Brain arteriovenous malformations

INTRODUCTION — Arteriovenous malformations (AVMs) are the most dangerous of the cerebrovascular malformations with the potential to cause intracranial hemorrhage and epilepsy in many cases. They have become the focus of scientific study leading to technological advances that have permitted these high-flow lesions to be treated, often with a multidisciplinary approach utilizing surgical, radiosurgical, and endovascular techniques.

This topic review will discuss brain AVMs. Three other general subtypes of congenital vascular malformations have been described: developmental venous anomalies, capillary telangiectasias, and cavernous malformations. These are discussed separately. (See "Vascular malformations of the central nervous system".)

PATHOGENESIS AND PATHOLOGY — The pathogenesis of brain AVMs is not well understood. Traditionally, brain AVMs were considered sporadic congenital developmental vascular lesions, but this notion has been disputed by many well-documented reports of de novo brain AVM formation [1-3]. The size of brain AVMs varies widely, and some undergo growth, remodeling, or regression over time [4,5]. Rare cases of familial brain AVMs have been reported but it is unclear if these are coincidental or indicate true familial occurrence [6]. However, genetic variation may influence brain AVM development and clinical course [7-9]. Somatic variants in the mitogen-activated protein kinase signaling pathway have been identified in several patients with sporadic brain AVMs [10,11].

The most common genetic cause of brain AVMs is hereditary hemorrhagic telangiectasia (HHT; Osler-Weber-Rendu syndrome), an autosomal dominant condition. Patients with HHT may have cerebral or spinal cord involvement with telangiectasias, brain AVMs, aneurysms, or cavernous malformations. The presence of more than one brain AVM, otherwise uncommon, is highly predictive of HHT [12]. (See "Clinical manifestations and diagnosis of hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)".)

The angioarchitecture of brain AVMs is direct arterial to venous connections without an intervening capillary network. Both the arterial supply as well as the venous drainage may be by single or multiple vessels. Gliotic brain is usually admixed with the vascular tangle, and calcification may be seen in the vascular nidus and surrounding brain. The high-flow arteriovenous communication potentiates a variety of flow-related phenomena such as the development of afferent and efferent pedicle aneurysms, which occur in 20 to 25 percent of patients, and arterialization of the venous limb. Aneurysms can be a source of bleeding in patients with brain AVMs and are thought to worsen their prognosis [13]. Abnormal flow and a vascular steal phenomenon have been suggested to underlie some clinical symptoms associated with brain AVMs [14]. Histopathologic studies demonstrate areas of chronic ischemia and gliosis in the region of the malformation.

EPIDEMIOLOGY — Brain AVMs are uncommon, occurring in approximately 0.1 percent of the population, one-tenth the incidence of intracranial aneurysms [4]. Supratentorial lesions account for 90 percent of brain AVMs; the remainder are in the posterior fossa. They usually occur as single lesions, but as many as 9 percent are multiple [15].

Brain AVMs underlie an estimated 1 to 2 percent of all strokes, 3 percent of strokes in young adults, and 9 percent of subarachnoid hemorrhages [16,17].

CLINICAL PRESENTATION — Brain AVMs usually present between the ages of 10 and 40 years. There are two peaks in age at presentation, one in childhood and then again at age 30 to 50. The presentation depends on symptoms produced (seizure, hemorrhage, or incidental) [18]. The patient's age, as well as the size, location, and vascular features of the AVM, influence the clinical presentation, which typically falls into one of five categories [2,4,19,20]:

●Intracranial hemorrhage (40 to 60 percent) – Hemorrhages are usually intraparenchymal, but isolated or concurrent intraventricular or subarachnoid hemorrhage may also occur, depending upon the location of the brain AVM [20]. Bleeding into the subarachnoid space is common for superficial AVMs. The clinical presentation of these events is described separately. (See "Spontaneous intracerebral hemorrhage: Pathogenesis, clinical features, and diagnosis" and "Aneurysmal subarachnoid hemorrhage: Clinical manifestations and diagnosis".)

Retrospective data suggest that initial presentation with hemorrhage may be more likely for children compared with adults (56 versus 43 percent in one study [21]).

●Seizure (10 to 30 percent) – Seizures are typically focal, either simple or partial complex, but often have secondary generalization. Patients with cortically-located, large, multiple, and superficial-draining AVMs are more likely to present with seizures [19,22]. The location of the AVM influences the seizure type and semiology. (See "Focal epilepsy: Causes and clinical features".)

●Headache (<1 to 6 percent) – There are no specific headache features that associate with AVM, which may be incidental to the headaches [4]. In one study, 0.2 percent of patients with headache and normal neurologic examinations were found to have an AVM [23].

●Focal neurologic deficit (<1 to 3 percent) – This presentation is fairly unusual for cerebral AVM. While a vascular steal syndrome has been hypothesized to cause this presentation, in most cases a focal neurologic deficit is caused by mass effect due to hemorrhage, or is a postictal effect of seizure [4].

●Incidental finding (10 to 20 percent) – A number of asymptomatic brain AVMs are discovered on imaging with brain magnetic resonance imaging (MRI) or computed tomography (CT) scan obtained for other reasons [2].

