Cerebral Arteriovenous Malformations And Their Surgical Treatment Background Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Arteriovenous malformations (AVMs) are haemodynamically active, fast-flowing vascular malformations composed of a knotted anastomosis of arteries and veins (McCormick 1978). There is a central nidus upon which the supplying arteries converge, and from which the draining veins leave (McCormick 1966). Brain AVMs are located solely within the brain parenchyma, thus distinguishing them from other AVMs. They are morphologically different from cavernous and venous intracranial vascular malformations, and also have a different prognosis and response to medical interventions (Al-Shahi et al. 2001; Tu et al. 2009). Brain AVMs are thought to occur due to a combination of both environmental factors and genetic polymorphisms including in the transforming growth-factor-beta related genes ALK-1 and ENG (Lasjaunias 1997; Kim et al. 2008), although the precise cause remains unknown. Molecular growth factors including basic fibroblast growth factor (bFGF) and also vascular endothelial growth factor (VEGF) are thought to be important in their pathogenesis (Moftakhar et al. 2009). Single nucleotide polymorphisms in IL-6, tumour necrosis factor-alpha, and apolipoprotein-E have been shown to be associated with AVM haemorrhage (Kim et al. 2008).

The most frequent presentation of an AVM is brain haemorrhage, but the data about average age at which bleeding occurs and the risk of bleeding remains inconsistent (Graf et al. 1983; Crawford et al. 1986; Stapf et al. 2003). Overall, brain AVMs account for approximately 4% of intracerebral haemorrhages (Al-Shahi et al. 2006); however they are thought to be responsible for up to 33% of intracerebral haemorrhages in young adults (Al-Shahi et al. 2006). Furthermore they are responsible for up to 2% of strokes and can cause focal and secondary generalised seizures (Perret et al. 1966).

AVMs affect both males and females equally, with a prevalence of about 18 per 100 000 adults (Al-Shahi et al. 2006). A significant proportion of AVMs remain asymptomatic, however, if they do cause symptoms, they tend to do so before 40 years of age, usually in adolescence, but they can present in early childhood. Recent advances in brain imaging and catheter angiography has increased detection of brain AVMs (Brown et al. 1996).

Clinical Course of AVMs

The most common clinical presentation of an AVM is intracranial haemorrhage, which occurs with a frequency of between 30 and 80% (Brown et al. 1996). Evidence from autopsies suggest that as little as 15% of cerebral AVMs are symptomatic during life (McCormick 1978). However, most AVMs will bleed at least once if left untreated, with an overall haemorrhage risk of approximately 2-4% per year which is also associated with a 50% chance of permanent neurological deficit and up to 10% chance of mortality (Graf et al. 1983; Ondra et al. 1990; ApSimon et al. 2002). Whilst the estimated mortality from AVM haemorrhage is thought to be 10-15%, the morbidity rate may be under 50% (Samson et al. 1991). Patients with parenchymal bleeds may have a worse prognosis than patients with an intraventricular or subacrachnoid haemorrhage (Hartmann et al. 1998).

In general, if the AVM is located deep within the brain or if there is an associated aneurysm, drainage to deep venous sinuses, venous stenosis or a single draining vein then the chance of haemorrhage is significantly increased (Spetzler et al. 1992; Pritz 1994; Redekop et al. 1998; Al-Shahi et al. 2001; Stapf et al. 2006). In addition, it has been shown that high pressures within the AVM nidus is also associated with an increased risk of bleeding (Duong et al. 1998). Small AVMs are believed to be at increased risk of bleeding in comparison to larger AVMs, although this risk may be overestimated (Mast et al. 1997; Stapf et al. 2006).

There is an estimated annual bleeding rate of 2-4% from AVMs, which was obtained by Ondra et al in a 24 year study of untreatable brain AVMs (Ondra et al. 1990). However most patients experienced only mild disability following the AVM bleeding, and this result is concordant with data from other studies (Brown et al. 1988; Auger et al. 1992; Pollock et al. 1996). In comparison, the annual rate of AVM rebleeding following an initial presentation of haemorrhage is thought to be around 18% (AVM-Study-Group 1999), hence AVM haemorrhage is considered a significant risk factor for subsequent bleeding (Brown et al. 1988; Pollock et al. 1996). The risk of AVM rebleeding is thought to be greatest during the first year after the first AVM bleed (Itoyama et al. 1989; Mast et al. 1997). This risk is then thought to decrease rapidly, however some studies have not found an association between the history of bleeding and further clinical course (Ondra et al. 1990).

