA, but this may in part reflect the higher relative prevalence of hypertension and hypertensive arteriopathy in this population [24]. By contrast with other cerebrovascular diseases, CAA does not appear to be related to conventional vascular risk factors. Although hypertension may aggravate bleeding risk in CAA [25], most patients with CAA-related ICH (up to 68% in one study) are not hypertensive [26,27]. Anticoagulation or antiplatelet treatments may also increase the likelihood of CAArelated ICH [28–31], but, apart from genetic variants, there are no other known strong risk factors that increase the presence of CAA or its bleeding risk. apoE alleles are associated with CAA. A recent meta-analysis showed a dose-dependent association between apoE e4 alleles and the presence of sporadic CAA [32]. A more recent study, taking into account CAA severity, found a possible association of severe CAA with apoE e4 but not 663
Review
Wilson, Charidimou & Werring
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Table 1. Types, histopathology and prevalence of cerebral small vessel disease. Name
Histology
Prevalence
Strength of relationship to intracerebral hemorrhage
Arteriosclerosis Common sporadic SVD Hypertensive arteriopathy
Arteriolosclerosis Lipohyalinosis Fibrinoid necrosis Microatheroma Microaneurysms
Very common
Common cause of deep or lobar ICH
Cerebral amyloid angiopathy
Ab amyloid protein Dilated and disrupted walls Double-barreled arterial wall
Age related 7% those aged 65–74 18% those aged 75–84 [20] 70% those aged over 85 [23]
Common cause of lobar ICH
Inherited or genetic SVD CADASIL COL4A1 Fabry’s disease CARASIL
Depends on underlying disease
CADASIL >2 per 100,000 COL4A1 very rare prevalence unknown Fabry’s disease 1 per 120,000 CARASIL very rare, unknown
CADASIL – ICH rare (CMBs common) COL4A1 – ICH common Fabry’s disease – ICH rare CARASIL – ICH rare (CMBs common)
Inflammatory and Immunologically mediated SVD. (primary angiitis CNS and systemic vasculitis)
Typical of vasculitis
Very rare
ICH common
Other (radiation, etc.)
Depends on etiology
Rare
Rare
ICH: Intracerebral hemorrhage; SVD: Small vessel disease. Modified with permission from [17].
apoE e2, but was limited by the number of participants with e2 genotypes [33]. In addition, there have been multiple studies linking both apoE alleles (e2 and e4) to the risk of recurrent hemorrhage [34–37]. apoE e2 has been shown to confer greater vessel fragility [38]: a large genetic association study has shown carriers of apoE e2 had increased ICH expansion, mortality (odds ratio [OR]: 1.50; 95% CI: 1.23–1.82) and poorer functional outcomes (modified Rankin scale score 3–6; OR: 1.52; CI: 1.25–1.85) compared with non-carriers after lobar ICH. By contrast, apoE e4 was not associated with lobar ICH volume, functional outcome or mortality [39]. A number of other genetic variants are also associated with CAA-related ICH, for example, complement component receptor, TGF-b1 and TOMM40, but further work is needed to confirm these [35]. New genetic associations with CAA are sure to emerge from large international collaborative efforts in the coming years. There is substantial evidence supporting the hypothesis that CAA is an important contributory factor in causing ICH. This association between CAA and ICH was noted as far back as the 1970s and 1980s in a series of pathologically confirmed case reports of ICH [40–44]; more recently a large systematic review and meta-analysis of published histopathological studies confirmed an association between CAA and lobar ICH (OR: 2.21; 95% CI: 1.09–4.45) [45]. Given the very high pathological prevalence of CAA in population-based studies, and the much lower incidence of CAA-related ICH, the vast majority of patients with CAA pathology do not suffer ICH. Elucidating which patients with CAA will develop CAA-related ICH is a key question for preventing CAA-related ICH. A pathological study comparing brains with CAA and ICH with those with CAA without ICH 664
found CAA was more severe in the brains with cerebral hemorrhage than in those without, and that fibrinoid necrosis was seen only in the brains with cerebral hemorrhage [46]. Microaneurysms occurred only in the presence of severe, rather than moderate or mild, CAA [47]. This suggests that mild CAA may not confer such a high risk of ICH, and that preventing progression of CAA may thus reduce the incidence of ICH. Serum & CSF markers in CAA
Circulating serum and CSF biomarkers may be another way to detect and monitor CAA, but results to date are inconsistent. Decreased levels of Ab40 and 42 in the CSF in patients with CAA have been reported compared with both controls and patients with Alzheimer’s disease (AD) [48]. Another study [46] found that in CAA cases CSF total and phosphorylated tau were significantly lower in CAA than AD, but total tau was higher than healthy controls. Ab42 levels were significantly decreased in CAA when compared with controls and marginally lower than patients with AD, while Ab40 was significantly lower in CAA when compared with both AD and controls. Ab(1–42) and truncated fragments Ab(N–42) were higher in probable CAA patients than in controls (p < 0.001 and p = 0.046, respectively) [49]. Ab(N–42) was higher in those patients with multiple cerebral microbleeds (CMBs) when compared with those with only one CMB, suggesting that circulating Ab fragments in the peripheral circulation may only be found with more severe CAA pathology. Further studies of CSF markers of CAA are needed to confirm and clarify associations of Ab levels, and to explore other potential markers relevant for amyloid or neurodegenerative Expert Rev. Neurother. 14(6), (2014)
Advances in understanding spontaneous ICH
pathways (e.g., soluble amyloid precursor protein a and b, total tau, phospho-tau, neurofilament light and ferritin). Promising biomarkers should be correlated with well-established imaging markers of CAA, cognitive status and clinical outcome over time to determine their sensitivity to disease progression.
