Posterior Fossa AVMs

Indications and contraindications for surgery

Brainstem and cerebellar AVMs have a higher annual rate of hemorrhage compared to supratentorial malformations (up to 15.1% and 11.6%, respectively) and should be considered for resection particularly if ruptured, low grade, or accessible from a pial or ependymal surface. Other indications for surgery include progressive neurological decline and residual AVM following stereotactic radiosurgery (SRS) or embolization. The risk of perioperative morbidity and mortality increases with AVMs of larger size, presence of deep venous drainage, and AVM location in an area of eloquence, as estimated by the Spetzler-Martin grading scale. The predictive accuracy of this grading system, however, has been demonstrated to be reduced when applied to AVMs in the cerebellum. This is likely owing to the fact that cerebellar AVMs, as compared to cerebral AVMs, commonly drain into the galenic system and reside within the small-volume space that is the posterior fossa. These factors, as well as the proximity to the brainstem and its delicate angioarchitecture, elevate the technical challenge of resection. A novel grading system was proposed by Nisson et al., who analyzed 120 cases of cerebellar AVMs treated with microsurgery and found that poor outcome was significantly associated with preoperative neurological status, the need for emergency surgery, presence of deep venous drainage, and advanced age. The 33% rate of poor outcome in the lowest-risk group is comparable to the morbidity predicted for treatment of AVMs with high Spetzler-Martin grades (grade IV or V), emphasizing the difficulty in treating these lesions. The association of outcome to the urgency of intervention also suggests that early diagnosis and treatment prior to rupture may improve the chance of achieving a favorable outcome. This updated grading scale also highlights that patients who present with poor neurological grade or are of advanced age are poor surgical candidates. Eloquence in the cerebellum is limited mostly to the deep nuclei and was not found to be a significant predictor.

Brainstem AVMs are typically classified as superficial or parenchymal. This distinction is critical as resection of parenchymal AVMs is associated with a high risk of neurological injury, rendering these lesions better treated with SRS. One exception would be in the setting of hemorrhage, where the resolving hematoma cavity creates a plane of dissection. As a result, mainly AVMs that are located on the pial surface and low grade are considered for surgery. Patients with higher-grade or larger AVMs may benefit instead from SRS with or without prior embolization.

Following hemorrhage, resection is usually performed within 4–8 weeks, providing a time period for the hematoma to liquefy, subsequent cerebral edema to diminish, and cerebral autoregulation to recover. This delay improves the ease and safety of surgery while exposing the patient to a minor risk of repeat hemorrhage, as the rate is modestly elevated compared to ruptured intracranial aneurysms. Acute management focuses on cerebrospinal fluid (CSF) diversion for hydrocephalus, reduction of intracranial pressure, and strict blood pressure control. An early digital subtraction angiogram is necessary to understand the AVM’s angioarchitecture if emergent hematoma evacuation becomes necessary, but also to identify the presence of flow-related or intranidal aneurysms as the source of hemorrhage, which requires urgent endovascular or surgical treatment. Embolic agents include coils or liquid embolic materials such as N-butyl cyanoacrylate (NBCA) or Onyx (Medtronic Inc., Minneapolis, MN).

Role for endovascular therapy

Endovascular therapy plays a critical role in the early treatment of flow-related and intranidal aneurysms in ruptured posterior fossa AVMs. These aneurysms are more commonly found in association with posterior fossa AVMs compared to supratentorial AVMs and are seen as weak points for potential recurrent hemorrhage. The endovascular approach provides a means of occluding the aneurysm without the risk of a craniotomy while the cerebellum is edematous or having to address the AVM emergently. Embolization is also beneficial as a preoperative adjunct to eliminate arterial pedicles and reduce the overall blood supply to the AVM. This aids in the reduction of the overall nidus size for larger AVMs to a surgically accessible size, while minimizing intraoperative blood loss. Deep feeders should be targeted preferentially, as these vessels are exposed later during the resection. It is critical to perform embolization as distally as possible to avoid occluding branches supplying the deep cerebellar nuclei or brainstem, ensuring a sufficient safety margin for reflux of the embolic agent.

While principally utilized as an adjunctive therapy, endovascular therapy with the intent to cure has been attempted successfully mostly for small AVMs supplied by a single, large-diameter arterial feeder, with a small nidus, and with sufficient intraoperative visualization of the draining veins. In a large series of 69 posterior fossa AVMs treated with transarterial embolization over a mean of 2.1 sessions, angiographic cure was achieved in 72.5%. In 21.7% of the cases, there was persistent residual AVM that was subsequently treated with microsurgical resection or SRS. Although 78.3% of the patients achieved a modified Rankin Scale score of 0–2 at last follow-up, 8.8% of patients had a temporary or permanent neurological deficit or died postoperatively. A more recent systematic review including 598 iAVMs treated with embolization demonstrated an even higher complication rate of 24.1%, with hemorrhage secondary to vessel perforation, venous occlusion, or nontarget embolization accounting for a majority of cases. This highlights the greater risk of treatment when endovascular therapy is utilized with curative intent as opposed to being used as an adjunct.

Role for stereotactic radiosurgery

Deep-seated AVMs within the parenchyma of the brainstem are inaccessible with surgery, outside the coexistence of a superficially tracking hematoma, but are excellent candidates for SRS. While this has been the major indication for this treatment modality in the management of posterior fossa AVMs, its use has expanded as a stand-alone or multimodal strategy to address cerebellar AVMs, which were traditionally surgical targets given their more superficial location. Angiographic obliteration rates ranging from 62.5% to 73% have been reported in treating cerebellar AVMs with a median nidus size of 2–4 cm and a majority presenting with hemorrhage. It is recommended that SRS for ruptured posterior fossa AVMs be delayed by 6–12 weeks to ensure accurate targeting of the nidus, which may be compressed or obscured by the hematoma. Higher rates of complete obliteration were observed in younger patients, lesions not treated with preoperative embolization, and smaller nidus diameter. The major disadvantage of SRS is the latency period to AVM occlusion, a median of 60 months, and therefore a delay in hemorrhagic protection. The annual rate of hemorrhage during this interval following SRS has been observed to be 0.85%–2% per year. One series reported a delayed hemorrhage occurring 9 years after treatment and angiographic evidence of complete cure with no residual lesion. Symptomatic perinidal T2 hyperintensities or radiation-induced changes are typically transient but remain permanent in 1.2%–3% of cases, typically occurring around 12 months after treatment. In a large, multicenter series of 162 cerebellar AVMs, 62.2% of patients achieved a favorable outcome with complete obliteration in the absence of hemorrhage or fixed deficit. SRS is also indicated for patients with significant comorbidities rendering them poor candidates for surgical treatment or for residual posterior fossa AVMs following embolization or microsurgical resection.