In fact AVM presentation through seizures occurs in between 20-50% of symptomatic cases and it is well established that AVMs may cause seizures independently of haemorrhage (Mast et al. 1995). The majority of such seizures are "partial complex" in nature, with many of the remainder being tonic-clonic (Osipov et al. 1997).

Headache may be the initial presentation in 7 to 48% of patients (Mast et al. 1995). However it appears such headaches lack any distinctive features which would lead to AVM diagnosis, including headache severity, frequency, nature or headache duration (Mast et al. 1995). Atypical migraines or recurrent unilateral headaches have not be shown to be associated with AVMs (Evans 1996; Frishberg 1997).

There is also some evidence that brain AVMs may cause transient, persistent or even progressive focal neurological deficits without haemorrhage, migraine or epileptic seizure (Stahl et al. 1980; Lazar et al. 1999). Focal neurological deficits without an underlying bleed have been reported in between 1-40% of patients, although only 4-8% have well-documented and progressive neurological deficits (Mast et al. 1995). The progression of such neurological deficits was believed to be caused by as so-called 'steal phenomenon' where ischemia in brain tissue adjoining the AVM occurs due to effective hypoperfusion of nearby ateries due to 'stealing' of cerebral bloodflow by the large AVM (Carter et al. 1995; Taylor et al. 2002). However, there is little definitive evidence that cerebral-artery hypotension is actually caused by the AVM (Mast et al. 1995). Alternative explanations for the progressive neurological deficit associated with AVMs include venous hypotension and the mass effect of the AVM (Miyasaka et al. 1997).

Indications for Surgical Treatment of AVMs

Currently, a decision on the management of a patient is reached by weighing the potential risks of any surgical intervention against the risk of leaving the AVM untreated. Lee et al proposed a multidisciplinary approach to the treatment of patients with AVM, and in their cohort of 76 patients found that most AVMs were "diffuse infiltrating extratruncular AVMs accompanied by a macro-arteriovenous (AV) shunting nidus" (Lee et al. 2004). Each case was subject to an MDT review following diagnosis, and a consensus reached about treatment. Absolute indications for treatment of AVMs in general include major or minor bleeds, gangrene, arterial or venous ulceration or ischaemic complications due to acute or chronic arterial insufficiency (Lee et al. 2004; Kim et al. 2006). In addition, congestive heart failure and lesions located at 'life-threatening vital areas of the brain' which may compromise vision, hearing, breathing or eating are also consider to be absolute indications for surgical intervention (Lee et al. 2004). Suggested relative indications include disabling, reduced quality of life (which includes severe cosmetic deformity and lesions with a potentially high risk of complication (Lee et al. 2004; Kim et al. 2006).

Management of AVMs

Medical or conservative management includes the treatment of epileptic seizures and headaches, and alleviating symptoms of the AVM. However medical management does not treat the underlying pathology of this condition. The treatment of AVMs differs in adults and children, and in addition relies on the clinical staging of the AVM based on the Spetzler-Martin classification (Table 1). In accordance with this grading system, AVM grading requires determination of the apparent size of the AVM and venous drainage of the lesion. In addition, the eloquence of the surrounding brain should be evaluated using CT and MRI imaging although angiography is required to evaluate the structure of an AVM (Horton 2007). The Spetzler-Martin grading system scores from Grade I to V, and surgical resection of Grade I and II lesions is generally associated with a very low incidence of neurological deficit (Spetzler et al. 1986). In comparison, neurosurgery on Grade IV and V brain arteriovenous malformations has been shown by multiple researchers to have a high risk of significant neurological complications (Spetzler et al. 1986; Chang et al. 2003; Han et al. 2003), whilst Grade III AVMs are a heterogeneous group with variable outcomes (Lawton 2003).

