Vascular Malformations of the Brain




(1)
Nuffield Department of Surgical Sciences, Oxford University, Oxford, UK

 




Preamble

This tutorial covers the topic of cerebral vascular malformation first in terms of their description and classification, and then their aetiology and natural history. Our understanding is limited by the relative rarity of these lesions and the bias of study from the perspective of a single centre or individual practice. Uncertainty about the nature of the most important type of vascular malformation is reflected in its utilitarian name, brain arteriovenous malformation (BAVM). This tells us little more than that arteries and veins (of the brain) are malformed.

In the first part of this tutorial, the emphasis is on the nature of cerebral vascular malformations. We could spend considerable time discussing attempts to classify vascular malformations of the brain, and we would still be unable to agree on a completely inclusive system. The problem is due to their individual variations and the need for definitions that accommodate extreme atypia. The tutorial will therefore attempt to balance good nosology with pragmatism in order to organise our thinking about clinical managements. But I hope readers will be left with a sense of uncertainty about the aetiology of these lesions and remember that despite their different names they are probably closely related.

A more pressing problem for endovascular therapy is the current controversy over interventions for people with BAVMs that have not bled. Its resolution remains a challenge because of the potential confounding effect of phenotypic variability. With this in mind, the data on which we base our advice and management protocols for patients diagnosed with these lesions should be seen as ‘work in progress’. As with unruptured intracranial aneurysms, natural history data, however good, is only a best guess of the individual’s prognosis.


9.1 Types of Vascular Malformations of the Brain


A discussion of the definitions of arteriovenous malformation (AVM) used by different writers is a good starting point because it highlights the ‘usual’, and a trap for the novice is not being able to recognise that a particular case is exceptional. Without a clear idea of commonly encountered features, detailed descriptions of all known variations of a disease risk creating misconceptions. Concentrating on the infrequent exceptions can bias perception, analogous to the blind man describing an elephant only after feeling its trunk.


9.1.1 Early Descriptions of Vascular Malformations of the Brain


There were case descriptions in the nineteenth century by Virchow [1] and Steinheil in 1895 [2]. Cushing and Bailey [3] described the pathology of a small series in 1928 and Bergstrand et al. [4] reported the first demonstration of AVM by angiography in 1936. Like aneurysms, the introduction of cerebral angiography led to rapid advances in the diagnosis, understanding and treatment of vascular malformations. Hilding Bergstrand, a Swedish pathologist, described vascular malformations of the brain as angiomas and separated them from hypervascular tumours (e.g. angioblastoma) [4]. He divided them into cavernous and racemose types on the basis that the vessels of the cavernous angioma connected without separating parenchyma. Whereas, racemose angiomas (the term means vessels arranged as a cluster) were separated by parenchyma. He further divided racemose angiomas into (a) telangiectasia, (b) Sturge-Weber disease, (c) angioma racemosum arteriale, (d) angioma racemosum venosum and (e) aneurysma arteriovenosum [5]. We would recognise the last type (called arteriovenous aneurysms by Olivecrona) as arteriovenous malformations.


9.1.2 Definitions of Arteriovenous Malformations


Several authors have proposed definitions of arteriovenous malformation of the central nervous system. These include:



  • Doppman 1971: AVMs are tangled anastomoses of blood vessels of varying calibre in which arteriovenous shunting occurs in a central nidus (Latin, nidus, nest), which is the area towards which one or multiple feeding arteries converge, and from which enlarged veins drain [6].


  • Valavanis 1996: Cerebral AVMs are inborn errors of vascular morphogenesis caused by a defect or malfunction of the embryonal capillary maturation process and resulting in the formation of abnormal arterial, venous or capillary channels with or without shunt [7].


  • The Arteriovenous Malformation Study Group 1999: Brain arteriovenous malformations are a complex tangle of abnormal arteries and veins linked by one or more fistulas [8].

These are useful in setting out the component features on which there is a consensus, i.e. arterial and venous feeders, a nidus and arteriovenous (AV) shunting. The last definition though includes fistulas and therefore AV connections without an intervening nidus, i.e. direct arteriovenous fistulas (AVF), described as cavernous by Bergstrand [4]. Thus, we can separate them from the definition of AVMs only if the nidus is absent. The 1996 Valavanis’ definition includes lesions without a shunt. So, the situation becomes more difficult and the process takes us further away from a description of a ‘usual’ lesion.


9.1.3 Classifications of Cerebral Vascular Malformations


Against this background, several classifications of cerebral vascular malformations have been proposed more recently. McCormick in 1966 defined five groups of cerebral vascular malformations: capillary telangiectasia, venous angioma, varix, cavernous angioma and arteriovenous malformation [9]; all subsequent authors have based their classifications on this and a more recent example is that of Chaloupka and Huddle [10], shown in Table 9.1. This separates lesions that grow from those that don’t, though it would be better stated as ‘don’t generally grow’ because examples of enlargement of lesions in this group have been described.


