A treatment paradigm for high-grade brain arteriovenous malformations: volume-staged radiosurgical downgrading followed by microsurgical resection

,

C ontemporary management of brain arteriovenous malformations (AVMs) requires the coordination of several treatment modalities including endovascular embolization, microsurgical resection, stereotactic radiosurgery (SRS), and observation.The Spetzler-Martin grading system helps individualize management by separating patients into 2 distinct groups: low-grade (Grades I, II, and most III) and high-grade AVMs (some Grade III, and Grades IV and V).Low-grade AVMs are generally small, superficial, and/or noneloquent with low morbidity rates associated with resection and high obliteration rates associated with SRS, making both modalities acceptable.In contrast, high-grade AVMs are large, deep, and usually involve eloquent brain areas.Not surprisingly, resection is associated with substantial morbidity, 18,26 while SRS is associated with low obliteration rates.Therefore, high-grade AVMs are some of the most challenging lesions because they require an understanding of when to simply observe, which runs against our clinical instinct to intervene to protect the patient from hemorrhage.The surgical indications for patients with high-grade AVMs include previous hemorrhage, an existing significant permanent deficit, progressive neurological deficit related to vascular steal, or an associated arterial or intranidal aneurysm. 9In young patients who do not have any of these features, observation is usually recommended, but this option is based on a lack of acceptable alternatives.A treatment paradigm is needed for these patients with substantial lifetime hemorrhage risks.
Single-session SRS (SS-SRS) is not effective for highgrade AVMs because nidal volumes greater than 14 or 15 cm 3 (approximately 3 cm in diameter) require reductions in the marginal dose below 16 Gy to avoid adverse radiation complications, 19 while 16-, 18-, and 20-Gy marginal doses are associated with 70%, 80%, and 90% obliteration rates for AVMs overall. 22Volume-staged SRS (VS-SRS) is a newer strategy that divides a large AVM into 2 or 3 smaller portions that are treated at separate stages enabling each portion to receive a higher dose. 12,22The higher dose may be associated with a greater likelihood of response, while the separation of stages by months and proper alignment of staged volumes may reduce complications.
Volume-staged SRS has improved rates of AVM obliteration compared with SS-SRS, but as a sole form of therapy, cure rates are low. 3,11,12,22,25,31VS-SRS of high-grade AVMs often results in a partial response, which may transform inoperable AVMs into low-grade lesions with a more acceptable surgical risk profile.Microsurgery has been used effectively and with reduced morbidity as a salvage technique after SS-SRS. 24,29Included in these reports are patients treated with additional sessions of SRS to residual AVM at the end of the latency period, which differs from upfront volume-staged therapy.However, the planned combination of VS-SRS plus microsurgery has not been evaluated as part of a treatment paradigm for inoperable high-grade AVMs.Although it may have been conceived as early as 1998 by the Pittsburgh group, 4 publications are limited to a single case report.Therefore, we reviewed our experience with microsurgical AVM resection after VS-SRS to advance the concept of downgrading inoperable AVMs and facilitating curative resection with radiosur-gery.This multimodality approach differs from salvage surgery after SS-SRS and from VS-SRS as a stand-alone approach.Although we and others have published case series on both VS-SRS alone and SS-SRS plus surgery, there are no case series published on VS-SRS plus surgery.Our experience with VS-SRS plus surgery described in this report establishes its effectiveness in select patients with AVMs that initially were overwhelming.

methods study design
This study was approved by the institutional review board and was performed in compliance with Health Insurance Portability and Accountability Act regulations.Patients with AVMs undergoing both VS-SRS and microsurgery and were identified from 2 prospectively maintained databases, one from the University of California, San Francisco Brain Arteriovenous Malformation Study Project and the other from the Gamma Knife stereotactic radiosurgery service.Data, including components of the Spetzler-Martin 26 and supplemented Spetzler-Martin 18 grading systems and modified Rankin Scale (mRS) outcomes, were collected from the prospective databases and additional data were gathered retrospectively.

patients
Volume-staged SRS was initiated at our institution in 1992 and patients included in this study were treated during a 20-year period (from 1992 to 2012).All VS-SRS was performed or supervised by senior faculty (M.W. McD.,  1992-2012) and AVM resections were performed by the senior author (M.T.L.) over a 16-year period from 1997 to 2013.