Table 1: The Spetzler-Martin Grading System of Arteriovenous Malformations

Size of AVM

Eloquence of Adjacent


Pattern of Venous


Small (<3cm)




Superficial only


Medium (3-6cm)




Deep Component


Large (>6cm)


† Eloquent areas of the brain include sensorimotor, language, visual, thalamus, hypothalamus, internal capsule, brain stem, cerebellar peduncles, and deep cerebellar nuclei

‡ Superficial lesions drain entirely through the cortical drainage system

Table taken from (Spetzler et al. 1986)

The mainstay of definitive treatment for brain AVMs are interventional radiology and surgical procedures including embolisation, sclerotherapy, surgical resection and reconstruction (Marler et al. 2005). Complete elimination of the brain AVM nidus is likely to cause a reduction in case fatality and the incidence of haemorrhage or epilepsy, and is the aim of treatment (Ogilvy et al. 2001; Chen et al. 2006). However, the benefit of any therapeutic treatment may be outweighed by the risk of surgical intervention depending on the grade of lesion and its location within the brain parenchyma. Superficial, uncomplicated, small lesions which drain to non-eloquent cortical areas of the brain may be treated using microsurgical excision. Gamma knife and linear accelerator stereotactic radiotherapy or radiosurgery can be used, but it is limited to AVMs which are under 3cm in diameter and are composed of a compact nidus. Another surgical option is endovascular embolisation, which can potentially completely occlude brain AVMs or can be used as a neoadjuvant before neurosurgery and radiosurgery. In addition, aneurysm treatment can be mediated through the use of detachable coils or glue. These surgical treatments can be used in isolation or in combination to attempt to achieve complete resolution of the AVM.

Surgical Resection

The surgical resection approach involves coagulating the arterial supply to the AVM. This is followed by a circumferential dissection around the lesion, working from superficial to deep layers within the gliotic gray plane (Yasargil 1988; Drummond et al. 1990). Venous drainage is maintained by the surgeon until after the arterial supply is removed. This avoids a build up of pressure which could cause rupture of the AVM and uncontrolled bleeding (Yasargil 1988). High-resolution digital angiography can be used during the procedure to determine how much of the AVM has been resected, minimising the risk of incomplete resection of the nidus (Ellis et al.; Munshi et al. 1999).

Whilst complete surgical resection of the AVM results in immediate cure, there are significant risks associated with surgery and this technique is not appropriate in all cases. Surgical resection is considered appropriate for Spetzler-Martin Grade I and II AVMs (Ogilvy et al. 2001; Andaluz et al. 2004), and this carries a good prognosis as a single stage operation (Pikus et al. 1998). Both microsurgery and new stereotactic techniques have significantly reduced the risk associated with surgical resection of an AVM (AVMStudyGroup 1999; Ogilvy et al. 2001; Ross et al. 2010). Stereotactic guidance can also be used to localise bone and scalp flaps, enable the localisation and dissection of deep AVMs as well as planning surgical trajectories (Germanwala et al.; Wang et al. 2007).

AVMs which extend into the basal ganglia or thalamus are at increased risk of bleeding and rebleeding, and even low grade AVMs with this anatomy are treated with multistage endovascular embolisation followed by resection or radiosurgery (Sasaki et al. 1998; Paulsen et al. 1999). Grade III AVMs are also often surgically resected after embolisation (Ogilvy et al. 2001). In addition, patient age and state of health should also be considered when deciding to perform surgical resection, along with the patient's preference for treatment. It is important to completely remove the AVM, as partial resection may not only lead to relapse and offers no protection against bleeding and may in fact increase the chance of haemorrhage (Ogilvy et al. 2001; Andaluz et al. 2004; Hoh et al. 2004).

There is a relatively high risk of haemorrhage during microsurgical resection. As a result, intra-operative monitoring should be employed and adequate amounts of blood made available for transfusion (Kim et al. 2008; Saleh et al. 2008). In addition this technique poses a significant risk of damage to adjacent tissues and of ischaemic stroke (Saleh et al. 2008). It is therefore important that surgical intervention for AVM should be elective operations (Ogilvy et al. 2001) as this permits time for pre-existing medical conditions to be optimised, and any pre-existing neurological deficits to be taken into account in the peri-operative management (Hashimoto et al. 2001; Saleh et al. 2008).