Table 9.1
Classification of Chaloupka and Huddle [10]





























Benign proliferating vascular anomalies: Haemangioma

Nonproliferating vascular anomalies:

 Capillary malformation [telangiectasias]

 Venous malformation

 Cavernous malformation [cavernoma]

 Arterial malformation [angiodysplasia and aneurysm]

 Arteriovenous shunting malformation

  Brain AVM

  Brain AVF

  Dural AVM

  Vein of Galen AVF

Mixed malformation

This classification includes vascular tumours, as proposed by Mulliken et al. [11] and a mixed lesion category, thus, reflecting a shift towards considering transitional lesions as a part of a continuum of vascular developmental abnormalities.

Having strayed into taxonomy, this part of the tutorial now refocuses on the four main divisions of vascular malformations included in the McCormick classification, though its main subject will be the brain arteriovenous malformation (BAVM) because they are the most common lesion types to be referred to endovascular therapists. Dural fistulas, vein of Galen malformations and non-Galenic arteriovenous fistulas are covered in separate tutorials.


9.2 Capillary Telangiectasia


These lesions are usually diagnosed post mortem and are typically found on the pial or immediate subpial surface of the brain stem or pons (Fig. 9.1).

A209602_2_En_9_Fig1_HTML.jpg


Fig. 9.1
Telangiectasia. Vertebral angiogram showing a collection of small irregular arteries on the left side of the brainstem, supplied by the anterior superior cerebellar arteries. Histological confirmation of this lesion was not possible


9.2.1 Pathology


The macroscopic appearance of cerebral telangiectasia is of a small collection of vessels on the cerebral surface. Microscopically, they are composed of thin-walled capillaries without smooth muscle layers or elastic lamina. The surrounding brain is normal.


9.2.2 Aetiology


This is unknown but multiple lesions occur in hereditary haemorrhagic telangiectasia (HHT) or Rendu-Osler-Weber syndrome, which is an autosomal dominant disorder characterised by multisystem vascular dysplasias in which those affected develop multiple telangiectasias. Telangiectasia occurs on skin, and in the mucosa of the nose and mouth, lung and gastrointestinal tract. Patients present with recurrent nosebleeds or other episodes of bleeding. The cerebral vascular abnormalities associated with HHT include brain AVMs, cavernous malformations and aneurysms [12]. Cerebral lesions were found on MRI in 23% of HHT patients screened for brain lesions [13].

In patients with multiple BAVMs reported by Willinsky et al. [14], the most frequent cause was HHT. In a series of 638 patients with cerebral AVMs, there were 14 (2%) patients with HHT; of these 50% were multiple, 42% of lesions were less than 1 cm (micro-AVM) and 29% were AVFs with high flow and venous ectasia [15]. Most of the patients were young.

HHT is caused by mutations in elements of the transforming growth factor-β (TGF-β) receptor complex [16]. The genes responsible for the commonest types of HHT (types 1 and 2) have been identified as ENG on chromosome 9q (which encodes for endoglin) and causes HHT type 1 and ALK1 (activin receptor-like kinase 1) on chromosome 12q, which causes HHT type 2 [17]. A relationship with cavernous malformations (see below) has also been proposed since telangiectasias may be found within cavernous malformations and be a mechanism for their enlargement.


9.2.3 Epidemiology and Natural History


Telangiectasias are not usually seen on catheter angiograms and therefore rarely diagnosed in vivo, so there is no useful epidemiological data concerning their frequency or natural history. However, Lee et al. [18] reported 18 patients diagnosed by MRI and calculated a prevalence of 0.4%. None of the patients developed haemorrhage during 3 years of follow-up.


9.3 Developmental Venous Anomaly (Venous Angioma)


The developmental venous anomaly (DVA) or venous angioma is recognised as an abnormally prominent collection of medullary veins, which drain to a single trunk (Fig. 9.2). The latter has, in the past, been termed a varix, but this term should be dropped since it is now generally agreed that these lesions are caused by abnormal venous development.

A209602_2_En_9_Fig2_HTML.jpg


Fig. 9.2
Developmental venous anomaly. MRI (coronal T1W sequence) showing an enlarged vein in the right frontal lobe. Small tributary veins are seen arising in the deep white matter and from the margin of the ventricle


9.3.1 Pathology


On microscopy, they have a thin endothelium, with thin smooth muscle cell and elastic tissue layers within a wall that is mainly composed of collagen but may be thickened by hyalinisation.