NATURAL HISTORY — An understanding of the natural history of brain AVMs, particularly in terms of hemorrhage rates, is critical to making decisions about management options.

Hemorrhage risk — A meta-analysis summarized the natural history of cerebral AVMs as reported in nine studies including 3923 patients and 18,423 patient-years of follow-up and reported an overall annual hemorrhage rate of 3 percent [24]. Similarly, in an individual patient level meta-analysis of four cohorts with 2525 patients and over 6000 years of follow-up, the overall annual rate of intracranial hemorrhage was 2.3 percent (95% CI 2.0-2.7 percent) [25], and in the observation arm of the ARUBA trial, the annual rate of ICH was approximately 2 percent [26].

Potential risk factors have been identified that appear to impact hemorrhage rates:

●Hemorrhage as the initial clinical presentation is the strongest predictor for subsequent hemorrhage in patients with untreated brain AVMs [24,25]. In the systemic review, the annual rate of hemorrhage was 2.2 percent (95% CI 1.7-2.7) for unruptured AVMs and 4.5 percent (95% CI 3.7-5.5) for ruptured AVMS [24]. In the individual patient level meta-analysis, the annual rate for unruptured and ruptured brain AVMs was 1.3 and 4.8 percent, respectively [25]. In one database, children were not at higher risk of subsequent ICH after initial hemorrhage when compared with adults [21]. In another series, clinically silent hemorrhage seen on neuroimaging was also a risk factor for subsequent hemorrhage [27].

●Increasing age at diagnosis was the only other risk factor identified in an individual patient level meta-analysis (n = 2525), with the risk increasing by approximately 30 percent per decade (hazard ratio [HR] 1.34, 95% CI 1.17-1.53) [25].

●Anatomic and vascular features of the AVM also appear to be risk factors for subsequent hemorrhage, as shown by the systematic review [24]. These include:

•Exclusive deep venous drainage (HR 2.4, 95% CI 1.1-3.8)

•Deep brain location (HR 2.4, 95% CI 1.4-3.4)

•Associated aneurysms (HR 1.8, 95% CI 1.6-2.0)

•In contrast, size of the brain AVM was not associated with hemorrhage risk in the systematic review or the patient level meta-analysis [24,25].

●Pregnancy is not a risk factor for hemorrhage from a brain AVM according to most of the available studies, but the issue is controversial, and the data are not definitive [28-31]. In a retrospective analysis of 393 pregnant patients 18 to 40 years of age with brain AVMs, the risk of hemorrhage during the pregnancy and puerperium was not increased compared with the control period (odds ratio 0.71, 95% CI 0.61-0.82) [29].

●Combinations of risk factors may identify patients at particularly low or high risk, as found in a study from the Columbia databank [32]. Using three risk factors (hemorrhage at initial AVM presentation, deep venous drainage, and deep brain location), the approximate average annual hemorrhage rates were as follows:

•No risk factor, 1 percent annually

•One risk factor, 5 percent annually

•Two risk factors, 10 to 15 percent annually

•Three risk factors, >30 percent annually

Seizure and epilepsy risk — Seizures and epilepsy may develop subsequent to presentation. In one population-based series, the five-year risk of a first-ever seizure risk for AVMs that were discovered incidentally was 8 percent [33]. In contrast, for patients who presented with ICH or focal neurologic deficits, the five-year risk of a first seizure was 23 percent. Additional factors associated with increased seizure risk were younger age, temporal location, cortical involvement, and nidus diameter >3 cm [22]. In patients lacking a history of intracranial hemorrhage or focal neurologic deficit, the five-year risk of developing epilepsy after a first seizure was 58 percent.

DIAGNOSIS — The diagnosis of brain AVMs should be suspected in patients with an unexplained etiology of intracranial hemorrhage or new onset seizures, as well as those with an uncertain cause of acute neurologic deficits or altered mental status. The diagnosis is typically made noninvasively on imaging with brain computed tomography (CT) or magnetic resonance imaging (MRI) and/or angiography with computed tomography angiography (CTA) or magnetic resonance angiography (MRA). In addition, a substantial proportion of brain AVMs are diagnosed incidentally when neuroimaging is obtained for another indication. (See 'Clinical presentation' above.)

Conventional contrast angiography (ie, digital subtraction angiography [DSA]) may be necessary to make or confirm the diagnosis when initial neuroimaging is nondiagnostic. In addition, DSA is typically required to plan interventional treatment, sometimes in combination with MRI. (See 'Angiography' below.)

There is no role for routine screening of the general population for brain AVMs or for screening family members of patients with brain AVMs [6,34]. Unlike cerebral aneurysms, brain AVMs are not typically hereditary, with the exception of individuals with hereditary hemorrhagic telangiectasia (HHT).

●Screening for associated conditions – The diagnosis of HHT should be suspected in patients with spontaneous and recurrent epistaxis, multiple mucocutaneous telangiectasia at characteristic sites, gastrointestinal telangiectasia, pulmonary or hepatic arteriovenous malformation, or a first-degree relative with HHT. In a retrospective review of 89 children diagnosed with a brain AVM, the history of >2 episodes per year of epistaxis and the presence of pulmonary AVM or telangiectasias were predictive of an HHT diagnosis on genetic testing [35]. The role of screening for patients with HHT or suspected HHT is discussed separately. (See "Clinical manifestations and diagnosis of hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)" and "Hereditary hemorrhagic telangiectasia (HHT): Evaluation and therapy for specific vascular lesions".)