Additionally, surgical resection of an AVM is associated with risks due to the anaesthesia used during the operation. No anaesthetic regimen has been shown to confer absolute cerebral protection during neurosurgery (Hashimoto et al. 2001; Ogilvy et al. 2001), but an agent must be selected which allows for excellent control of blood pressure, rapid emergence and brain relaxation along with euvolaemia, isotonicity and mild hypocapnia (Hashimoto et al. 2001; Saleh et al. 2008). Intracranial compliance may be abnormal and therefore anaesthetic agents which cause cerebral vasodilatation should be avoided (Hashimoto et al. 2001; Saleh et al. 2008). Induced hypotension is often used during resection of an AVM, especially those with a deep blood supply (Szabo et al. 1989; Langer et al. 1998). Postoperatively, bleeding from the surgical operation site or brain swelling with postoperative oedema and haemorrhage may be due to normal perfusion pressure breakthrough. Alpha-adrenergic receptor blockers may be used to prevent and treat this condition (Bloomfield et al. 1996; Hashimoto et al. 2001; Olsen et al. 2002). However this cause of bleeding or malignant brain swelling should be diagnosed by exclusion of all other correctable causes. Recently it has been demonstrated that the normal perfusion pressure breakthrough phenomenon may be due to an inappropriate neurogenic response to vasodilatation, and that this may in fact be a relatively rare event (Young et al. 1996).

The literature reports a serious adverse event risk of permanent paralysis or weakness, aphasia or hemianopia ranging from 0 to 15% (Hamilton et al. 1994; Pikus et al. 1998). There has been a low mortality rate reported in these case studies. However, surgical resection of Spetzler-Martin Grade IV or V AVMs has been found to have much greater rates of complications and death, and hence surgical resection is often not used on these grade AVMs (Hamilton et al. 1994; Ogilvy et al. 2001; Friedlander 2007).

Endovascular Embolisation

Endovascular embolisation plays a significant role in the current surgical treatment of cerebral AVMs. It is rare for endovascular embolisation to be used alone in brain AVM treatment, and generally it is used as an adjunctive treatment to surgical resection or radiosurgery (Ogilvy 1990; Sun et al. 2010). There are currently no studies which compare the effect of surgical resection with and without adjunctive endovascular embolisation, since when this technique was introduced it was almost immediately deemed to be advantageous (Ogilvy et al. 2001). Preoperative endovascular embolisation is used to decrease the risk of blood loss and surgical risk by decreasing the volume of the arteriovenous shunt and the vascularity of an AVM (Jafar et al. 1993). Multiple embolisations may be required, and this seems to depend on the size and complexity of the specific AVM (AVM-Study-Group 1999; Ogilvy et al. 2001). Large AVMs with multiple feeding arteries may need many embolisations (Valavanis et al. 1998; Ogilvy et al. 2001). In comparison, pre-radiosurgical embolisation aims to reduce the AVM nidus to a total volume less than 10ml, which makes treatment with stereotactic radiotherapy easier, as well as removing aneurysms, fistulae and varices which are potential sources of bleeding after radiosurgery (Fournier et al. 1991; Valavanis et al. 1998; Gailloud 2005; Starke et al. 2009). In addition, endovascular embolisation may be used palliatively in patients with large, inoperative subcortical or cortical AVMs (Ogilvy et al. 2001). Partial embolisation may reduce symptoms in these patients, including seizures and progressive neurological deficit (Vinuela et al. 1983; Fox et al. 1985). However improvement is usually only a temporary as collateral vessels rapidly develop, although interestingly, partial embolisation in these patients does not seem to have an effect on the long term risk of bleeding from the AVM (Ogilvy et al. 2001).

Currently there are two main methods of endovascular embolisation, divided into either solid or liquid therapeutic agents. Solid agents are comprised of fibers, microcoils, microballoons and polyvinyl alcohol particles (Fournier et al. 1991; Wallace et al. 1995; Gailloud 2005). Catheter techniques allow selective approaches to the arteries that supply the AVM and also the AVM nidus (Richling et al. 1991). Current microcatheters have the ability to guide the catheter tip combined with the pulling capacity of the blood flow by microguidewires. Advances in neuroimaging and increased knowledge of the angioarchitecture of AVMs has contributed to the increased success in using solid embolisation techniques (Valavanis et al. 1998).