9.3.2 Aetiology


The conclusion that they are caused by a development failure of a normal section of the cerebral venous system is based on their having been identified in neonates and their nonprogressive behaviour. The initiating event probably occurs at about the third week of intrauterine life during formation of medullary veins. A malconnection of deep and superficial veins also occurs in Sturge-Weber syndrome, and DVA may be a variation of the same failure of normal development [19]. Crucially, the DVA drains normal brain and therefore should not be embolised.


9.3.3 Epidemiology and Natural History


They are found in up to 2.5% of autopsies and are the most common vascular abnormality of the brain. They comprised 63% of McCormick’s series of 165 malformations found at 4669 autopsies [20]. They are rare causes of spontaneous haemorrhage. Their natural history was studied by Garner et al. [21], who reported a haemorrhage in only 1 of 100 patients followed for 14 years. They calculated a lifetime haemorrhage risk of 0.22%. Other presentations were seizure, headache and transient deficits.

They are increasingly recognised as incidental findings on MRI, alone or in association with cavernous malformations. Since intervention is not indicated (except for the rare situation of an emergency evacuation of haematoma), their importance to the endovascular therapist is that they should be correctly diagnosed so that embolisation can be avoided.


9.4 Cavernous Malformations


These are hamartomatous lesions containing thin-walled vessels with more circumscribed borders than capillary telangiectasias and without intervening normal brain (Fig. 9.3). There are no associated feeding arteries or veins, so they are not detectable on catheter angiography or CT if non-calcified but are easily demonstrated on MRI because they contain haemosiderin.

A209602_2_En_9_Fig3_HTML.jpg


Fig. 9.3
Cavernous malformation. On CT performed after contrast enhancement (a) calcification is seen. T2W MRI (b) shows the typical appearance of this lesion and a gradient echo sequence (c) shows additional lesions because of the magnet susceptibility effect of blood breakdown products


9.4.1 Pathology


The macroscopic appearance is a capsulated blood-filled tumour resembling a grape. On microscopy, they are composed of channels, which have an endothelial lining and thin fibrous adventitia without elastin, smooth muscle cells or the other elements of mature vessels. They contain old haemorrhage, haemosiderin, calcification or cholesterol crystals. The walls may show hyaline thickening and the adjacent brain gliosis. Endothelial leakage causes microhaemorrhages [22]. They can occur anywhere in the brain but are most frequently found in subcortical white matter, periventricular white matter, the pons and the external capsule.


9.4.2 Aetiology


Sporadic and familial forms are recognised. Multiple lesions probably always occur at some stage in the familial disease, and they are found in at least 30% of patients without a positive family history. The genetic basis of the familial disease has been recently linked to chromosomes 7 and 3. The condition has an increased frequency in Americans of Mexican descent, in whom a defective CCM1 gene has been described. Associations have also been described with capillary telangiectasia, DVAs and previous radiotherapy [23].


9.4.3 Epidemiology


There is no gender difference, and patients may present at any age, but most do so in the second to fourth decades. Reported estimates of prevalence in the general population are 0.5% at autopsy [24] and 0.4% on MRI [25]. They account for 5–16% of central nervous system vascular malformations.


9.4.4 Symptoms and Natural History


Diagnosis is by MRI and the various possible appearances were classified as four types by Zabramski et al. based on signal characteristics [26]. Small lesions can be distinguished from telangiectasia by their lack of enhancement after gadolinium administration. Asymptomatic cavernous malformations may appear de novo and enlarge or regress on serial MRI studies [27]. Symptoms are presumably initiated by enlargement. These are, in descending order of frequency, seizures, focal neurological deficit, headache and haemorrhage. Symptoms are often progressive, in a stepwise fashion, and presentation is more commonly due to haemorrhage in children and seizures in young adults.

When discussing rates of haemorrhage, symptomatic bleeding has to be distinguished from the more common asymptomatic lesion demonstrable on MRI. Symptomatic haemorrhage at presentation has been reported in 10–26% of patients and annual rates of symptomatic bleeding on follow up calculated as 0.7–1.6% [25, 28]. Bleeding is usually intraparenchymal and only rarely subarachnoid or intraventricular.


9.4.5 Treatment


Since there is no endovascular access, these lesions are not referred to the endovascular therapists for treatment. Management is generally conservative and intervention only for symptomatic lesions. Interventions are surgical resection, which is generally reserved for symptomatic large accessible lesions, or focused radiotherapy. Stereotactic radiotherapy is performed for control of seizures and is generally not considered effective in preventing symptomatic haemorrhage [29].