NEUROIMAGING

Computed tomography — In the absence of acute hemorrhage, noncontrast computed tomography (CT) has a lower sensitivity for detecting brain AVMs and vascular anomalies than magnetic resonance imaging (MRI) [20].

In patients who present with hemorrhage, CT characteristically demonstrates intraparenchymal hemorrhage without significant edema. However, compression of the nidus by the hematoma often precludes CT diagnosis of brain AVM in the setting of acute intracerebral hemorrhage; more sensitive techniques such as MRI or angiography are required in such cases (image 1). The ability of CT to identify brain AVMs in the acute setting may be improved by computed tomography angiography (CTA), which has a high sensitivity and specificity (95 and 99 percent, respectively) for the diagnosis of an underlying intracranial vascular malformation [36].

Magnetic resonance imaging — MRI is sensitive for delineating the location of the brain AVM nidus and often an associated draining vein. It also has unique sensitivity in demonstrating remote bleeding related to these lesions [2,20]. Dark flow voids are appreciated on T1- and T2-weighted studies (image 2). Similar to CTA, magnetic resonance angiography (MRA) a high sensitivity and specificity (98 and 99 percent, respectively) for the diagnosis of an underlying intracranial vascular malformation [36].

Angiography — Catheter contrast angiography (ie, digital subtraction angiography [DSA]) may be needed to confirm the diagnosis of a brain AVM and is essential for treatment planning and post-treatment follow-up of brain AVMs (image 3 and image 4) [20]. It has the highest spatial and temporal resolution of all the neuroimaging methods used to evaluate and diagnose brain AVMs. Anatomic and physiologic information such as the nidus configuration, its relationship to surrounding vessels, and localization of the draining or efferent portion of the brain AVM are readily obtained with this technique. DSA is required to evaluate for the presence of an early draining vein without a visible nidus, which is a risk factor for subsequent hemorrhage and cannot be detected using CTA or MRA [20]. Contrast transit times provide additional useful information regarding the flow state of the lesion; this can help in identifying venous outflow obstruction, which can influence endovascular treatment planning. The presence of associated aneurysm suggests a lesion at higher risk for subsequent hemorrhage.

Angiography is associated with a low risk of immediate neurologic complications, mainly ischemic stroke [20]. In addition, the high frame rates, magnified views, and multiple injections required for some diagnostic and interventional procedures may lead to high doses of radiation exposure with potential for long-term adverse effects [37].

ACUTE MANAGEMENT ISSUES

Acute intracranial hemorrhage — Brain AVM rupture typically causes intracerebral hemorrhage (ICH); isolated or concurrent intraventricular or subarachnoid hemorrhage may also occur, depending upon the AVM location. When an ICH is large and causing severe deficits, urgent clot evacuation surgically (with or without AVM treatment) is appropriate.

If the ICH is smaller with minimal deficits or if the AVM is in an area with high risk for worsening deficits, conservative observation can be instituted with consideration for later treatment of the AVM, depending on the lesion and patient specific factors.

The acute management of acute ICH, intraventricular hemorrhage, and subarachnoid hemorrhage are reviewed in detail elsewhere. (See "Spontaneous intracerebral hemorrhage: Acute treatment and prognosis" and "Intraventricular hemorrhage" and "Nonaneurysmal subarachnoid hemorrhage" and "Aneurysmal subarachnoid hemorrhage: Treatment and prognosis".)

Acute seizures — Individuals with AVMs may present with or develop seizures (see 'Clinical presentation' above and 'Hemorrhage risk' above).

Prophylactic antiseizure medication therapy to prevent a first seizure occurrence is generally not recommended. If a seizure occurs, appropriate antiseizure medication treatment should be administered to prevent recurrent seizures. The choice of the initial antiseizure medication depends upon individual circumstances and contraindications. The treatment of seizures and epilepsy is discussed in detail separately. (See "Initial treatment of epilepsy in adults" and "Overview of the management of epilepsy in adults".)

Although antiseizure medication therapy is generally successful in preventing recurrent seizures [33], drug-resistant epilepsy (also known as intractable or medically refractory epilepsy) develops in a significant minority of patients with brain AVMs and seizures, 18 percent in one series [38]. The treatment of seizures and epilepsy is discussed in detail separately. (See "Initial treatment of epilepsy in adults" and "Overview of the management of epilepsy in adults" and "Evaluation and management of drug-resistant epilepsy".)

MANAGEMENT OF THE AVM

Treatment goals and options — The major goals of interventional treatment are to reduce the risk of AVM-related hemorrhage, seizures, and other neurologic impairments. However, the efficacy of interventional methods for improving outcomes remains unproven and high-quality evidence is lacking.

The main options for brain AVM treatment are conservative medical management versus microsurgical excision or stereotactic radiosurgery; endovascular embolization is typically used as an adjunctive intervention to surgery, or less often for stereotactic radiosurgery. The exact technique chosen for treatment is contingent on a number of patient- and lesion-specific variables. (See 'Choice of treatment' below.)

