Radiation Oncology/CNS/Arteriovenous malformation


  • Originally described by Luschka and Virchow in the mid 1800s, AVMs are abnormal communications between arteries and veins without the normal capillary flow.
  • The incidence in the US population is 0.14% (140 cases per 100,000 persons or 1 case per 700 persons). This is approximately one fifth to one seventh the incidence of intracranial aneurysms.
  • Abnormal communication between artery and vein, with disproportionate and unbalanced hydrodynamic stress across them. The direct connection between the arterial and venous systems exposes the venous system to abnormally high pressures and results in the formation of enlarged feeding vessels and enlarged draining venous structures.
  • Point of abnormal communication is known as the nidus
  • Annual risk of hemorrhage 2-6%, with associated fatality rate 10-20%
  • Primary treatment approach is typically surgical resection
    • Risk evaluated by Spetzler-Martin score (PMID 3760956) (size, location, pattern of venous drainage), which estimates post-op morbidity and mortality
  • SRS typically used if surgery is considered too risky: surgical inaccessibility, expected high morbidity due to eloquent location or large size, medically inoperable
  • Goal of SRS treatment is to eliminate risk of catastrophic intracranial hemorrhage
  • Probability of obliteration ("cure") is ~80% for smaller lesions, and depends on dose
  • There is a definite association between AVM and aneurysm with about 8% of AVM patients developing a cerebral aneurysm and about 1% of patients with an aneurysm found to have an AVM. Aneurysms are most commonly found on arteries feeding the AVM.
  • AVMs are congenital lesions that result from the failure of the embryonic vascular plexus to fully differentiate and develop a mature capillary bed. While the structure may change or grow postnatally, they are thought to only be seen in the presence of a prenatal lesion. Tissues adjacent to the AVM may be mildly hypoxic due to “stealing” blood from adjacent healthy tissue. This may lead to expression of pro-angiogenic factors such as VEGF and bFGF.
  • Hemorrhage is the most common reason for presentation (~50%). It is often associated with the acute onset of a severe headache that may be described as “the worst headache of my life”. Depending on the location, it may also be associated with a new neurologic deficit. Seizure can be seen in about 45% of patients, headache in 33%, and neurologic deficit in 20%.
  • Pediatric patients can present with heart failure, macrocephaly, and/or prominent scalp veins.

Management overview

  • Surgery - Cure is immediate and permanent after complete resection. Is generally recommended for grade 1, 2, and (sometimes) 3 lesions. One of the primary risks is that of hemorrhage into healthy parts of the brain.
  • Endovascular neurosurgery -- Involves the obliteration of vessels with glues or particles delivered via arterial catheter. Its main use is “neoadjuvantly” prior to craniotomy to decrease intraoperative bleeding. It can also be used prior to SRS to decrease the size of an AVM. Embolization can be curative for lesions <1cm that are fed by a SINGLE artery. The main risk involves occluding a vessel that also supplies normal brain. The pathologic vessels usually recanalize over time, so this is rarely used by itself.
  • Stereotactic radiosurgery -- Can be used if surgery is considered too risky: inaccessible location, expected high morbidity due to location, large size, medically inoperable. The goal of SRS is to eliminate the risk of catastrophic intracranial hemorrhage. Some groups have used staged or fractionated SRS to treat larger lesions. It can take 2 or more years for a full destructive effect and the risk for hemorrhage is not reduced during this time. Cure rates for lesions <3cm are 80-90%.
  • Controversies-- Given the increased use of cranial imaging estimates are that 50-60% of newly diagnosed AVMs are detected incidentally. Although the traditional stated risk of hemorrhage is 2-3% per year, this is typically calculated from the time of diagnosis. More recent estimates place the risk of hemorrhage at closer to 1%. Predictors of AVM hemorrhage include increasing age, deep brain location, associated aneurysms, and deep venous drainage. This has led to: ARUBA (A Randomized Trial of Unruptured Brain Arteriovenous Malformations), a multicenter randomized trial enrolling 800 patients, that compared treatment (sugery, radiation, or both, physician decision) of unruptured lesions with conservative management. Randomization was ended early because of increased risk of stroke and death in the intervention group.


  • Helsinki, 1990 (1942-75) PMID 2384776 -- "The natural history of symptomatic arteriovenous malformations of the brain: a 24-year follow-up assessment." (Ondra SL, J Neurosurg. 1990 Sep;73(3):387-91.)
    • Retrospective series on 160 patients with untreated symptomatic AVMS with a mean follow-up of 23.7 years (seen between 1942 and 1975 at Helsinki University). The rate of major rebleeding was 4% per year and the mortality rate was 1% per year. Nearly 25% of patients were dead from AVM hemorrhage. There was no significant difference based on whether a patient presented with hemorrhage, seizure, or other.


