Natural History of Brain Arteriovenous Malformations: A Systematic Review

Isaac Josh Abecassis, M.D.; David S. Xu, M.D.; H. Hunt Batjer, M.D.; Bernard R. Bendok, M.D., M.S.C.I.


Neurosurg Focus. 2014;37(3):e7 

In This Article

Abstract and Introduction


Object The authors aimed to systematically review the literature to clarify the natural history of brain arteriovenous malformations (BAVMs).

Methods The authors searched PubMed for one or more of the following terms: natural history, brain arteriovenous malformations, cerebral arteriovenous malformations, and risk of rupture. They included studies that reported annual rates of hemorrhage and that included either 100 patients or 5 years of treatment-free follow-up.

Results The incidence of BAVMs is 1.12–1.42 cases per 100,000 person-years; 38%–68% of new cases are first-ever hemorrhage. The overall annual rates of hemorrhage for patients with untreated BAVMs range from 2.10% to 4.12%. Consistently implicated in subsequent hemorrhage are initial hemorrhagic presentation, exclusively deep venous drainage, and deep and infrantentorial brain location. The risk for rupture seems to be increased by large nidus size and concurrent arterial aneurysms, although these factors have not been studied as thoroughly. Venous stenosis has not been implicated in increased risk for rupture.

Conclusions For patients with BAVMs, although the overall risk for hemorrhage seems to be 2.10%–4.12% per year, calculating an accurate risk profile for decision making involves clinical attention and accounting for specific features of the malformation.


First described by Steinheil in 1895, brain arteriovenous malformations (BAVMs) are a complex of abnormal arteries and veins that directly fistualize without an intervening capillary bed. From a purely categorical level, BAVMs differ from other fistulous vascular malformations, such as vein of Galen malformations, dural arteriovenous fistulas, or secondary malformations that arise from trauma, or neovascularization that occurs after chronic cerebral venous occlusion. Because BAVMs can differ in size, location, morphology, and angioarchitecture, clinical management varies substantially from patient to patient. The decision to proceed with treatment of a BAVM ultimately hinges on weighing the subsequent risk for intracranial hemorrhage with the immediate risks from intervention. Unfortunately, the natural history of BAVMs is still largely unknown. Current data are mostly limited to isolated single-center case series. Recently, ARUBA (A Randomized Trial of Unruptured Brain Arteriovenous Malformations)—a trial that randomly assigned AVM patients to either a conservative medical management group or an intervention group (any sort of intervention, including surgery, embolization, or Gamma Knife)—was stopped by the National Institute of Neurological Disorders and Stroke because after a mean follow-up time of 33 months, the event rate (death or symptomatic stroke) was more than 3 times higher among patients in the intervention group than among those in the medical management group.[30] Thus, it is imperative that clinicians have a detailed understanding of the existing literature so that they can effectively counsel patients about the precise and individualized natural history of BAVMs. Gross and Du recently published a meta-analysis of the risk factors that predict hemorrhage in AVMs. They found that rates of future rupture were significantly increased among patients with prior hemorrhage, deep brain location, exclusive deep venous drainage, and associated aneurysms.[12] Our goal in this study was to more broadly and inclusively survey the literature on this topic and to summarize reasonably powered studies of the natural history of BAVMs and the independent rates of rupture. Additionally, we review BAVM epidemiology and presenting symptoms.