Management decisions are most appropriately made by a multidisciplinary team of experienced clinicians who consider AVM features in the context of individual patient values and preferences [39].

Who should be treated? — The management of AVMs in any given patient is individualized based on factors such as patient age and medical comorbidities as well as the anatomic and vascular features of the AVM. Many experts believe that interventional treatment is reasonable for most patients with ruptured brain AVMs (see 'Ruptured AVMs' below) and for selected patients with unruptured AVMs (see 'Unruptured AVMs' below) [40]. In all cases, however, the natural history of brain AVMs must be weighed against the risks and benefits of any intervention [2,20,39,40]. Interventions for brain AVMs are associated with considerable risks, including permanent neurologic complications or death in approximately 5 to 7.5 percent. (See 'Risk of interventions' below.)

In determining whether a patient should undergo medical management or interventional AVM treatment, several factors are considered. Besides a history of previous rupture, other important factors influencing management include patient age, AVM size, AVM location, venous drainage pattern, and associated aneurysms. (See 'Other factors influencing treatment' below.)

Several of these parameters are incorporated into the Spetzler-Martin grading scale (table 1), which can be used to estimate surgical risk. Conservative medical management is usually preferred for Spetzler-Martin grade 4 or 5 lesions, which are considered to be at high risk of surgical morbidity. (See 'Spetzler-Martin grading scale' below.)

Ruptured AVMs — For most patients with ruptured AVM, we suggest intervention. Patients with a prior history of AVM rupture are at higher risk of subsequent hemorrhage than those without this history (see 'Hemorrhage risk' above). Thus, interventional AVM treatment is indicated for most patients who present with acute intracranial hemorrhage or have remote intracranial hemorrhage demonstrated on imaging, particularly those with other risk factors that favor treatment, such as deep location or exclusively deep venous drainage. Conservative medical management may be reasonable for older patients without additional risk factors or patients with an unacceptable risk of surgery. However, some experts recommend intervention for any ruptured AVM when technically feasible. (See 'Other factors influencing treatment' below and 'Spetzler-Martin grading scale' below.)

AVMs with angiographic features suggesting an increased risk of recurrent hemorrhage, such as an associated aneurysm, are treated acutely. Other AVMs are generally treated four to six weeks after the hemorrhage; absorption of the hematoma and resolution of any surrounding edema improve access to the AVM [39]. In patients undergoing radiosurgery, hematoma resolution may improve accuracy of the targeting of the radiation.

Unruptured AVMs — For patients with unruptured AVM who are not at high risk of rupture, we suggest conservative management. Interventional treatment of unruptured AVMs may be performed in select patients if the treatment-related risks are thought to be lower than the natural history risk of hemorrhage. Patients with symptoms (eg, seizures) refractory to medical treatment or with AVM features posing high risk for rupture should also be considered for treatment.

The treatment must be considered on a patient by patient basis, taking into account the various factors that influence the natural history of the lesion and the estimate of treatment-related risks. Typically, a multi-modality team evaluation is performed, and the risks and benefits of each treatment modality (surgical excision, radiosurgery, embolization) are considered alone or in combination. The angiographic and anatomic details of each AVM are carefully evaluated to estimate the potential for post-treatment morbidity. Intervention is often favored for young individuals at low risk of treatment-related adverse outcomes. Conservative medical therapy is a reasonable option for patients with unruptured brain AVMs who have no additional risk factors for hemorrhage such as deep location or exclusively deep venous drainage (see 'Hemorrhage risk' above), whereas microsurgical excision is reasonable for patients with additional risk factors for recurrent hemorrhage and Spetzler-Martin grade 1 or grade 2 lesions (table 1), suggesting low surgical risk. (See 'Spetzler-Martin grading scale' below.)

Data from a single randomized controlled trial (the ARUBA trial) suggest that interventions for unruptured brain AVMs are not beneficial and may be harmful compared with medical management, but the results are somewhat controversial. In the ARUBA trial, patients with unruptured AVMs were assigned to medical versus interventional (surgery, radiotherapy, and/or endovascular therapy) treatment [26]. The trial was halted early with outcome data on 223 patients followed for a mean of 33 months. By intention-to-treat analysis, the composite rate of symptomatic stroke (ischemic and hemorrhagic) and death was lower in the medical treatment group compared with the interventional group (10 versus 31 percent; hazard ratio 0.27, 95% CI 0.14-0.54). The rate of neurologic disability was also lower in medically treated patients (15 versus 46 percent). In an extended follow-up (median 50.4 months), the superiority of medical therapy appeared durable; the incidence of death or symptomatic stoke was lower for patients assigned to medical management than for those assigned to interventional therapy (3.4 versus 12.3 per 100 patient-years; HR 0.31, 95% CI 0.17-0.56) [41]. Two patients died in the medically treated group and four assigned to interventional therapy.

Criticism of the ARUBA trial includes small patient numbers, early stopping, and the nonstandardized and variable treatment approaches used in the interventional arm [42].