Liquid embolisation agents include polymer solutions (for example ethylene vinyl alcohol) and cyanoacrylate monomers including NBCA (N-butyl cyanoacrylate) and IBCA (I-butyl cyanoacrylate), as well as absolute alcohol (Wallace et al. 1995; Gailloud 2005; Starke et al. 2009). NBCA is used with an oily dye to ensure that it is visible under x-ray, but the ratio of the dye to the glue affects the speed of polymerisation and setting. This variability, along with operator controlled variation in speed of injection, led to penetration of the monomer into the venous system causing complications including an increased incidence of haemorrhage (Richling et al. 1991; Richling et al. 2006). Newer agents have now been developed, including Glubran-2 and Neuroacryl which have more controlled embolisation (Richling et al. 2006).

Recently a new polymer, ethylene-vinyl alcohol copolymer or Onyx, has emerged as a new liquid polymerising agent available for use in the treatment of cerebral AVMs. Theoretically, Onyx allows for slower filling and better penetration into the AVM (Ayad et al. 2006). Onyx also has the ability to provide a solid cast of the AVM nidus due to its lack of adherence and slower polymerisation, and thereby completely obliterate the AVM (Ayad et al. 2006; Panagiotopoulos et al. 2009). Obliteration rates with Onyx have varied in studies, with rates between 4 and 28% (Pierot et al. 2005; Tevah et al. 2005; Mounayer et al. 2007; Panagiotopoulos et al. 2009). This rate is higher than the reported rate of complete obliteration with NBCA in the literature. Complete obliteration with Onyx has been reported to be associated with a supratentorial or cortical location, a compact nidus, a small number of supplying arteries and one superficial draining vein (Weber et al. 2007; Panagiotopoulos et al. 2009). However the efficacy of Onyx as a curative treatment of cerebral AVMs needs additional evaluation with larger studies and periods of longer clinical follow-up (Panagiotopoulos et al. 2009).

Endovascular embolisation carries multiple risks, including technical complications during the embolisation as a result of the AVM architecture or blood flow within the AVM (Fournier et al. 1991; Ellis et al. 1998). Significant harm to the patient can occur with endovascular embolisation, especially when the venous drainage is occluded before the arteriovenous shunt of the brain AVM (Richling et al. 1991; Ogilvy et al. 2001). If embolisation is being used as an adjunct to surgical resection then it is important that the combined risk of both techniques is not greater than the risk of surgical resection alone.


Stereotactic radiosurgery attempts to achieve complete obliteration of the AVM within the brain whilst maintaining neurological function. A focused volume of radiation is delivered to a defined target within the brain and is usually used for AVMs that are not readily amenable to surgical resection (Ogilvy et al. 2001; Sun et al. 2010). In addition, stereotactic radiosurgery is used in AVMs which are located in deep areas of the brain, where a surgical approach is likely to be very hazardous, or in small AVMs (Kurita et al. 2000). Stereotactic radiotherapy is most effective when treating AVMs which have a diameter of under 3.5cm and is curative in 64% and 95% of these cases (Colombo et al. 1994; Friedman et al. 1995; Pollock et al. 1998). Small AVMs, young patients, few draining veins and a hemispheric location of the AVM are associated with successful stereotactic radiotherapy treatment (Pollock et al. 1998). Stereotactic radiosurgery can be used in isolation or as an adjunct therapy after surgery and/or endovascular embolisation (Maruyama et al. 2005). However, the larger the AVM the smaller the chance of cure using radiosurgery (Friedman et al. 1995; Fleetwood et al. 2002; Maruyama et al. 2005).

Stereotactic radiation delivers a relatively high dose of therapeutic radiation to the target area of the brain whilst minimising the dose of radiation which is given to other areas of the brain (Flickinger et al. 1996; Flickinger et al. 2002). Multiple beams of external radiation are focused on a stereotactically defined intracranial target area (Flickinger et al. 1996). The various beams intersect within the area of the AVM and the particle beams are engineered to deposit a well-defined maximal dose of radiation in the target area, a phenomenon known as the Bragg Peak effect (Flickinger et al. 2002). There is a dose-response relationship when using stereotactic radiation between obliteration of the AVM and toxicity (Flickinger et al. 1996; Flickinger et al. 2002), and large AVMs are treated with smaller doses to minimise toxicity although this reduces the efficacy of AVM obliteration (Flickinger et al. 1996).