9.5 Brain Arteriovenous Malformations (BAVM)


The definitions of this lesion had been discussed above but can be summarised as a vascular lesion composed of an abnormal tangle of vessels (nidus) with pathologic shunting of blood flow from the arterial to the venous tree, without a normal intervening capillary bed (Figs. 9.4, 9.5 and 9.6). The nidus of BAVMs occurs in supratentorial brain (85–90%) or the cerebellum (10–15%) and involves superficial (70%) or deep (30%) structures of the brain. They vary in size from micro-AVM (<1 cm) (Fig. 9.9) to large lesions (>6 cm).

A209602_2_En_9_Fig4_HTML.jpg


Fig. 9.4
Sulcal brain arteriovenous malformation. The lesion lies in the posterior temporal lobe and is shown on internal carotid angiograms as the nidus starts to fill (a) and after shunting to cortical vein (b). The arrows on (b) indicate veins


A209602_2_En_9_Fig5_HTML.gif


Fig. 9.5
Brain arteriovenous malformation. Internal carotid DSA in the frontal (a) and lateral (b) projections. This typical wedge shaped nidus (a) is seen in arteriovenous malformations which extend into white matter. The origin (foot) of the draining vein is marked (arrows)


A209602_2_En_9_Fig6_HTML.jpg


Fig. 9.6
Brain arteriovenous malformation of the corpus callosum. Sagittal T2-weighted MRI (a) and lateral DSA (b) show the nidus in the corpus callosum which drains to a massively enlarged internal cerebral vein. On (b), this vein and a cortical vein on the medial surface of the frontal lobe are marked (long arrows). There is also a nidal aneurysm (short arrow) shown on the DSA


9.5.1 Pathology


The macroscopic findings are of a variety of vessels ranging from well-differentiated arteries and veins to highly malformed, hyalinised, poorly differentiated vessels with thick or thin walls. The abnormal vessels are variably dilated with saccular aneurysms or narrowed with segmental or focal areas of stenosis.

Microscopy of the feeding arteries shows irregular endothelium and elastic layers with vacuolisation and necrosis of smooth muscle cells, invasion of the adventitia by foreign cells and small blood vessel as well as changes in the mural matrix. Aneurysms presumably develop in areas where the elastic tissue and smooth muscle is thin or absent. In some areas, the vessel walls are thickened by medial hypertrophy, collections of fibroblasts and thickening of the basal lamina and interstitial tissue. These changes are induced by high blood flow and shear stress.

The nidal vessels are indeterminate as either artery or vein. Draining veins may resemble arteries because of hypertrophy of smooth muscle and media, but they are distinguished by the lack of an organised elastic lamina. They may be massively dilated with ectasia and varices. Any intervening brain is typically gliotic, and gliosis may be seen in the immediately adjacent brain. Calcification may be seen in the vessel walls, and haemosiderin staining in the surrounding tissue indicates prior bleeding.

Locations: There are several systems in use for describing BAVMs by their location within the brain, and the advent of MRI and 3D reconstruction of angiograms has helped to show the relationship of feeding vessels and the nidus to normal brain structures. Valavanis and Yasargil [30] developed a system based on the principal location of nidal vessels. This divides superficial BAVMs into sulcal (nidus located in the subpial space within a sulcus), gyral (nidus completely surrounded by a cortical mantle) and mixed sulcal-gyral lesions. Deep lesions are found in the subarachnoid space (within basal cisterns or fissures), the parenchyma (within deep nuclear structures) and the ventricle (originating in choroid plexus). Individual lesions can involve one or more of these compartments, i.e. mixed. An alternative system proposed by Lasjaunias, Berenstein and Ter Brugge [31] recognises similar deep and superficial locations but emphasises the role of the feeding arteries and draining veins. Thus, lesions confined to cortex are defined as being supplied exclusively by cortical arteries and veins whilst subcortical lesions are supplied by cortical arteries but may drain to both superficial and deep veins. Cortico-ventricular BAVMs are supplied by both perforator and cortical arteries and drain to superficial and deep veins. This classification separately defines cortico-callosal BAVMs as not being supplied by perforated arteries and, like Valavanis and Yasargil, choroid and deep BAVMs with centrally directed (deep) venous drainage.

The angioarchitecture of BAVMs has been extensively studied, principally to identify features that relate to the risk of bleeding. These will be discussed below. One final point on the macroscopic appearance of lesions and their location is the effect of a low pressure, high blood flow shunt on adjacent vessels. Superficial lesions are not infrequently additionally supplied by transpial arteries. These are generally considered to be secondary and recruited as a result of the primary pial-based arteriovenous shunt. They are sometimes found to ‘appear’ after partial embolisation but whether this is a response to the abnormal haemodynamics of the shunt is uncertain and needs to be confirmed from longitudinal observational data, which currently remains incomplete. Another feature, which is presumed to be secondary and may be stimulated by partial treatment, is angiomatous change in arteries adjacent to the nidus, often termed neovascularisation. These vessels though tortuous and appearing similar to those of the nidus should not be confused with them because they show normal contrast transit times on angiography. Cure of the AV shunt leads to their disappearance, but our understanding of their aetiology and the underlying haemodynamic factors is incomplete [32].