  • Florida, 1995 (1988-93) PMID 7815144 -- "Linear accelerator radiosurgery for arteriovenous malformations: the relationship of size to outcome." (Friedman WA, J Neurosurg. 1995 Feb;82(2):180-9.)
    • Report of 158 patients at UF treated between 1988 and 1993. Mean dose of 1560 cGy to the periphery with mean volume of 9 mls. Mean f/u of 33 months during which patients were followed with MRI until findings suggested complete thrombosis at which time an arteriogram was performed to verify occlusion. If a persistent nidus was found at 36 months, patients were re-treated. The goal was to relate size to outcome. Only 56 patients had "definitive endpoints" (death related to hemorrhage, retreatment, angiographic cure) so this data may be quite biased. Due to the significant selection of whom they reported on, the more interesting data may be their complications. Seizure within 48 hrs of SRS was seen in 4.6% of patients (all of these initially presented with seizure). Hemorrhage was seen in 4% of patients between 2 and 11 months after treatment. Transient delayed complications (HA, dysphasia) was seen in 2% of patients with resolution after steroids.
  • Pittsburgh, 1996 (1987-92) -- "A dose-response analysis of arteriovenous malformation obliteration after radiosurgery." (Flickinger JC, Int J Radiat Oncol Biol Phys. 1996 Nov 1;36(4):873-9.)
    • Retrospective, 216 patients with gammaknife radiosurgery for AVM at U Pitt. Of these, 197 had angiographic followup with a median treated volume of 4.1 ml. Complete obliteration was seen in 72% of patients. On multivariate analysis, a correlation with minimum dose was seen, but not with volume or maximum dose. Very few patients were treated with <15 Gy or >25 Gy. Very nice figures in the paper.
  • Pittsburgh, 2000 -- "Development of a model to predict permanent symptomatic postradiosurgery injury for arteriovenous malformation patients. Arteriovenous Malformation Radiosurgery Study Group." (Flickinger JC, Int J Radiat Oncol Biol Phys. 2000 Mar 15;46(5):1143-8.)
    • Review from 85 AVM patients who developed symptomatic complications and 337 control patients without complications to develop a method to better predict permanent complications from AVM radiosurgery. Essentially, the risk of developing complications is related to the location and volume receiving 12 Gy. History of prior hemorrhage and marginal dose did not significantly predict injury.
  • Tokyo, 2005 PMID 15647577 -- "The risk of hemorrhage after radiosurgery for cerebral arteriovenous malformations." (Maruyama K, N Engl J Med. 2005 Jan 13;352(2):146-53.)
    • Retrospective analysis of 500 patients treated with gammaknife RS to assess the effects of RS on rates of hemorrhage. RS dose was 20 Gy with use of the 50% isodose line. RESULTS: 42/500 patients had a hemorrhage before radiosurgery. 23/458 patients had a hemorrhage during the latency period (the interval between RS and angiographic obliteration). 6/250 patients had hemorrhages after obliteration. The annual rate of repeated hemorrhage prior to radiosurgery was about 6% among those patients who presented with hemorrhage.


  • Pittsburgh, 1998 (1987-94) PMID 9457809 -- "Analysis of neurological sequelae from radiosurgery of arteriovenous malformations: how location affects outcome." (Flickinger JC, Int J Radiat Oncol Biol Phys. 1998 Jan 15;40(2):273-8.)
    • 332 pts treated with GKRS; minimum f/u 2 yrs (median 45 mo).
    • Symptomatic neurologic sequelae seen in 9%; resolved in 58% by 27 months (more likely to resolve if Dmin < 20 Gy). 7-yr rate of persistent symptoms 3.8%. Increased risk associated with location and 12 Gy volume.
    • Conclusion: 12 Gy volume associated with long-lasting neurologic sequelae

Proton Therapy



  • Harvard; 2003 (1965-1993) PMID 12924697 -- "Dose-volume prediction of radiation-related complications after proton beam radiosurgery for cerebral arteriovenous malformations." (Barker FG 2nd, J Neurosurg. 2003 Aug;99(2):254-63.)
    • Retrospective. 1,250 patients. DVH information reviewed. Median size 33.7 cm3, 23% <10 cm3, largest 927 cm3. Median dose 10.5 Gy. Median F/U 6.5 years
    • Toxicity: permanent late toxicity 4% (hemiparesis 2%, visual field defects 1%, other <1%: cognitive dysfunction, ataxia, speech deficits, hemisensory deficits, hearing loss). Median time-to-complication 1.1 years (0.2 - 6.8 years)
    • Predictors: Median dose complications 17 Gy vs no complications 10 Gy. If dose <12 Gy, complication rate 0.5% but a complication also occurred at 5 Gy. Complication rate related to dose, volume, thalamic/brainstem location, and age (median age 31). Significantly more complications than predicted by Kjellberg's model (e.g expected 2-2.5% complication bin had 34% actual complications)
    • Conclusion: Estimate of risks after proton SRS for AVM; dose-volume model developed


  • 2007 PMID 17596605 -- "Clinical practice. Arteriovenous malformations of the brain." (Friedlander RM, N Engl J Med. 2007 Jun 28;356(26):2704-12.)