Stereotactic radiosurgery causes gradual obliteration of the cerebral AVM through thrombosis induced by gradual vascular changes (Ogilvy 1990). These include vessel wall thickening within the AVM nidus and occur over a period of up to 3 years after treatment (Ogilvy 1990). However, because the changes leading to AVM obliteration are gradual, radiosurgery does not affect the risk of bleeding from an AVM in the period immediately after the procedure (Pikus et al. 1998; Pollock et al. 1998). In addition, multiple studies have found that stereotactic radiotherapy does not guarantee complete obliteration of brain AVMs (Pikus et al. 1998; Kurita et al. 2000; Pollock et al. 2003; Pollock et al. 2004; Ross et al. 2010; Sun et al. 2010).

Patients who still have a residual and patent AVM nidus despite radiosurgery should consider that this treatment modality has failed. They should therefore undergo surgical resection, endovascular embolisation or repeated stereotactic radiotherapy to eliminate the risk of bleeding from the AVM (AVM-Study-Group 1999; Ogilvy et al. 2001). Failure of stereotactic radiotherapy is associated with many factors including high or increasing Spetzler-Martin grades, decreasing treatment dose or increasing size of the AVM (Pollock et al. 1998; Horton 2007). The risks of radiosurgery depend on the size and the complexity of the AVM, as well as its specific location within the brain (Ogilvy 1990; Ellis et al. 1998; Flickinger et al. 2002). The dose of radiation given during treatment also affects the risk to the patient (Flickinger et al. 2002). Local alopecia of the scalp, seizures, oedema around the AVM, neurological deficits can occur following radiosurgery and additionally children treated using radiosurgery are at risk of developing secondary tumours later in life (Kurita et al. 2000; Ogilvy et al. 2001; Pollock et al. 2003; Ross et al. 2010; Sun et al. 2010). Unfortunately the reappearance of cerebral AVMs after apparent complete obliteration by radiosurgery has also been reported in the literature (Ellis et al. 1998).

There are 3 main methods of performing radiosurgery - using a gamma knife, a high energy linear accelerator-based (LINAC) radiosurgery or less commonly a charged particle unit (Sun et al. 2010). LINAC radiosurgery is performed most frequently; however gamma knife treatment is becoming more common in the USA (Sun et al. 2010). A recent study found that there was no difference in toxicity between LINAC and gamma knife radiosurgery (Sun et al. 2010). This study did not find a statistically significant difference in the efficacy of treatment between these techniques, but there was a trend that favoured gamma knife surgery (Sun et al. 2010).

Management of Aneurysms Associated with AVMs

Intracranial aneurysms are found in association with AVMs in 7 to 17% of patients with a cerebral AVM (Perret et al. 1966; Brown et al. 1988), although some papers have reported their presence in up to 58% of cases (Turjman et al. 1995). These aneurysms may occur on the arteries which feed the AVM, in which case small aneurysms are likely to involute after obliteration or resection of the AVM (Ogilvy et al. 2001). However, these aneurysms may also rupture after treatment of the AVM (Batjer et al. 1986). There may also be saccular intracranial aneurysms located on the Circle of Willis in individuals with AVMs (Friedlander 2007). Current recommendations suggest approaching these aneurysms at the same time as the AVM if the surgical operating field is large enough, with larger aneurysms over 5mm being clipped providing the surgical risk is acceptable (Mayberg et al. 1994). If the operating field is not large enough to approach the aneurysm at the same time as the AVM, then treating any aneurysms separately with endovascular or open surgical obliteration is suggested (Ogilvy et al. 2001; Friedlander 2007). However if there is a high risk of bleeding from the aneurysm after treatment of the AVM, then the aneurysm is often treated with microsurgical clipping or endovascular coiling before the AVM is treated (Redekop et al. 1998; Friedlander 2007).