9.5.2 Aetiology


The majority of lesions are thought to be ‘congenital’ rather than acquired. Implicated acquired causes are trauma (which includes surgery) and ionising radiation, but the links are vague. Congenital causes involve an embryological cause or a genetic cause. Postulated theories can be summarised as involving a congenital predisposition, which is triggered by extrinsic factors. The observations that any theory has to explain are that AVMs are rarely (if ever) identified in the foetus and that the adult phenotype is unusual in children. AVMs have been reported to appear in previously normal (on imaging) brain and recur after successful surgery or radiotherapy [33]. The vast majority of patients don’t give a positive family history.

Yasargil (1987) postulated a proliferative capillaropathy [34], whilst Mullan et al. [35] suggested that an initiating event occurs in foetal life (possibly due to failure of regression of pial-dual veins at the 40–80-mm stage) and that the changes are too small to be detected at birth. An abnormality of normal capillary development has been proposed by Lasjaunias [36], and Mullan, in a second paper [37], postulated they were caused by a form of venous angioma.

If you accept that they are developmental anomalies, it is then likely that the combination of a genetic predisposition and extrinsic factors leads to their formation. First the congenital defect expressing itself as the malformation and then a vascular response to the presence of the malformation. As far as a genetic factor is concerned, we have the observation of the occasional report of patients with a positive family history and the associations of BAVMs with HHT and conditions such as Wyburn-Mason and Sturge-Weber syndromes.

The focus of research has been to identify abnormalities of gene expression in the cells of AVMs and genetic studies of familial case pedigrees. Rhoton et al. [38] found that the preproendothelin-1 gene is locally repressed in HHT lesions. This results in a lack of endothelin-1 peptide, which is a vasoconstrictor with a role in vascular cell growth.

A more recent finding is increased endothelial expression of VEGF-R receptors Flk-1 and Flt-1 in surgically resected BAVM vessels compared to controls [39]. This finding leads to the attractive theory that these agents contribute to the maintenance and slow growth of BAVMs.

To explain their role, we have to remember the embryology of vascular development. Vascular morphogenesis is a two-stage process in which angioblasts differentiate into endothelial cells to form the primary vascular plexus (see Tutorial 1). The second stage is angiogenesis when the primary vascular plexus undergoes remodelling and organisation including recruitment of periendothelial cell support. Two main systems are involved in these processes:


  1. 1.


    Vascular endothelial growth factors (VEGF-R1, VEGF-R2), which mediates endothelial cell proliferation, migration, adhesion and tube formation.

     

  2. 2.


    Angiopoietins 1 and 2 (a group of cytokines) and their receptors Tie-1 and Tie-2, which play an important role at the later stage of vascular development.

     

These findings suggest a link between what happens in embryonic life with the growing realisation that BAVMs ‘appear’ towards the end of childhood and ‘grow’ in response to haemodynamic drivers. Once a shunt is present, the altered haemodynamics literally drive the required alterations in gene expression. Genetic expression research probably holds the key to this puzzle. The role of epigenetic mechanisms was recently reviewed by Thomas et al. [40], who propose that BAVMs result from aberrant epigenetic modifications in the genome of endothelial cells.


9.5.3 Demography


Brain AVMs affect men and women equally [41]. Earlier studies reported a small male dominance: Crawford et al. [42] 1.2:1 and the Cooperative study [43] 1.1:1, but these reports included all types of cerebral vascular malformations. No racial variation has been reported in the largest series, but anecdotally, brain AVMs are considered more common in China than Japan. The commonest age at diagnosis is consistently reported as between 20 and 40 years. Deruty et al. [44] reported the age distribution at presentation in their series as 33.5% (<30 years), 49% (31–50 years) and 15.5% (>50 years).


9.5.4 Epidemiology


Brain AVMs are relatively rare lesions. Their prevalence in the general population of Scotland (>16 years) has been estimated as 16–18/100,000 based on a retrospective study [45]. The incidence, i.e. proportion of the population newly diagnosed over a 1-year period, can only be calculated with complete ascertainment of cases from a population of known size. It is therefore difficult to obtain. Al-Shahi and Warlow reviewed this subject in 2001 [45], and they calculated an incidence of approximately 1/100,000, which was largely based on data from two large studies. These were a study performed over 10 years in the Dutch Antilles [46] which found incidences of 1.1/100,000 person-years and a longer study performed in Olmstead County, Minnesota [47], which found an incidence of 0.82/100,000 per year. Autopsy studies, which theoretically provide more reliable data on prevalence, have reported a range of 0.04–0.60% (i.e. up to 600/100,000) [4850] and 3% in autopsies performed after cerebral haemorrhage [49]. This wide discrepancy has stimulated further research.