treatment strategy
Patients with AVM were reviewed at weekly multidisciplinary conferences attended by neurosurgeons, neurologists, neuroradiologists, interventional neuroradiologists, radiation oncologists, and Gamma Knife coordinators.Patients with high-grade AVMs deemed unfavorable for open microsurgery were initially treated with VS-SRS.AVM response to VS-SRS was evaluated at the end of an approximately 3-year latency period, and AVMs with a decreased Spetzler-Martin grade or significant radiation response were reconsidered for microsurgical resection.Postradiosurgical AVMs deemed more favorable for open microsurgery were offered this additional treatment, with or without preoperative embolization.

volume-staged radiosurgical technique
Our technique for VS-SRS for large AVMs was tabulated in 69 patients in a separate report (unpublished data, Seymour et al., International SRS Congress, June 19,  2013).Patients underwent stereotactic MRI, MR angiography, and cerebral angiography prior to the first VS-SRS session.During the first session the entire AVM volume was planned and then the portion of the AVM closest to the primary arterial input was targeted first.In subsequent sessions repeat MRI/MR angiography was performed, and the treatment plan from the first session was coregistered to the new imaging data set to avoid overlap of dose in brain adjacent to the AVM. 6In general, these patients had AVMs larger than 10 cm 3 in volume in eloquent and/ or deep locations with minimal or no neurological deficits, or the patients had refused surgery.Individual radiosurgical sessions were separated by an interval of 3-6 months.In those patients treated in the second half of the experience (after 2005, considered Era 2) the volume per stage was decreased to less than 8-10 cm 3 and the dose per stage increased to ≥ 17 Gy.MRI was performed at 12-month intervals following completion of VS-SRS, and angiography was performed at 36 months.AVMs that were downgraded or favorably changed by VS-SRS after 3 years were then treated by open microsurgery.Some AVMs that were not downgraded or were favorably changed by VS-SRS after 3 years were considered for salvage radiosurgery.Patients with AVMs that were not downgraded or favorably changed by VS-SRS, and were still deemed unfavorable for open microsurgery, were not selected for microsurgery.

outcomes
Patient outcomes were assessed using the mRS during posttreatment clinic visits, subsequent hospital admissions, or telephone interviews by research personnel not involved in the patients' care.A good outcome was defined as a final mRS score of 0-2, while a poor outcome was defined as mRS score greater than 2. Functional improvement was defined as a decrease in mRS score from the initial examination at presentation to the final followup examination.Angiography demonstrating complete AVM resection and no residual arteriovenous shunt was required for AVM obliteration.

statistical analysis
Comparisons between groups were made using the Fisher exact test for categorical variables, using the chisquare test for categorical variables with more than 2 categories, and with the Wilcoxon-Mann-Whitney test for comparison of continuous variables; p values < 0.05 were considered statistically significant.

results overall treatment results
The treatment paradigm for AVMs with Spetzler-Martin Grades III, IV, or V is shown in Fig. 1.Of 402 patients with high-grade AVMs managed since 2000, 54 were selected for observation alone, 199 underwent embolization and resection, 37 underwent embolization alone, and 112 underwent some form of SRS (either SS-SRS or VS-SRS).In the period between 1992 and 2012, 74 patients were treated with VS-SRS.Eight AVMs (10.8%) were cured, 50 (67.6%)were unchanged or remained unfavorable for microsurgery, and 16 (21.6%)were selected for microsurgery.Microsurgical treatments were performed during the years 2000 to 2012 in all patients except one (1998).

vs-srs treatment
The mean number of VS-SRS sessions was 2.7 (range 2-5 sessions) (Table 1).Before VS-SRS, the average Spetzler-Martin grade was 4.0 and the average supplemented Spetzler-Martin grade was 7.1, with an average AVM diameter of 5.9 cm.After VS-SRS, the average Spetzler-Martin grade decreased to 2.5, a downgrade of 1.5 points.Similarly, the average supplemented Spetzler-Martin grade decreased to 5.6, a downgrade of 1.5 points.The average AVM diameter decreased to 3.0 cm, with a mean reduction in nidus size of 2.9 cm.After VS-SRS, 14 of 16 AVMs had Spetzler-Martin grades less than or equal to Grade III, and 13 of 16 AVMs had supplemented Spetzler-Martin grades less than or equal to Grade 6, the usual cutoffs for surgical intervention.

