Approach

Management decisions are made in the light of the lifetime risk of haemorrhage versus the risk of treatment. The risk of arteriovenous malformation (AVM) rupture is reduced only by complete exclusion of the AVM from the intracranial circulation, and not by partial resection/obliteration. Treatment is therefore highly individualised and dependent on the angioarchitecture, the location of the AVM, the age and comorbidity of the patient, and the relative risks of different treatment modalities for that particular treating centre.[59]

A European consensus statement concluded there is sufficient indication to treat Spetzler-Martin grade 1 and 2 AVMs with an intention to cure, while the decision to treat patients with higher grades is on a case-by-case basis.[60] Surgery can be done in a semi-elective setting in patients with previous AVM-related haemorrhage, progressive neurological deterioration from steal syndrome, or epilepsy that is resistant to antiepileptic drugs.[61]

Associated haematoma and hydrocephalus

In patients with a ruptured AVM, emergent surgical evacuation of the intracerebral haematoma and control of acute bleeding may be required.[61] Simultaneous resection of small (≤3 cm) superficial AVMs can be attempted during the emergency operation. Resection of deep or complex AVMs should be deferred and undertaken as a semi-elective procedure.[61] During an acute ICH episode, blood pressure lowering and reduction in blood pressure variability can reduce haematoma expansion with improved functional outcomes.[45]​ See Stroke due to spontaneous intracerebral haemorrhage.

Hydrocephalus secondary to intraventricular rupture of the AVM may require treatment with an external ventricular drain.

Not suitable for surgery

Very large AVMs in eloquent locations (areas of the brain that control speech, motor function, and senses) with deep venous draining veins from the intracranial circulation should be managed conservatively with symptomatic treatment of the effects of the AVM such as seizure control.

Occasionally, palliative embolisation can be offered with the aim of reducing shunt volume in the nidus to control seizures or reduce focal hypoxia ('vascular steal').

Surgical candidates

In patients with AVMs amenable to treatment, the principal treatment modalities are surgical resection, stereotactic radiosurgery, and embolisation.[61]

Often a combination of two or more modalities is required to completely obliterate an AVM. Which treatment modalities to use should be decided in a multidisciplinary setting. Following any intervention, angiography should be performed to either confirm complete obliteration or plan the next stage of treatment.

Surgical resection

Surgical resection without embolisation may be the only treatment modality required for small, superficially placed AVMs in non-eloquent locations. Larger AVMs are more likely to require multimodality treatment.

A craniotomy is performed to expose the AVM, which is removed using standard microsurgical techniques to circumferentially excise the nidus. Feeding arterial vessels are sacrificed to the nidus itself using bipolar diathermy forceps and microscissors until the nidal draining veins are completely dearterialised. Once this has been achieved the draining vein may be taken and the nidus removed. Intraoperative neuronavigation is often used to localise the AVM nidus; alternatively, where a superficial arterialised draining vein is present on the cortical surface, this can be followed into the nidus.

Stereotactic radiosurgery

Patients with AVMs that are not surgically accessible, or in whom the overall risk of surgery outweighs that of other treatment modalities, may require treatment with stereotactic radiosurgery (SRS) with or without embolisation. European consensus guidelines consider eloquent location of an AVM to be a strong indication to consider SRS.[60]

SRS using either linear accelerator-based (LINAC) radiosurgery or the 'gamma knife' enables precise delivery of a high dose of radiation to a small intracranial target while sparing the surrounding normal brain. It is usually given as a single dose. Although non-invasive, the procedure does carry risks. In particular, LINAC radiosurgery takes between 2 and 5 years to obliterate the AVM, so the patient is at risk of rebleeding during this period.[62]

The success of SRS is inversely correlated to the size of the nidus. Typically, AVMs with a diameter of less than 3 cm (volume <120 cm³) are suitable for SRS, and effective rates of obliteration of the AVM of up to 80% can be achieved.[59] Small size, non-eloquent location, low-flow pattern, and absence of perinidal angiogenesis are predictors of obliteration by radiosurgery.[14] The use of SRS specifically in Spetzler-Martin grade 1 and 2 AVMs appears to achieve obliteration in 80% of patients, with post-treatment haemorrhage occurring in 6%.[63]​​

Larger lesions may be amenable to staged treatment: that is, treating different anatomical components of the AVM at intervals staged between 3 and 6 months.[64] Staged SRS of large AVMs may reduce adverse effects of radiation: in one study, staging of SRS did not affect temporary adverse radiation effects, but permanent adverse effects fell from 15% to 6.5%.[65]

AVM-associated aneurysms are strong predictors of post SRS haemorrhage. It is recommended to treat AVM associated aneurysms via microsurgery or endovascular therapy before SRS to reduce risk of haemorrhage.[61]

Embolisation

Smaller AVMs with few arterial feeders are most amenable to curative embolisation. However, the cure rate with embolisation alone is moderate, with an average of 20% with n-butyl cyanoacrylate (n-BCA) in older studies, and up to 50% with newer embolic agents.[59] Larger AVMs usually require planned, often staged, embolisations followed by surgical excision or SRS for any residual AVM. 