Two prospective population studies in Scotland and New York have produced interim reports. The Scottish study documented incidences of 0.56/100,000 per year for BAVMs compared with 0.43/100,000 per year for cavernous malformations and an overall detection incidence of 1.2/100,000 per year for all cerebral vascular malformations [51]. The New York study found higher detection rates for BAVMs at 1.34/100,000 a year and an incidence of haemorrhage of 0.51/100,000 per year [52].

The frequency of the diagnosis of BAVMs as the cause of first ever presentation with stroke was found to be 1.4% in the Lausanne Registry [53]. This is similar to a 10-year study of hospital admissions in Northern California which reported the detection rate as 1.4 (95% CI 1.3–16) per 100,000 person-years [54]. BAVMs are more likely to be the underlying cause in younger patients and a prospective population study found that they accounted for about 3% of stroke in young adults [55]. This rate is probably an underestimate since approximately 10% of all stroke is haemorrhagic [56, 155] and about 15% of spontaneous cerebral haemorrhage is due to BAVMs [49, 156].


9.5.5 Natural History


In 1949, Olivacrona [57] wrote ‘in the end, probably most, if not all patients (with arteriovenous malformations) die of haemorrhage or are completely incapacitated’. This statement has influenced medical management for over 50 years but is it correct?


9.5.5.1 Symptoms at Diagnosis


Most patients present after spontaneous intracranial haemorrhage or the onset of seizures. The relative rates for symptoms at diagnosis are 50–60% with haemorrhage and 25–30% with seizures [56, 58, 59]. Of the rest, about 10% present with focal neurological symptoms or signs without haemorrhage and about 3–5% present with migraine or other types of headache referable to the BAVM [60]. Headache may be co-incidental and there remain a proportion of asymptomatic people, estimated at up to 15% of all diagnosed BAVMs [45]. In children (<16 years), bleeding is the most common presentation and accounts for 30–50% of haemorrhagic stroke in this age group. Epilepsy tends to present in younger patients: 44% in the second decade, 30% in the third decade and only 6% in the 30–60-year age group [61]. BAVM is not uncommonly discovered during pregnancy. Crawford [42] reported that 25% of women in the third decade of life were pregnant at diagnosis. AVMs are diagnosed in 20–50% of women presenting with spontaneous intracranial haemorrhage during pregnancy and accounts for 5–12% of maternal deaths. The incidence is highest early in the third trimester, and vaginal delivery is not considered to increase the risk of haemorrhage [62].

Patients presenting with seizures alone are more likely to be young with cortical lesions in the temporal lobe [63]. Seizures amongst BAVM patients may occur in association with haemorrhage, so the presence of other symptoms are important in considering the likelihood of future epilepsy. Crawford [42] reported an overall rate of epilepsy in 18% of unoperated patients over 20 years and rates of 22% in patients presenting with haemorrhage, 44% for the age range 10–19 years and 37% for temporal lobe lesions. The Scottish Audit of Intracranial Vascular Malformations reported that patients presenting with a first-ever seizure developed epilepsy over the next 5 years in 58% (95% CI 40–76%) and 8% of those presenting without seizure, haemorrhage or neurological deficit experienced at least one seizure [64].

Though focal neurological deficits without haemorrhage are uncommon at presentation, subsequent progressive neurological deficits are common in patients on observation and usually attributable to the effects of repeated haemorrhage. However, there are several possible causes for progression of neurological deficits without haemorrhage, including secondary effects of epilepsy, focal brain compression by dilated vessels, vein thrombosis, steal effects on adjacent brain and chronic venous hypertension. Mass effect is relatively frequently observed on MRI and may be related to the size of the nidus and vessels [65]. A decrease in cerebral blood flow has been demonstrated in the brain surrounding AVMs, but the functional significance is uncertain [66, 67]. Because the potential effects causing deficits are so various, it is difficult to generalise and often difficult to draw firm conclusions in practice.


9.5.5.2 Risk of Haemorrhage


The diagnosis of a brain AVM makes the individual at risk for future adverse events. The important issue for deciding how to advise patients about treatment is the risk of future bleeding since this is the major contributor to acute and long-term morbidity associated with the diagnosis. In what is probably the best clinical paper written on this subject, Crawford et al. [42] showed that without surgical treatment, the risk of death was 29%, risk of haemorrhage 42%, risk of neurological handicap 27% and risk of epilepsy 18% during a mean follow-up period of 10.4 years for symptomatic AVM.