Other factors influencing treatment — The patient's age, presentation, anatomic and vascular features of the AVM, presence of associated seizures, and Spetzler-Martin grading scale score (table 1) are considered when making treatment decisions; each of these factors influences the choice between interventional treatments versus conservative medical management.

Age — Patient age is an important factor in the decision to treat brain AVMs; those with a longer life expectancy will accrue a higher lifetime risk of hemorrhage [43]. Thus, therapy is more likely to be recommended for children and young adults, while older individuals with shorter life expectancies may be managed more conservatively. However, long-term outcomes to guide treatment are lacking [44]. (See 'Hemorrhage risk' above.)

The cumulative hemorrhage risk can be estimated by the formula:

Lifetime risk of hemorrhage = 1 - (1 - P)^N

where P is the annual probability of hemorrhage and N is the expected years of life remaining [45]. As an example, a 60-year-old female with a newly diagnosed unruptured brain AVM and no other contributing comorbid conditions would expect to live approximately 20 years. If the annual risk of AVM hemorrhage is 3 percent, the formula gives a cumulative risk of hemorrhage over her expected life span of 46 percent. This compares with an overall risk of treatment morbidity and mortality of approximately 5 percent, with much or most of the treatment risk involving the periprocedural period.

AVM characteristics

●AVM location – AVMs located in eloquent brain or brainstem regions (ie, areas that control language, motor, sensory, or visual functions) present a challenge for risk assessment; significant clinical morbidity is likely to result if a surgical complication occurs or if the AVM ruptures. Such patients may be more likely to be considered for radiosurgery.

●Deep venous drainage – Similarly, deep venous drainage is a risk factor for both surgical complications and for AVM rupture.

●AVM size – Large brain AVMS are not clearly at higher risk of bleeding than smaller lesions, but are more difficult to treat; larger size is associated with higher risk by surgical and radiosurgical techniques. Obliteration rates fall with radiosurgery when larger lesions are treated with safe dosages of radiation. Lesions greater than 6 cm are likely to be managed conservatively. In some patients with large lesions, endovascular treatments may be useful to decrease the size of the aneurysm if the vascular anatomy is determined to be amenable to this approach.

Associated aneurysm — Treatment of aneurysms associated with AVMs varies depending on aneurysm location and diameter [17]. When believed to be the source of hemorrhage, aneurysms are generally treated with surgery or endovascular therapy, depending on their location and size, according to the expertise of the available experienced clinicians. Aneurysms associated with unruptured AVMs do not necessarily require treatment, depending on their size the other anatomic features.

Seizures — The effect of surgical resection on managing AVM-related epilepsy is uncertain. Available data, largely observational, suggest that interventional management of brain AVMs does not reduce the risk of subsequent seizures or epilepsy, but definitive conclusions cannot be made. A 2016 systematic review and meta-analysis identified only two controlled observational studies that compared interventional AVM treatments with antiseizure medication management alone for patients with brain AVMs and epilepsy [46]. In the pooled data from these two studies, rates of seizure freedom were similar between treatment groups (risk ratio 0.99, 95% CI 0.69-1.43); the overall proportion of patients who achieved seizure freedom with medical management in the two studies was 57 percent [47] and 46 percent [48] respectively. In the ARUBA trial, the rate of seizures per patient year was similar for the interventional group compared with medical management, but confidence in this finding is limited; among other issues, seizures were not the focus of the trial, and follow-up was only 33 months [26].

By contrast, some patients who undergo resection may develop new onset seizures and epilepsy [49-51]. In an retrospective study of 536 patients without prior seizures who underwent AVM resection, the cumulative rate of epilepsy at one year follow-up was higher than matched control patients with AVM who did not undergo resection (11.5 versus 3.4 percent) [52]. Postoperative seizure risk may vary by surgical technique, AVM characteristics, and presence of intracerebral hemorrhage.

Risk of interventions — Interventions for brain AVMs are associated with considerable risks. In a 2011 systematic review and meta-analysis of observational studies, findings included permanent neurologic complications or death in approximately 5 to 7.5 percent, and incomplete obliteration of the AVM in 13 to 96 percent [53]. By intervention, persistent neurologic deficits or death after microsurgery occurred in a median 7.4 percent of patients, after embolization in a median 6.6 percent, and after stereotactic radiosurgery in a median 5.1 percent.

In the ARUBA trial, which enrolled 223 patients with unruptured brain AVMs, subjects assigned to interventional therapy had a higher number of strokes (45 versus 12) and neurologic deficits unrelated to stroke (14 versus 1) compared with those assigned to medical management [26].

Spetzler-Martin grading scale — The Spetzler Martin grading scale classifies the risk of surgical AVM removal according to AVM size, location in eloquent or noneloquent brain areas, and whether deep venous drainage is present or absent (table 1) [54]. AVM size is determined by the largest dimension of the nidus in centimeters on imaging with computed tomography (CT), magnetic resonance imaging (MRI), or digital subtraction angiography (DSA). Eloquent brain areas include regions of cortex devoted to sensorimotor, visual, and language functions, and certain deeper structures: the internal capsules, basal ganglia, thalamus, hypothalamus, brain stem, cerebellar peduncles, and deep cerebellar nuclei [54,55]. Deep venous drainage is considered present if any or all of the outflow occurs via deep veins such as internal cerebral veins, basal veins, and precentral cerebellar vein [20].