A detailed angiographic analysis of the arteries supplying the AVM, supplemented if necessary with superselective angiography, is an essential precursor to treatment planning. Embolisations are generally performed under general anaesthesia through a femoral artery approach. There has been interest in using a venous approach, and more aggressive venous embolisation with controlled hypotension has been described, but this has not yet been widely adopted.[66]

n-BCA is a fast-polymerising liquid adhesive embolic agent. However, its use has been largely supplanted by the Onyx liquid embolic system, which is less adhesive and polymerises slowly, allowing for a more controlled embolisation of the nidus.[67] Other liquid embolics, such as precipitating hydrophobic injectable liquid (PHIL) and squid (a non-adhesive liquid embolic agent composed of ethylene vinyl alcohol copolymer), are also available.[68][69] Regardless of choice, there is a risk of reflux of the embolisation agent into a feeding artery, which can result in stroke, and early obliteration or thrombosis of the draining veins can lead to periprocedural AVM rupture.[67][69] There appears to be a 4.5% recurrence rate after embolisation of AVMs, so patients with obliterated AVMs need to be followed up with angiography at regular intervals.[70]

Embolisation may form part of a multimodality approach, before surgical excision or stereotactic radiosurgery. When subtotal embolisation before surgical excision of an AVM is planned, the aim is to reduce the risks associated with surgery by targeting areas that are difficult to reach with open surgery.

Pre-stereotactic radiosurgery embolisation

Systematic reviews and meta-analyses report lower AVM obliteration rates in patients who have undergone embolisation followed by SRS than in those who have undergone SRS alone.[71][72][73] Increased treatment failure in patients who received pre-SRS embolisation may be attributable to several causes: a failure to account for differences in AVM characteristics between patients who underwent embolisation followed by SRS and those who had SRS alone (most studies are non-randomised and retrospective); patients with complex AVMs being more likely to be candidates for pre-stereotactic radiosurgery embolisation; embolisation agents causing significant imaging artifact, thereby obscuring AVM visualisation; and recanalisation after embolisation.[74][75][76][77][78]

The specific goals of pre-SRS embolisation include making SRS feasible by reducing the nidus volume, and minimising bleeding risk in the latency period by embolising weak elements in the angioarchitecture of the nidus, such as flow-related aneurysms or high-flow fistulas.[79][80] The embolisation should aim to produce a compact, stable nidus.

AVMs associated with intranidal or extranidal aneurysms or arteriovenous fistulas may be resistant to radiosurgery, and have a higher incidence of perioperative haemorrhage.[81] When performed by experienced surgeons, embolisation prior to radiosurgery may be considered for carefully selected patients with large, complex AVMs.[82]

Unruptured AVM

There is no evidence that invasive treatment for unruptured AVMs is beneficial.[83]​ Haemorrhage risk seems to be overestimated in patients without haemorrhagic presentation (<1%), and the risks of treatment may outweigh the risk of rupture.[22][36]​ One multicentre study reported a significantly lower risk of death or stroke among patients with unruptured brain AVM who were randomised to medical management compared with those randomised to neurosurgery, embolisation, or SRS, alone or in combination (hazard ratio 0.27, 95% confidence interval 0.14 to 0.54).[84] Outcome data were available for 223 patients with a mean follow-up of 33.3 months when the trial was stopped by the National Institute of Neurological Disorders and Stroke-appointed data and safety monitoring board. The composite endpoint of death or symptomatic stroke was reached in 10.1% of patients in the medical arm versus 30.7% of patients in the interventional arm.[84] A subsequent analysis of the 5-year follow up data showed that these differences in the composite endpoint persisted.[85]

However, the study has been widely criticised for numerous methodological limitations, including the systematic error caused by the abbreviated follow-up period, and a probable selection bias prior to randomisation, with a large number of eligible patients not enrolled, and therefore a lack of generalisability of the trial results.[86]​ Moreover, none of the patients had AVMs with a Spetzler-Martin grade of more than 4, and only 18 patients had surgery, while 30 patients had embolisation alone, which at that time was not widely considered a curative procedure. In addition, the follow-up period was relatively short for a disease that harbours a lifetime risk of haemorrhage. The study did, however, confirm a low spontaneous rupture rate (2.2% per year) for unruptured AVMs. A larger study is underway.[87]​ 

Further uncertainty exists regarding a sub-population of AVMs that have been classified as 'unruptured' in most studies. As many as 30% of 'unruptured' AVMs show evidence of prior silent intralesional microbleeds that may be predictive of further, more serious, rupture.[88][89] Newer imaging techniques are being evaluated for their ability to define this potential sub-population.

Pregnancy and labour

Management of pregnancy and labour in women with AVMs requires a multidisciplinary team. Risk of intrapartum intracranial bleeding is considered low if the AVM is fully treated or intracranial bleed occurred more than 2 years ago.[90] Women at low risk of intracranial bleed can base decisions on mode of delivery based on their usual preference and obstetric indications. Risk of intrapartum intracranial bleed is high if the mother has an untreated or complex AVM or haemorrhagic episode in the past 2 years. Mothers at high risk of intracranial bleed should be offered the option of caesarean section after full discussion of the benefits and risks of each option. Women at high risk who prefer to attempt vaginal birth should be offered regional analgesia and offered the option of assisted second stage of delivery.[90]​ 

Use of this content is subject to our disclaimer