Observational studies of untreated AVMs have estimated an annual rate of haemorrhage of 2–4% per annum; Crawford 2.5% [42], Brown 2.2% [47] and Ondra 4% [60]. Kim et al. reported a meta-analysis of untreated patients and calculated an overall annual rate of haemorrhage of 2.3% (95% CI 2.0–2.7) [68]. In Crawford’s cohort, patients presenting with haemorrhage were found to have an increased risk of subsequent haemorrhage compared to patients presenting without haemorrhage (36 vs. 17% over 10 years). In the Kim pooled data, the haemorrhage annual rates were 1.3% (95% CI 1.0–1.7) for unruptured and 4.8% (95% CI 3.9–5.9) for ruptured BAVM patients at presentation [68]. Graf et al. [69] reported that this increased risk was mainly during the first year after bleeding. A range of factors that different reports have recognised as either being associated or not with haemorrhage are listed in Table 9.2. The most consistent factor is the occurrence of prior haemorrhage.


Table 9.2
Reports of factors associated with haemorrhagic episodes






































































 
Significant association with haemorrhage

Contradictory data

Age

Increasing

Crawford et al. [42]

Graf et al. [69]

Mast et al. [56]

Stapf et al. [70]

Halim et al. [71]

Sex

Male

Mast et al. [56]
 

Size

Small

Khaw et al. [72]

Mast et al. [56]

Graf et al. [69]

Stefani et al. [73]

Crawford et al. [42]

Stapf et al. [70]

Kader et al. [74]

Langer et al. [75]

Large

Stefani et al. [73]
 

Presenting symptom

Previous haemorrhage

Halim et al. [71]

Stefani et al. [73]

Kondziolka et al. [76]

Mast et al. [56]

Pollock et al. [77]

Headache

Kondziolka et al. [76]

Mast et al. [56]

Hypertension

Langer et al. [75]
 

Brown et al. [78] followed patients with unruptured BAVMs and used Kaplan Meier curves to calculate the actuarial risk of haemorrhage as follows:













1.3% per year at 1 year

1.7% per year at 5 years

1.5% per year at 10 years

2.2% per year at 15 years

These data suggest that the risk of bleeding is steady overtime. On this assumption, Kondziolka et al. [76] constructed a probability of bleeding equation as follows:



$$ \begin{array}{c}\mathrm{Risk}\;\mathrm{of}\;\mathrm{bleeding}=1-{\left(\mathrm{risk}\;\mathrm{of}\;\mathrm{no} \mathrm{bleed}\right)}^{\mathrm{x}}:\\ {}\left(\mathrm{X}=\mathrm{expected}\;\mathrm{years}\;\mathrm{of}\;\mathrm{life}\right).\end{array} $$




$$ \begin{array}{l}\mathrm{Based}\;\mathrm {on}\; \mathrm{an}\;\mathrm{annual}\;\mathrm{risk}\;\mathrm{of}\hfill \\ {}\kern2em \mathrm{bleeding}\;\mathrm{of}\ 3\%,\mathrm{i}.\mathrm{e}.\kern0.5em 0.03\hfill \end{array} $$




$$ \begin{array}{l}\mathrm{Chance}\;\mathrm{of}\;\mathrm{remaining}\;\mathrm{bleed}\;\mathrm{free}\;\mathrm for\ 1\ \mathrm{year}\hfill \\ {}\kern2.5em =1-0.03=0.97.\hfill \end{array} $$




$$ \begin{array}{l}\mathrm{Chance}\;\mathrm{of}\;\mathrm{remaining}\;\mathrm{bleed}\hfill \\ {}\kern2.25em \mathrm{free}\; \mathrm for\ 2\ \mathrm{years}={0.97}^2=0.94.\hfill \end{array} $$




$$ \begin{array}{l}\mathrm{Chance}\;\mathrm{of}\;\mathrm{remaining}\;\mathrm{bleed}\hfill \\ {}\kern2.25em \mathrm{free}\; \mathrm for\ \mathrm{x}\ \mathrm{years}={0.97}^{\mathrm{x}}.\hfill \end{array} $$




$$ \begin{array}{l}\mathrm{Therefore}\;\mathrm{the}\;\mathrm{probability}\hfill \\ {}\kern3.25em \mathrm{of}\;\mathrm{bleeding}=1-{0.97}^{\mathrm{x}}.\hfill \end{array} $$
This exercise has been refined by Brown who proposed a formula in which the lifetime risk (%) = 105 − patient’s age in years. From these equations, risk/benefit tables can be constructed to assist patients and physicians in calculating individual risk rates [79].