The main goal of surgical treatment of cerebral AVMs remains the complete obliteration of the AVM, and it remains unclear whether partial resection provides a benefit to patients in cases where it is used to reduce mass effects or uncontrolled seizures. The choice of therapy used to treat AVMs depends on specific features about each case, including the presence of bleeding, pattern of venous drainage and the existence of associated aneurysms and their diameter and location. In addition the age of each patient should be taken into account, along with any co-morbidities and foreseeable complications of anaesthesia.

Current North American multidisciplinary guidelines suggest that surgical excision should be used in Spetzler-Martin Grade I and II AVMs unless the anatomy makes them unsuitable for surgery, in which case small lesions under 3cm should be treated with radiotherapy (Ogilvy et al. 2001). Embolisation should be used as part of a combined approach to obliterate cerebral AVMs before surgery or radiosurgery for Spetzler-Martin grade III-V lesions (Ogilvy et al. 2001). These guidelines also state that surgery alone is neither suitable nor recommended for grade IV and V AVMs, and that palliative embolisation is beneficial when there is a venous outflow obstruction or true 'steal' (Mast et al. 1995; Ogilvy et al. 2001). Endovascular therapy is recommended to remove deep feeding arteries in an attempt to decrease the size of the cerebral AVM, thereby making the brain AVM suitable for surgical resection or radiotherapy (AVM-Study-Group 1999; Ogilvy et al. 2001). It is also used in AVMs which are not easily resectable by surgery or are too large to be treated by radiosurgery (Ogilvy et al. 2001). It may also be an option in small AVMs where the vessels can be completely occluded by embolisation (AVM-Study-Group 1999).

However, these guidelines are based on non-randomised evidence, and there is now much controversy about treatment of AVMs (Ross et al. 2010). This is partially because of a lack of evidence about which treatment modality to use and partially because there it remains unclear whether or not the risks of surgical treatment are less than the risk posed by the AVM if it is left untreated (Al-Shahi et al. 2006; Ross et al. 2010). Non-controlled trials suggest a surgical treatment related complication rate of approximately 16%, with 7% of patients suffering permanent neurological deficits and 1% dying from complications of the surgery (AVM-Study-Group 1999). Endovascular embolisation is associated with a 13% morbidity and a mortality rate of about 2% (Gobin et al. 1996). Staged embolisation followed by surgical resection has a combined mortality risk of 4% and 9% of patients treated with this technique are left with a permanent neurological deficit (Vinuela et al. 1991). Most AVMs will bleed if left untreated, with an overall haemorrhage risk of approximately 2-4% per year associated with a 50% chance of permanent neurological deficit and up to 10% chance of mortality (Graf et al. 1983; Ondra et al. 1990; ApSimon et al. 2002). Whilst adverse treatment associated effects are lower with radiotherapy approaches to AVM treatment, this may in partly be because the effects radiotherapy are delayed and they may not be captured by studies in their follow up periods (AVM-Study-Group 1999). In addition it remains unclear at precisely what time radiotherapy should be used after presentation with a bleed, and the role of this technique in patients with repeated AVM bleeds (Pollock et al. 1996).

Whilst there are three main approaches to surgical management of cerebral AVMs, there are few published randomised controlled trials (RCT) which have assessed the clinical effects of these interventions. There is currently an ongoing RCT which is comparing interventional treatment against medical management of cerebral AVMs which have not previously bled. This trial is called 'A Randomised trial of Unruptured Brain Arteriovenous Malformations' (ARUBA) and it is an "international, multi-centre, randomised, controlled trial" which aims to compare endovascular, surgical and/or radiation therapy against medical management of AVMs in adults (Mohr et al.). The designated endpoint is a death from any cause or stroke, with secondary outcomes including risk of death or clinical impairment with clinical outcome status measured through the Rankin Scale (Mohr et al.). Patients were randomised to best possible treatment and will be followed for at least 5 years. Results from this trial are due to be published after 2012, with an additional proposed follow up at 10 years after patient recruitment (Mohr et al.). It is hoped that this RCT will finally determine whether prophylactic intervention is superior to medical management until bleeding occurs in the treatment of cerebral AVMs as measured through patient outcomes of death, stroke or functional outcome status as measured on the modified Rankin Scale (Mohr et al.).