A higher Spetzler-Martin grading scale score correlates with increased risk of surgical morbidity and neurologic deficits.

A modification of the Spetzler-Martin grading scale that supplements the neuroimaging information with clinical features (age, sex, baseline disability), may perform better in predicting surgical risk, but requires independent validation [56]. Other grading scales are also being evaluated for this purpose [57].

Choice of treatment — For patients selected for intervention, microsurgical excision is often preferred for patients with brain AVMs associated with a low risk of poor treatment outcomes, correlating with Spetzler-Martin (table 1) grade 1 or grade 2 lesions [40], with radiosurgery as an alternative for small lesions based upon location or other vascular or anatomic features. Stereotactic radiosurgery is also preferred for small grade 3 lesions. Large grade 3 lesions that involve eloquent cortex have a high surgical morbidity and available evidence suggests that treatment is no better than observational medical management. Conservative medical management is usually preferred for grade 4 or 5 lesions, although some may benefit from partial obliteration with endovascular treatment for high-risk features such as associated aneurysms located within the nidus or on feeding arteries.

Microsurgical resection appears to have the highest success rate for seizure control among treatment options for brain AVMs. When analyzed by type of intervention in systematic reviews and meta-analyses, the median rate of seizure freedom was highest for resective surgery (73 to 78 percent), followed stereotactic radiosurgery (52 to 63 percent) and embolization (49 to 54 percent) [46,58]. However, as previously noted, observational data suggest that interventional management of brain AVMs may not reduce the risk of subsequent seizures or epilepsy. (See 'Other factors influencing treatment' above.)

Particular advantages and disadvantages of the interventional modalities for brain AVMs are discussed in the sections that follow.

Microsurgical excision — Open microsurgical excision has the longest history of use for the definitive treatment of selected patients with AVMs and offers the best chance immediate cure in patients considered to be at high risk of hemorrhage [39]. The surgery is complicated and often requires detailed planning with review of imaging studies.

The main advantages of microsurgical excision compared with other interventions (radiosurgery and endovascular embolization) include a high success rate of complete nidus obliteration, the immediate elimination of hemorrhage risk, and long-term durability [20]. Disadvantages include the invasive nature of the treatment with a risk of neurologic impairments related to dissection of normal adjacent brain parenchyma and neurovascular structures needed to reach the AVM, and a longer recovery period. Postoperative complications include arterial or venous infarction or hemorrhage from the resection cavity. Typically, an arteriogram is performed after the surgery to document complete resection of the lesion. There have been isolated cases of hemorrhage into a resection bed despite obliteration on angiography. The mechanism of such bleeding is uncertain, but may be venous in nature. In addition to cortical injury, deep fiber tract injury can cause transient or permanent morbidity. Pre-operative assessment utilizing functional MRI studies with tractography may help minimize such complications.

An important factor in recommending therapy is an assessment of surgical risk. Multiple or large lesions, those in eloquent brain areas, and those with deep venous drainage are more difficult to safely resect. Many surgeons use the Spetzler-Martin grading scale (table 1) to assess the surgical risk [20]. Several studies have correlated surgical risk with higher Spetzler-Martin grading scale score [59,60]. Female sex may also increase surgical risk [61].

Stereotactic radiosurgery — Stereotactic radiosurgery is most successful when used to treat small brain AVMs eg, <3 cm in diameter [2].

●Latency period – Stereotactically focused high energy beams of photons or protons to a defined volume containing the brain AVM nidus induces progressive thrombosis of properly selected lesions via fibrointimal hyperplasia and subsequent luminal obliteration. The time course of these changes is usually one to three years, sometimes longer, and the time between treatment and obliteration is referred to as the latency period. During this latency period, the patient remains at risk for hemorrhage [2]. (See "Stereotactic cranial radiosurgery".)

The magnitude of brain AVM hemorrhage risk during the latency period between treatment and obliteration is uncertain; however, the best evidence suggests that it gradually declines during this interval [62]. This is an important consideration, particularly for lesions at higher bleeding risk. Existing studies are generally limited by retrospective and observational design. Earlier studies reported conflicting results, including increased [63-65], decreased [66,67], and unchanged [68,69] bleeding rates during the latency period. One of the largest studies including 500 patients found that the risk of hemorrhage declined by 54 percent during the latency period and by 88 percent after obliteration. The risk reduction was greater among patients who presented with hemorrhage than those who presented without hemorrhage. Another study including 657 patients had similar findings [70]. An untreated cerebral aneurysm was found to be a risk factor for hemorrhage in the latency period in another consecutive case series; the hemorrhage rate was 6.4 percent per year for untreated aneurysms versus 0.8 percent per year for clipped or embolized aneurysms [71].

Once the lesion is completely obliterated, the hemorrhage risk from the brain AVM is very low [72-74], but not totally eliminated [73,75-77]. Incomplete lesion obliteration, hypertension, and prior hemorrhage are risk factors for late bleeding [77].