Expert Review of Neurotherapeutics Downloaded from informahealthcare.com by National University of Singapore on 05/24/14 For personal use only.
Treatment of CAA
With no effective disease-modifying treatment for CAA, trials to date have focused on reducing the risk of recurrent ICH by influencing modifiable risk factors. Results from the PROGESS trial showed a striking reduction in presumed CAA-related ICH by 77% (95% CI: 19–93%) for those patients treated with antihypertensives [25], but this study did not use MRI to classify the likely cause of ICH and requires confirmation in other trial populations of ICH survivors. Reducing blood pressure is now the mainstay of secondary prevention treatment, but the optimum treatment target is not known. There is an urgent need for further trials to improve outcome after CAArelated ICH, for example, intensive long-term blood pressure lowering, use of statins and antithrombotic drugs. A Phase II randomized disease modification trial of the safety and efficacy of the monoclonal antibody ponezumab in CAA has been conducted. Ponezumab is a humanized monoclonal antibody originally used in AD trials. Unlike other monoclonal antibodies used in AD trials, ponezumab did not cause antibodyinduced microhemorrhage or other amyloid-related imaging abnormalities. Although this agent lacked efficacy in AD, it may be more effective in CAA as the amyloid burden is vascular in CAA (thus not protected by the blood–brain barrier, potentially allowing easier binding of the antibody to antigen) than in AD. The study is a Phase II, randomized, double-blind placebo-controlled trial to evaluate the safety, tolerability, pharmacokinetics and efficacy of ponezumab, with saline being used as the placebo. The primary end point of this study is an increase in cerebrovascular reactivity measured by visual BOLD fMRI, which is a potentially reversible marker of amyloid-related vascular functional compromise [50,51]. Hypertensive arteriopathy
Hypertensive arteriopathy is a term often used to describe multiple different (non-CAA) pathologies affecting mainly the deep arterial perforators (including arteriolosclerosis, lipohyalinosis and fibrinoid necrosis) [52–54], some of which are not clearly directly related to hypertension [55,56]. Sporadic SVD is a more encompassing term [55], but may also include CAA depending on how a small vessel is defined. The most direct histopathological correlate of hypertensive injury is fibrinoid necrosis, which is much more commonly found in brains from hypertensive patients than in those without hypertension [57–59] as well as in arterioles adjacent to deep ICH [58–61]. Furthermore, a study from the late 1970s showed that marked elevation in blood pressure can produce fibrinoid necrosis acutely [62]. Historically, hypertension has been considered the cause of the majority of non-lobar (i.e., deep) ICH. This observation first arose from autopsy series from the late 50s and early 70s of patients with deep ICH, in which classical pathological vessel informahealthcare.com
Review
changes suggested hypertensive arteriopathy [58,60]. These presumed hypertensive-related changes, including microaneurysms (‘Charcot-Bouchard’ or ‘miliary’ aneurysms) occurred in small penetrating arteries and arterioles subjected to high pressures such as the lenticulostriate arteries emanating from the middle cerebral artery, basilar and posterior cerebral artery perforators. These small arteries feed deep central areas of the brain such as the basal ganglia, thalami and brainstem [63] corresponding to the common locations of deep ICH. Thus, deep ICH has for many years been commonly attributed to hypertension and termed ‘hypertensive hemorrhage’. This theory has been challenged, primarily by the finding of a low prevalence of hypertension in some ICH cohorts [64,65] as well as pathological studies in which findings previously attributed to hypertension have been observed in the absence of hypertension [66] and associations of deep perforator SVD with other factors (e.g., reduced cerebral perfusion) [55,67–73]. Nevertheless, for simplicity and consistency, we will use the term ‘hypertensive arteriopathy’ to include nonCAA SVD affecting the small deep perforating arteries. Genetic studies have shown associations with the occurrence and outcome of deep ICH presumed due to hypertensive arteriopathy. A recent study showed a strong association between size of hemorrhage, clinical outcome and known hypertension risk alleles in those with deep ICH [74]. This study demonstrated an increase in hematoma volume by 28% and an increased risk of poor clinical outcome by 71% for every standard deviation rise in a blood pressure-based genetic risk score. Attributing lobar ICH to CAA and non-lobar ICH to hypertensive arteriopathy, while attractive for clinicians, is clearly an oversimplification. There is evidence from a recent large systematic review and meta-analysis showing lobar hemorrhages are significantly associated with CAA [45], but there was no statistically significant negative association between CAA and deep ICH. Furthermore, hypertensive arteriopathy can affect the white matter perforators and thus cause a proportion of lobar ICH in addition to deep ICH [75]; importantly, older individuals may in fact have a mixture of both CAA and hypertensive arteriopathy, as both are age-dependent [17]. Treatment