9.5.5.3 Risk of Death in Patients with BAVM


The long-term crude annual case fatality rate has been estimated at 1–1.5% by Al-Shahi and Warlow [45]. The observed combined morbidity and mortality in observational studies is between 3% and 4% per annum. In Brown et al.’s patients [47], the mortality rate was 3.5% per annum (29% over 8.2 years in the patients who haemorrhaged). The overall combined incidence of major morbidity and mortality was 2.7% in Ondra’s patients [60] followed for a mean period of 23.7 years. This risk was higher in patients who had haemorrhaged (85% versus 34%), and this trend was also seen in Brown’s series [47] with a 23% risk of significant morbidity in survivors of haemorrhage compared with 7% in those with no haemorrhage. Fatality rates of 5–10% per episode of bleeding [45, 69] are not as high as after aneurysm subarachnoid haemorrhage, but up to a third of survivors are left with a permanent disability after each event. Bleeding is therefore the major factor in the poor long-term prognosis of this diagnosis.

The mean age of death in Ondra’s patients was 51 years (compared with 73 years in the general Finnish population) [60]. Thus, BAVMs reduce the life expectancy of patients, and the issue that has to be resolved in recommending treatment is how the natural history compares with the risk of iatrogenic morbidity following intervention. The risks of both rebleeding and death are at least a factor of 2 higher in the patients who haemorrhage and at 3% annually, sets the level of risk that justifies recommending direct intervention to cure BAVMs.


9.5.6 Angioarchitecture


Identifying the vessels supplying a BAVM is obviously crucial to management. Traditionally, this has depended on catheter angiography (DSA) and this continues to provide both anatomical and flow data. We now have the option of contrast-enhanced planar scanning for angiography and multimodality imaging (e.g. MRA/CTA/Flat Detector CT) and reconstructions of 3D and 4D data with image fusion techniques. These can reliably separate arteries and veins from nidal vessels to provide a complete anatomical display of the angioarchitecture. Additionally, MRI techniques, such as phase contrast sequences and arterial spin labelling, can interrogate blood flow patterns and pathways within the nidus (Fig. 9.7). Several authors have emphasised the additional value of superselective angiography (i.e. injection of individual arterial pedicles) to the definition of a lesion’s angioarchitecture but this is now unnecessary, except as part of an embolisation procedure.

A209602_2_En_9_Fig7_HTML.jpg


Fig. 9.7
Brain arteriovenous malformation with reconstructed flow data following partial embolisation. A reconstructed lateral view from a phase-contrast MRI has been coloured to show areas of decreased blood flow velocity (green) and intravascular NBCA in red (a). The 2D DSA (b) is shown for comparison. Note that some NBCA has migrated to the distal portion of the large draining vein and is shown in red on (a) and at the arrow on (b)


9.5.6.1 Identification of Prognostic Features


Various authors have described features of the angioarchitecture that were associated with haemorrhage. These are shown in Table 9.3. The list seems to grow with each report, as does the degree of controversy over particular features.


Table 9.3
Angioarchitecture factors associated with haemorrhagic episodes




































































































 
Significant association with haemorrhage

Contradictory data

Location of nidus

Deep

Crawford et al. [42]
 

Stefani et al. [73]

Ventricular/periventricular

Nataf et al. [80]
 

Miyasaka et al. [81]

Basal ganglia

Brown et al. [82]
 

Posterior fossa

Khaw et al. [72]
 

Stapf et al. [70]

Brown et al. [82]

Venous drainage

Single vein

Nataf et al. [80]
 

Miyasaka et al. [81]

Fewer veins

Todaka et al. [83]
 

Stefani et al. [73]

Deep

Kader et al. [74]

Halim et al. [71]

Only deep

Khaw et al. [72]
 

Duong et al. [84]

Mast et al. [56]

Brown et al. [82]

Marks et al. [85]

Nataf et al. [80]

Pollock et al. [77]

Miyasaka et al. [81]

Stapf et al. [70]

Stefani et al. [73]

Langer [75]
 

Ectasia

Stefani et al. [73]

Nataf et al. [80]

Stenosis

Mansmann et al. [86]

Marks et al. [85]

Nataf et al. [80]

Arterial feeder

High pressure

Todaka et al. [83]

Henkes et al. [87]

Leblanc et al. [88]

Duong et al. [84]

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Aug 17, 2017 | Posted by in NEUROSURGERY | Comments Off on Vascular Malformations of the Brain

Full access? Get Clinical Tree

Get Clinical Tree app for offline access