microsurgical treatment
The mean interval between the initiation of VS-SRS and microsurgical resection was 5.7 years (range 0.5-18.7 years) (Tables 1 and 2).Four patients experienced a hemorrhage during the latency period, prompting surgery before the usual 3-year reevaluation.Six patients underwent preoperative embolization after VS-SRS (Table 2).
Of the 16 patients undergoing surgery, 1 had a residual AVM after all treatment.One patient with a 5-cm-diameter AVM located in Wernicke's area in the dominant hemisphere underwent an awake craniotomy with speech mapping.The AVM was not completely resected to preserve the language cortex.Postoperative angiography confirmed curative AVM resection in 15 patients (93.8%).A summary of angiographic images, initially, before surgery, and following surgery is shown in Fig. 2. The location of the AVMs treated in this series is graphically depicted in Fig. 3.
Four patients had poor outcomes (mRS Score 3, all experienced deterioration after treatment).One patient's condition deteriorated after VS-SRS (mRS Score 2 to 3, Case 2); one patient's condition deteriorated after a hemorrhage during the latency period following VS-SRS (mRS Score 1 to 5, Case 7); and one patient's condition deteriorated after resection of an AVM in the motor cortex (mRS Score 2 to 3, Case 1).An additional patient with an insular/basal ganglia AVM underwent uncomplicated AVM resection without neurological sequela, but worsened 2 days after surgery because of a delayed venous infarction (mRS Score 0 to 3, Case 13).
Two patients died (treatment mortality, 12.5%).One patient with a paramedian frontal AVM involving the motor cortex presented with a hemorrhage and minor neurological deficits but then experienced rehemorrhage 6 months after completing VS-SRS.Although she was aggressively treated with hemicraniectomy at an outside hospital, and AVM resection after transfer to our institution, she did not emerge from coma and died.The second patient had a similar but much larger AVM that presented after VS-SRS with steno-occlusive changes in parasagittal veins and marked hemispheric edema.Resection was complicated by her limited venous drainage, intraoperative AVM rupture, and postoperative intracranial hypertension.