Treatment of hypertensive arteriopathy has focused on treating hypertension. The original PROGRESS trial showed the addition of perindopril-based antihypertensive medication regimen reduced the risk of all types of ICH by 50% (95% CI: 26–67%) in those with a previous stroke or transient ischemic attack [76]. Further analysis of this trial has shown that blood pressure lowering can reduce risk of ICH attributed to hypertensive arteriopathy (i.e., non-CAA-related ICH) by 46% (95% CI: 4–69%) [25], which, interestingly is a smaller treatment effect than seen in presumed CAA-related ICH. The reduced apparent effectiveness of lowering blood pressure in hypertensive-related ICH compared with CAA-related ICH likely reflects the notion that not all ‘hypertensive arteriopathy’ is related to hypertension, and not all ICH attributed to possible or possible CAA in this study was actually solely due to CAA, since advanced MRI was not used to 665
Review
Wilson, Charidimou & Werring
has increased fivefold [11], which closely corresponds to the fourfold increase in warfarin use (per capita) over the same period [11]. Restricted Anticoagulants should not ‘cause’ ICH diffusion per se but it is hypothesized that in vulnerable individuals (e.g., those with SVD) they Cortical lead to larger ICH volumes due to impaired superficial hemostasis. If this is the case then patients Leukoaraiosis siderosis with ICH on anticoagulants should have Deep CMBs larger hematoma volumes, and more hematoma expansion, than those not taking oral anticoagulants. The available studies are summarized in the online SUPPLEMENTARY TABLE 1 (supplementary material can be found online at www.informahealthcare.com/ suppl/14737175.2014.918506). Indeed, ICH anticoagulation is generally associated with larger baseline hemorrhages [63,79–84] and Figure 3. Small vessel disease. Schematic of small vessel disease markers seen on MR hematoma growth [83–85]; all but one of the modalities. CMB: Cerebral microbleed. studies with over 100 patients [86] show a Reproduced with permission from [183]. significant relationship between anticoagulation and hematoma size. One study revealed establish ICH etiology. Further large clinical trials are needed to an association between international normalized ratio intensity of confirm the effect of intensive blood pressure lowering on the anticoagulation and the volume of deep bleeds, but not for lobar risk of recurrent ICH. bleeds [79], suggesting differing underlying pathology (e.g., amyloid angiopathy and hypertensive arteriopathy) in different brain regions may respond differently to anticoagulation. However, this finding Changing epidemiology of spontaneous ICH & the role has not been replicated, and further studies are warranted. Three of anticoagulation The epidemiology of ICH has changed over the years with an studies to date show that hematoma expansion is increased by antiaging population, more aggressive treatment for hypertension coagulation; two of these studies were observational hospital based and increasing use of anticoagulation. It may well still be studies [84,85] without standardized CT timing scanning. changing: with the advent of the novel oral anticoagulants, we may expect to see an even higher use of anticoagulants in Neuroimaging markers in ICH elderly patients. Although these drugs have lower ICH risks Neuroimaging, particularly with MRI, is the most useful way than warfarin, the increased use may still overall lead to a to visualize the consequences of SVD. Although the small arterhigher prevalence of ICH in the elderly population. This is of ies are generally beyond the resolution of MRI, their complex particular importance as higher age is associated with worse out- effects on the brain (from both ischemia and hemorrhagic procomes [77,78]. One paper comparing ICH in the very elderly cesses) can be clearly seen. Here, we consider each neuroimagfound those aged greater than 85 years had increased incidence of ing marker in terms of its diagnostic and prognostic utility in neurological deficit at hospital discharge (89 vs 58%; p < 0.005) regard to future ICH. A schematic illustration of the main and in-hospital mortality (50 vs 27%; p < 0.01) than those youn- MRI markers of the SVDs relevant to ICH are shown ger than 85 years [77]. However, the cutoff age could be consid- in FIGURE 3 and example images are shown in FIGURE 4. ered arbitrary, and there was no significant difference in either neurological deficit or mortality when comparing those aged White matter hyperintensity (leukoaraiosis) greater than 85 to those aged between 65 and 85. Leukoaraiosis is a broad term originally used to describe confluWhile global ICH rates have remained stable [6], the improve- ent areas of low density on CT scans, but subsequently applied ments seen from improved hypertension control seem to be offset to high signal areas on T2-weighted MRI scans. More recently, by the increasing number of anticoagulation and CAA-related these changes on T2 and FLAIR MRI have been termed hemorrhages [9–11]. Large cohorts, each over a 20-year period from ‘White matter hyperintensities of presumed vascular origin’ Oxfordshire and Dijon have shown a decrease in the incidence of (WMH) [12]. WMH has a heterogeneous pathological substrate ICH in the younger age group (