●Success rate – Successful brain AVM obliteration with radiosurgery depends upon lesion size and dose of radiation [76]. Complete cure is considerably higher with smaller lesions; an overall 80 percent obliteration rate by three years occurs with lesions that are 3 cm or smaller [65,78-81]. Larger lesions have reported obliteration rates of 30 to 70 percent at three years [78,80,82]. Brain AVMs with a diffuse nidus or associated neovascularity were less likely to achieve a radiographic cure in one clinical series of 248 patients [83]. Despite the lower initial success rates for angiographic obliteration seen with larger brain AVMs (>3 cm), some amount of lesion volume reduction (mean 66 percent) typically occurs [84,85].

The success rate of radiographic brain AVM obliteration also varies with the amount of radiation delivered to the margin of the lesion. Doses of 16, 18, and 20 Gray (Gy) are associated with obliteration rates of about 70, 80, and 90 percent, respectively [85-87]. In one series, brainstem lesions were associated with lower obliteration rates, perhaps because of more conservative radiation dosing [88]. Radiation dose is generally selected according to a protocol which takes in account the size and location of the lesion.

Retreatment with radiosurgery is effective for complete obliteration in about 60 to 80 percent of patients with residual brain AVMs, depending on the size and other factors [84,89].

●Complications – Complications after radiosurgery include radiation necrosis, which can produce new neurologic deficits and seizures. In a multinational study that included 1255 patients undergoing radiosurgery for cerebral AVMs, therapy-related complications developed in 102 (8 percent) and included radiographic parenchymal lesions, cranial nerve deficits, seizures, headaches, and cyst formation [90]. Symptoms were disabling in 21, fatal in two, and resolved completely in 42 (41 percent). The risk of radiation necrosis with permanent neurologic deficit is 1 to 3 percent in most reports [78,80,85,91]. In another case series, 10 of 75 patients who had not had seizures and were not on antiseizure medications before radiosurgery had provoked seizures after radiosurgery [92].

The risk of hemorrhage following angiographically confirmed AVM obliteration appears low. In a retrospective cohort of 1607 patients treated with radiosurgery, hemorrhage occurred in 16 patients (1 percent) [93]. Only two of these 16 patients had developed a recurrent AVM, suggesting that hemorrhagic recurrence may also be due to other factors such as dysregulated neovascular proliferation following treatment.

The incidence of complications are related to the brain AVM location and the volume treated [81,94]. Thalamic, basal ganglionic, and brainstem locations are particularly prone to development of deficits after radiosurgery [88,94,95]. The risk of complications is also related to the radiation dose directed to the surrounding tissue.

The risk of complications is also increased in large brain AVMs that require larger treatment volumes. In a series of 73 patients, in whom one-half of the brain AVMs were >3 cm in diameter, the incidence of post-treatment imaging abnormalities and clinical complications rose with increasing treatment volume [96]. In patients whose treatment volumes were >14 mL and who received a dose ≥16 Gy, the incidence of post-treatment MRI abnormalities was 72 percent, and 22 percent required resection for radiation necrosis. The rate of post-treatment hemorrhage was also higher for treatment volumes ≥14 mL (7.5 versus 2.7 percent per person-year).

Repeated radiosurgery is also associated with increased complications; but the rate is not clearly prohibitive [89,97]. In one series of 15 patients who underwent two radiosurgeries with a mean dose per session of 18 Gy and 21 Gy, three (20 percent) had permanent radiation-induced complications [97]. No rebleeding occurred over 137 patient-years of follow-up.

In contrast to standard fractionation cranial irradiation, radiosurgery does not appear to impact cognitive function. One study of 10 patients found no effect of radiosurgical treatment of brain AVMs upon neuropsychological performance 11 months after treatment [98].

Endovascular embolization — Despite initial optimism that embolic agents such as microparticles and cyanoacrylates could cure brain AVMs, less than 25 percent of lesions are cured by this approach alone [99]. Typically, AVMs successfully treated with endovascular embolization are small and have a single draining vein [2].

Some experts suggest that embolization can be an effective adjunct to surgery and radiosurgery [100]. Embolization prior to surgery is employed to reduce blood loss and to occlude vessels that may be difficult to control during surgery [101]. Embolization prior to radiosurgery is controversial; it has been used to reduce the nidus size of large brain AVMs to less than 10 cm³, as these large AVMs have a lower cure rate with radiosurgery alone. Embolization may also reduce overall flow to the AVM. However, there is some evidence suggesting that embolization may reduce the obliteration rate and the rate of favorable outcomes after radiosurgery [102-106]; this drawback may be caused be embolic material in the AVM that can make it difficult to accurately target the nidus and can shield the AVM from the effects of radiation [20,40].

Endovascular therapy may also be used as primary treatment for intranidal aneurysms. (See "Treatment of cerebral aneurysms", section on 'Endovascular therapy'.)

A meticulous analysis of angiographic information (size, eloquent location, deep versus superficial venous drainage, vascular anatomy/number of feeders) determines the suitability for embolization [100,107]. Generally, only afferent pedicles to the nidus are embolized in an attempt to avoid occlusion of branches irrigating normal brain.

