Presenter: Jona Hattangadi, MD
Presenter's Affiliation: Harvard University
Arteriovenous malformation (AVM) is a congenital abnormality of arterial and venous communication, caused by failure of the vascular plexus to develop into a mature capillary bed.
The direct connection between the arterial and venous systems exposes the venous system to abnormally high pressures, resulting in the formation of enlarged feeding vessels and draining venous structures. The point of abnormal communication is called a nidus.
While AVM is a rare entity, with prevalence of < 0.5%, the risk of hemorrhage for patients with AVM is 2-4% per year. Fatality associated with hemorrhage is 10-20%.
The goal of treatment is to minimize the risk of fatal hemorrhage.
Most commonly, patients present with seizure, acute onset of severe headache, or new neurologic deficit.
The primary modality of treatment has been surgical resection, however other techniques include endovascular neurosurgery (obliteration of vessel mass with glues/particles used pre-operatively to decrease bleeding from other interventions), embolization (if AVM fed by single artery), or stereotactic radiosurgery (SRS).
SRS has typically been used for patients deemed inappropriate for surgical resection, i.e. AVM in an inaccessible location, high anticipated morbidity (due to large size or eloquent location), or medically unfit for surgery
The probability of cure ("total obliteration") after SRS depends on the size of the lesion and dose delivered, with cure being more likely in smaller lesions and with higher total dose.
While surgical results are immediate and permanent, it can take 2 or more years for a full destructive effect to occur after radiation. The risk for hemorrhage is not reduced during this time.
Obliteration rate for lesions < 3 cm is 72-96%, whereas for large lesions (>10 cm), the cure rate is less impressive and more variable (0-71%), and unfavorable dose-volume parameters are more common. AVMs in eloquent areas of the brain have an obliteration rate of 25-82%, but are prone to hemorrhage causing damage to critical structures.
Some groups have used staged or fractionated SRS to decrease normal tissue toxicity and risk of hemorrhage in large AVMs.
High-risk AVMs are defined as being either > 10 cm, eloquent (located in motor strip, basal gang, thalamus, or brainstem), or both.
Proton beam SRS (P-SRS) has been used at Harvard to treated AVMs since 1965. Institutional data has shown similar obliteration rates to photon-based SRS and Gamma Knife.
The principal advantage of P-SRS is the sharp dose fall-off in areas of critical normal tissue in the brain.
The purpose of this study was to evaluate the AVM obliteration rate and treatment-related toxicity rate for P-SRS using a two-fraction, multi-stage approach.
Materials and Methods
Retrospective study assessing 59 pts with high-risk AVM treated from 1991-2009 with two-fraction, multi-stage, P-SRS.
Patients were eligible if AVMs were large or in an eloquent area of the brain.
Prior to CT simulation, fiducial markers were implanted in the outer skull to assist in treatment planning. At the time of CT simulation, patients were immobilized in a head frame in the treatment position, and CT was fused with MRI and CT angiogram.
Two fractions were delivered, each with 2-4 convergent beams.
The median total dose was 16 RBE (range 12-28 RBE) prescribed to the 90% isodose line.
Follow-up included a clinical exam and reimaging on a yearly basis.
Primary outcome: AVM obliteration classified as total (100% obliteration), partial (< 100% obliteration), or no change.
Finally, factors associated with favorable outcomes were examined.
Kaplan Meier statistics were used for analysis.
The median age of the cohort was 47 yrs, but 19% were < 18 yrs of age.
92% of patients were symptomatic at presentation, 37% had experienced prior hemorrhage, and 40% had undergone prior surgery or embolization.
39% had tumor in an eloquent region and median tumor volume was 23 cc, making these patients very high risk. The median size of the nidus was 4 cm.
The AVM score is a compilation of age, AVM volume, and location, and the number predicts for outcome after SRS. The median AVM score for this cohort was 3.2, one of the highest reported in the literature.
Of the 59 patients, 15% had total obliteration, 34% had partial obliteration, and 51% had no change. The 5 yr obliteration rate was 8%.
Factors significantly associated with total obliteration included higher dose and lower AVM score.
In terms of side effects, 15 patients experienced a hemorrhage after radiation, 13 of which were due to rupture of the AVM and 2 due to radiation dose to aneurysm. Patients were more likely to experience a rupture if their AVM score was greater than 3.2. 5 yr rate of hemorrhage was 21%.
Of the total cohort, 9 patients died of fatal hemorrhage and 3 had unrelated deaths
With respect to acute toxicity, 14% of patients had grade 1 toxicity (headache), which resolved with over-the-counter medications, 7% had neurologic changes, which resolved 2 weeks after the last proton treatment, and 12% had mild seizure within 48 hours after treatment resolved. This led to a practice change of administering prophylactic benzodiazepines during the treatment period to prevent seizure.
In terms of late toxicity, 12% had long-term grade 1 headache, 3% had lasting neurologic changes, and 5% developed seizure disorders requiring medications.
The overall rate of toxicity (acute and late combined) was 49%.
Dosimetrically, comparison of proton vs. photon dose distribution showed significant dose sparing of optic chiasm and brainstem with proton therapy
Of note, 25 pts (42% of cohort) were retreated at 2-3 yrs due to lack of response, residual presence of nidus, or ongoing symptoms.
High-risk AVM can be safely treated with a two-fraction, multi-stage approach.
Though there is a low percentage of complete obliteration, there is also low toxicity.
The hemorrhage rate was higher than expected, however not surprising given the high-risk nature of the cohort.
Higher doses led to a higher chance of complete obliteration.
With longer follow-up, the rate of complete obliteration may increase.
Given the increased use of cranial imaging in recent years, estimates suggest that roughly 50-60% of newly diagnosed AVMs are detected incidentally. More recent estimates of hemorrhage risk are about 1%, lower than traditionally thought, largely due to earlier detection of AVMs.
This study highlights the important point that P-SRS is feasible with potentially better dosimetric parameters than photon-based therapy. However, the data do not imply that P-SRS is preferable to either photon SRS or Gamma Knife.
Given the retrospective nature of the study, small number of patients, and short follow-up, direct comparisons between surgery, photon SRS, or Gamma Knife are not possible.
Additionally, almost 20% of the patients in this cohort were children, which may have confounded the results.
40% of the patients had received prior treatment with surgery or embolization and nearly 40% presented after initial hemorrhage. In a sense, this study assessed the utility of P-SRS as a "salvage" therapy for high-risk patients, and it is unclear whether or not these results would translate to treatment-naïve patients with lower-risk disease.
Normal tissue sparing is of critical importance when treating patients with benign disease because they are likely to live long enough to experience side effects from normal tissue complication. This study examined a particularly high-risk group treated with P-SRS, for whom complications rates were quite high. In this group, the authors found a 49% rate of overall toxicity for a 15% chance of total obliteration. Further study would help elucidate the risk-benefit ratio for patients with lower-risk AVM – an essential question regarding this treatment for non-malignant disease.
A cost-benefit analysis would also be helpful, given the cost of proton therapy as compared to other potentially equally effective modalities.
An ongoing multicenter randomized trial with 800 patients compares surgery, radiation, or both for unruptured lesions. Comparison of P-SRS to other radiation modalities in such a randomized controlled fashion would be useful to determine its true v