avm grading and risk prediction with vs-srs plus surgery
In this series, VS-SRS downgrading plus microsurgery was an effective combination in the treatment of high-grade AVMs, with cure observed in 15 of 16 patients (93.8%).This cure rate was significantly higher than VS-SRS alone (8 of 58 patients, 13.8%; p < 0.001, Fisher exact test [Table 4]).Rates of mortality and latency hemorrhage were considerable for VS-SRS plus surgery (12.5% and 25%, respectively; Table 4) but not significantly different from the rates of mortality and hemorrhage associated with VS-SRS alone (19% and 29.3%, respectively; Table 4).
Based on our patients' initial Spetzler-Martin grade and data from the original publication of the Spetzler-Martin grading system, neurological morbidity or mortality would be expected in 4.0 patients (Table 5).However, after VS-SRS downgrading and using the same Spetzler-Martin data, neurological morbidity/mortality would be expected in 1.7 patients.Actual surgical neurological morbidity/mortality was observed in 3 patients, thus between these expected values, indicating that VS-SRS lowered the risk associated with AVM resection compared with AVM resection alone.Conversely, these morbidity and mortality figures indicate that the conventional risks predicted by the Spetzler-Martin grading system may be inaccurate with downgraded AVMs and may underestimate surgical risks.Total morbidity/mortality (6 patients) exceeded the highest expectations predicted by Spetzler-Martin grade when radiosurgical morbidity/mortality was included, due to radiation injury and latency hemorrhages (Total morbidity and mortality, Table 5).
Patients with high-grade AVMs usually have 3 management options: 1) observation, 2) staged embolization plus microsurgical resection, and 3) VS-SRS.Staged embolization of large AVMs with high blood flow and cerebral steal occludes large feeding arteries in different vascular territories and redirects blood flow to dysautoregulated arteries in adjacent brain in a stepwise manner, rather than all at once during surgery. 27Although Spetzler and colleagues championed this strategy and reported an experience with 20 patients with morbidity and cure rates of 15% and 90%, 27 respectively, they now favor observation. 9In their consecutive series of 73 patients with Grade IV and V AVMs, only 4 patients (5%) underwent complete surgical removal and 55 patients (75%) were observed with an overall hemorrhage rate of 1.5% per year, which is less than that reported for low-grade AVMs. 9Other neurosurgeons and institutions have embraced a "no treatment" posture for high-grade AVMs, in recognition of a more benign natural history.
Some AVM patients are young with a long life expectancy but they have high-risk features such as silent intralesional hemorrhage, 7 deep location, or deep drainage, all of which increase the risk of future hemorrhage. 5,23,28or these patients, VS-SRS offers a noninvasive option between resection and observation.5]17,21,22,24,25,[29][30][31][32] In a group of 47 patients with large AVMs treated with VS-SRS, "obliteration" rates were 23% for initial VS-SRS after a mean follow-up of 7.3 years; 16 patients required salvage SRS at a mean interval of 61 months, and overall rates of complete obliteration were 18%, 45%, and 56% at 5, 7, and 10 years, respectively. 12In our experience with 69 AVM patients reviewed in a separate report, complete obliteration was observed in just 8 patients (11.6%) after VS-SRS alone, which is an unsatisfactory cure rate (unpublished data, Seymour et al., International SRS Congress, June 19,  2013).
Volume-staged SRS downgrading followed by microsurgical AVM resection combines 2 treatment strategies that are ineffective or risky alone, but are effective together.This approach was previously reported in 1998 by the Pittsburgh group as a case report. 4At our center, this combination was curative in 93.8% of patients, and the one incomplete resection was deliberate after speech mapping demonstrated involvement of language cortex.The rate of surgical morbidity/mortality in this experience (3 of 16, 18.8%) is similar to that reported with other surgical paradigms like staged embolization and resection, but lower than that reported for high-grade AVMs (Grades IV and  * Values correspond to the number of points for each grading system. V) in our previous review of 300 patients for the supplementary grading system (11 of 31, 35.5%). 18m downgrading A dictum of the Spetzler-Martin grading system 26 is that an AVM's grade is defined by discrete anatomy and remains constant and immutable throughout ongoing interventions.Our experience demonstrates that VS-SRS can change AVM grade, with 94% of our cases having lower Spetzler-Martin scores after initial treatment.Most downgrades were due to reductions in nidus size: 13 AVMs lost points for size, including 3 that lost 2 points.Size reduction converted 5 eloquent AVMs into noneloquent AVMs, and occlusion of deep draining veins decreased AVM grade in 3 patients.Our experience also demonstrates that VS-SRS changes an AVM's supplementary grade, but unlike the Spetzler-Martin grade that only decreases after radiation, the supplementary grade can move in either direction.Patient age can only increase, and 5 patients added a point during long latency periods.Hemorrhage during the latency period resulted in the loss of a point in 3 patients with previously unruptured AVMs, and 3 diffuse AVMs compacted as   the nidus shrank.This variability in the direction of point changes resulted in new supplementary grades in only 6 patients (37.5%), of whom 3 upgraded and 3 downgraded.Arteriovenous malformation downgrading complicates surgical risk prediction, which is the objective of the Spetzler-Martin grading system. 26The initial Spetzler-Martin grade overestimates surgical morbidity associated with downgraded AVMs, whereas the preoperative downgrade underestimates surgical morbidity.Therefore, VS-SRS facilitates surgery and reduces operative morbidity, but not to the full extent of the downgraded AVM.Downgraded AVMs retain some of their original complexity that made them more dangerous.

microsurgical avm resection after vs-srs
This treatment paradigm with VS-SRS and microsurgery is effective because radiation induces biological changes that facilitate resection. 24,29Intimal hyperplasia and medial hyalinization thicken arterial walls, narrow their lumens, and occlude feeding arteries. 24Components of the nidus may obliterate and shrink its active volume.Blood flow through the AVM is reduced, sclerotic arteries are easier to coagulate, and surrounding gliosis creates favorable dissection planes.Perforating arteries that supply the deep borders near white matter tracts are transformed from thin and friable to thick and coagulable. 24Diffuse margins can be obliterated by radiation. 24hese benefits notwithstanding, VS-SRS has important risks and half of the morbidity in our experience occurred after radiosurgery and before surgical intervention.Four hemorrhages occurred in this experience.All were within the 3-year latency period; one caused permanent morbidity, and another resulted ultimately in death.Latency hemorrhage is therefore a concern.This hemorrhagic risk following multiple SRS sessions may be more significant than the risk of radiation-induced complications. 16These complications are particularly difficult with high-grade AVMs noted for their lower rate of spontaneous hemorrhage, 9 and in this cohort selected for its young age and good condition.It remains unclear whether these latency hemorrhages were due to a more dangerous natural history risk attributable to radiation, or to the natural history of an unobliterated AVM.