The risk of new neurologic deficits following endovascular treatment ranges from 8 to 20 percent [99,107,108]. Disabling treatment complications appear to be uncommon [107,108]. The most feared complications are ischemic stroke from embolic material occluding vessels to healthy cerebral adjacent tissue, and periprocedural hemorrhage. Hemorrhage with AVM embolization is usually the result of venous occlusion from the embolic material.

Follow-up — Long-term follow-up is recommended after AVM treatment with brain MRI, magnetic resonance angiography (MRA), and in some cases with DSA [109].

●For adult patients treated with embolization and microsurgical excision, DSA is obtained immediately after the procedures to document complete obliteration of the AVM.

●For adult patients treated with radiosurgery, interval MRI and MRA studies can be obtained at six months and one year. The regression of the AVM nidal volume can be accurately measured and followed with MRI, and the patient can be monitored for white matter change consistent with small vessel occlusion, post-therapy edema, or radiation necrosis [110]. Once an AVM is no longer visualized on an MRI and MRA, DSA can be performed to confirm complete obliteration.

●For adult patients treated with microsurgical excision, radiosurgery, and/or embolization, no further studies are needed if complete obliteration is documented, unless new symptoms occur.

●For children treated with microsurgical excision, radiosurgery, and/or embolization, DSA is obtained after both treatment and in a delayed fashion, typically at six months and at five years [109].

●For patients who are managed conservatively, we defer follow-up imaging unless new symptoms develop.

SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Stroke in adults" and "Society guideline links: Hereditary hemorrhagic telangiectasia (Osler-Weber-Rendu syndrome)".)

INFORMATION FOR PATIENTS — UpToDate offers two types of patient education materials, "The Basics" and "Beyond the Basics." The Basics patient education pieces are written in plain language, at the 5^th to 6^th grade reading level, and they answer the four or five key questions a patient might have about a given condition. These articles are best for patients who want a general overview and who prefer short, easy-to-read materials. Beyond the Basics patient education pieces are longer, more sophisticated, and more detailed. These articles are written at the 10^th to 12^th grade reading level and are best for patients who want in-depth information and are comfortable with some medical jargon.

Here are the patient education articles that are relevant to this topic. We encourage you to print or e-mail these topics to your patients. (You can also locate patient education articles on a variety of subjects by searching on "patient info" and the keyword(s) of interest.)

●Basics topics (see "Patient education: Arteriovenous malformations in the brain (The Basics)")

SUMMARY AND RECOMMENDATIONS

●Definition and epidemiology – Brain arteriovenous malformations (AVMs) are cerebrovascular malformations characterized by direct arterial to venous connections without an intervening capillary network. They occur in about 0.1 percent of the population but may be the cause of an estimated 1 to 2 percent of all strokes, 3 percent of strokes in young adults, and 9 percent of subarachnoid hemorrhages. (See 'Pathogenesis and pathology' above and 'Epidemiology' above.)

●Clinical presentation – Brain AVMs usually present between the ages of 10 and 40 years with intracranial hemorrhage, seizure, headache, or focal neurologic deficit; a substantial number are asymptomatic and are found incidentally. Hemorrhage is the most common presentation, particularly in children. (See 'Clinical presentation' above.)

●Risk of hemorrhage – Overall, annual hemorrhage rates from brain AVMs are between 2 and 3 percent. After an initial hemorrhage, annual hemorrhage rates are approximately 5 percent. Combinations of risk factors (eg, hemorrhage at initial AVM presentation, deep venous drainage, and deep brain location) may identify patients at particularly low or high risk. Patients with none of these risk factors have an estimated annual hemorrhage rate of approximately 1 percent, while those with all three risk factors may have an annual hemorrhage rate of >30 percent. (See 'Hemorrhage risk' above.)

●Imaging diagnosis – Brain AVMs can often be detected on computed tomography, magnetic resonance imaging, and/or noninvasive angiography with computed tomography angiography or magnetic resonance angiography. Digital subtraction angiography (DSA) may be necessary to make or confirm the diagnosis when initial neuroimaging is nondiagnostic and is essential for treatment planning and follow-up after treatment of brain AVMs. (See 'Neuroimaging' above.)

●Management – The management of AVMs in any given patient is individualized based on risk factors such as patient age, medical comorbidities, and the anatomic and vascular features of the AVM (table 1) as well as the risks of morbidity with intervention (see 'Natural history' above and 'Who should be treated?' above):

•For most patients with a ruptured AVM, we suggest intervention (Grade 2C). In some cases, intervention may not be technically feasible or safe or desired by the patient. (See 'Ruptured AVMs' above.)

•For patients with an unruptured AVM who are not at high risk of rupture, we suggest conservative management (Grade 2C). However, interventional treatment of unruptured AVMs may be performed in select patients at low treatment-related risks, with symptoms (eg, seizures) refractory to medical treatment, or with AVM features posing high risk for rupture. (See 'Unruptured AVMs' above.)

•For patients selected for intervention, microsurgical excision is often preferred for patients with brain AVMs associated with a low risk of poor treatment outcomes, with radiosurgery as an alternative. Conservative medical management is usually preferred for patients with brain AVMs associated with a high risk of poor treatment outcomes, although some may benefit from partial obliteration with endovascular treatment. (See 'Choice of treatment' above.)