limitations
The indications for radiosurgical downgrading plus microsurgery are narrow and exclude patients who are older, have comorbidities, have previously experienced significant hemorrhage, or have truly inoperable AVMs unlikely to be deemed otherwise after radiosurgery.Pa-  institutional experience with radiosurgery and microsurgery created a unique opportunity to accumulate a sizable group of rare patients within a single center.This treatment paradigm for high-grade AVM patients should be studied further in a multicenter, prospective, observational trial, but this will be difficult in the aftermath of the ARU-BA trial. 20We conclude that volume-staged radiosurgical downgrading with microsurgical resection is a reasonable option to propose to selected patients with long life expectancy and high-risk anatomical features.

conclusions
Volume-staged SRS can be considered a prelude to surgery for large AVMs, with the aim of reducing nidus size.For those patients with favorable response to radiation therapy without cure, the surgical cure rate is high (94%) and the complication profile seems acceptable.This experience demonstrates that high-grade AVMs can be treated with a strategy that begins with VS-SRS and the intent to downgrade the AVM for microsurgical resection, offering an acceptable treatment paradigm for patients with these challenging AVMs.

Fig. 1 .
Fig. 1.Flow diagram illustrating the overall management of high-grade AVMs.Embo = embolization; S-M = Spetzler-Martin.Figure is available in color online only.
ganglia; Embo = embolization; est = estimated; FU = follow-up; HA = headache; med = medial; NA = not available; paramed = paramedian; Sz = seizure.* Prior to VS-SRS, all patients except one had AVMs in eloquent locations.† During the latency period.‡ A good outcome is considered an mRS score of 0-2.§ Delayed second stage of radiosurgery.¶ Complicated by microwire fracture during embolization with right middle cerebral artery occlusion, right temporal lobe subarachnoid hemorrhage, and posterior temporal lobe infarct.table 3. spetzler-martin, lawton-young, and supplemented spetzler-martin avm grades prior to vs-srs and prior to surgery (after vs-srs)* = Lawton-Young.

Fig. 2 .
Fig. 2. Initial, postradiosurgical/preoperative, and final angiograms obtained in patients in this experience.N/A = not available.

Fig. 3 .
Fig. 3. Visual depiction of all AVMs in this series based on initial locations.upper: Approximate locations of the high-grade AVMs prior to treatment are given.The solid circles denote convexity or cortical location, and the broken circles denote medial or deep location.lower: The approximate locations are projected onto the coronal plane.The numbers represent the case numbers in this series as listed in Table 2 and Fig. 2. The upper half of figure has been modified (built upon) from an open license for artwork under Creative Commons Attribution 2.0 License.No usage restrictions.Creative credit: Brain by _DJ_ via Creative Commons.http://www.flickr.com/photos/flamephoenix1991.http://creativecommons.org/licenses/by/2.0.The lower half of figure has been modified (built upon) from an open license for artwork under Creative Commons Attribution 2.5 License 2006; no usage restrictions.Creative credits: Patrick J. Lynch, medical illustrator; C. Carl Jaffe, MD, cardiologist.http://creativecommons.org/licenses/ by/2.5/.

table 4 . outcomes of vs-srs without and with microsurgical avm resection*
* The value in boldface is statistically significant.

table 5 . expected morbidity versus actual morbidity based on spetzler-martin grade
As reported in the paper by Spetzler and Martin. *

table 7 . literature review of srs for high-grade and eloquent avms or those requiring more than one stage of srs
In a clinical experience limited by its sample size, it is difficult to clearly measure morbidity associated with this VS-SRS plus surgery paradigm, or compare this morbidity with natural history risks in a matched control group.Our