Surgical Management of Cerebral Arteriovenous Malformations

Introduction

Intracranial vascular malformations represent approximately 6% of all diseases that affect the central nervous system. They are classified according to their hemodynamic patterns in terms of arterial malformations (aneurysms), arteriovenous shunts (cerebral arteriovenous malformations), dural and pial fistulae, venous malformations (cavernous and venous angiomas), and capillary malformations (telangiectasias).

Another type of vascular malformations that involve different angioarchitecture and preponderance in the pediatric age group (although it is possible to see these cases in adults) are proliferative malformations. These are diffuse lesions and include moyamoyadisease, cerebral proliferative angiopathy, and hemangiomas. It has been hypothesized that the underlying pathogenic mechanism here involves altered endothelial proliferation and angiogenesis. The incidence is very low.

This chapter will be dedicated to the treatment of arteriovenous malformations (AVMs); cerebral proliferative angiopathy also will be briefly discussed, as it is closely related to AVMs and may sometimes require surgery if hemorrhagic complications arise.

By definition, an AVM is a direct communication between arterial and venous vessels without the interposition of a capillary bed. AVMs can be plexiform or fistulous. Macroscopically, they present as a tangled cluster of abnormal vascular channels with afferent arteries, a nidus (which is compact in most cases), and draining veins. There is no intervening brain parenchyma between the abnormal vessels. AVMs are surrounded by a modified gliotic tissue that is very useful at the time of surgical resection.

Pathologic dilated veins, secondary to nidus hypertension, often lead to the formation of venous aneurysms that are frequently responsible for bleeding. The brain parenchyma around an AVM is subject to chronic hypoperfusion caused by steal phenomena and also is subject to venous hypertension. This leads to neuronal loss and, in some cases, even regional atrophy and ischemic events that often explain the clinical manifestations of an AVM, other than bleeding.

AVMs can be found all along the cerebrospinal axis, but they largely predominate in the supratentorial space. Less frequently, they can originate in the posterior fossa (15%) and rarely in the spinal cord (less than 5%). It is estimated that the annual cumulative risk of bleeding for an AVM ranges between 2% and 4%. ^{, , , ,}

Even though they are congenital lesions, AVMs become symptomatic between the third and fifth decades of life. However, symptomatic AVMs may be diagnosed at any age.

The first clinical manifestation is intracranial hemorrhage in 50% of the cases—mainly intracerebral bleeding and less frequently subarachnoid hemorrhage or subdural hematomas. Typically, another clinical presentation is the occurrence of seizures or focal deficits due to ischemia, secondary to blood stealing. Large AVMs might present with murmurs or persistent headaches, explained by venous hypertension. With the widespread use of CT scans and magnetic resonance imaging, a nonnegligible portion of AVMs are incidentally diagnosed. ^,

Writing about the surgical management of AVMs goes far beyond opening the skull and resecting the nidus. Appropriate management involves knowledge of preoperative or palliative endovascular therapies and radiant treatments—mainly stereotactic radiosurgery.

It is likely that the treatment guidelines that we present herein might change or even become obsolete in the near future, considering the constant advances in microsurgery, new procedures, and devices for endovascular therapy and the advances in radiosurgery that we see nowadays. It is thus somewhat controversial to propose treatment guidelines in such a dynamic and constantly evolving field. New studies with a long-term follow-up of our patients are essential to assess the efficacy and outcomes of a certain treatment. The management of AVMs, and particularly some subtypes, poses a true challenge, where multimodal approaches and combined treatments are necessary.

History

We should consider the fact that AVM treatment is a relatively new field of study. Before the 1960s, there were very few papers on this subject, and most of them referred to isolated cases. After that decade, surgery for AVMs became more commonplace, and the first endovascular treatments as a preoperative adjunct appeared.

Since the first publication on a preoperative embolization of an AVM performed by Luessenhop and Spence (1960), endovascular therapy has continuously evolved, with a drastic change in the last 20 years that nowadays allows almost complete occlusions of AVMs with this type of treatment. ^{, ,} Also in the 1960s, the surgical microscope was introduced, allowing for a 180-degree-change in microneurosurgery. These advances in microsurgery, along with the development of more sophisticated and intelligent microscopes that can even perform intraoperative vascular studies, have led to a significant improvement in surgical outcomes. Furthermore, some AVMs that would have been considered inoperable in the past can now be operated on. ^,

Laboratory training in microsurgery and neuroanatomy has also advanced the practice of treatment of AVMs. The further development of microsurgical instruments and expertise in vascular neurosurgery has led to low mortality and morbidity rates in AVMs treatment. At present, these rates are so low in some cases that they would have been impossible to imagine 15 years ago.

Finally, in the decade of the 1980s, radiosurgery appeared as another option for treatment. The use of radiosurgery as a complementary treatment or as the only treatment in some cases increased the number of AVMs that could be treated, and the outcomes improved even further. In conclusion, one could say that the practice of treating AVMs is only about 55 years old.

Classification

Knowledge of AVMs’ angioarchitecture and hemodynamics has increased over the years. The experience of vascular neurosurgeons has made it possible for them to propose different grading systems for AVMs, taking into account various parameters, including imaging studies. In addition, tools such as angio-CT scans with 3D reconstruction, magnetic resonance with angio-MRIs, functional MRIs, and diffusion tensor imaging have all become very useful for the planning of surgical approaches.

However, digital subtraction angiography (DSA) is still the most important tool for use in the practice. It offers great anatomic detail and a dynamic view of the AVM’s hemodynamics. A detailed assessment of angioarchitecture and hemodynamics allows the neurosurgeon to decide if exclusive surgical treatment is indicated or if preoperative endovascular therapy is needed.

Careful analysis of the pre-nidal, nidus, and post-nidal components of the AVM in digital angiography is very important. The pre-nidal component includes the afferent arteries that feed the AVM. One should determine the number and characteristics of these vessels. The main vessel should be identified, and the existence of flow aneurysms should also be ruled out. The shape and size of the nidus should then be assessed, along with the presence of intra-nidal aneurysms. The post-nidal component is crucial for the management of AVMs. Draining veins must be recognized. One should determine the length of the veins and if they drain to the superficial and/or deep venous systems. Venous stenosis, occlusions, tortuous dilations, or venous aneurysms must be identified, as they predict a higher risk of bleeding. ^{, , ,}

In 1977, Luessenhop and Gennarelli published a grading system for AVMs based exclusively on their arterial feeding pedicles. Later on, in 1984, Luessenhop and Rosas published a new classification that also included the size of the nidus. The most recent grading scale that takes into account only the angiographic findings is Nataff’s scale ( Table 57.1 ). It defines four groups of AVMs, according to their angiographic bleeding risk factors. ^,

Classifying AVMs has proven to be very useful for the attending team when the moment arrives to decide whether to treat a particular case. The aforementioned classification of Luessenhop and Rosas was historically easy to apply, but it was not practical for surgery, as it did not consider venous drainage or proximity to eloquent areas. Sugita classified AVMs of the sylvian fissure according to their relation to it (lateral, medial, deep, or within the sylvian fissure). In the early 1980s, B. Borovich proposed another classification that divided AVMs into deep or superficial and considered the number of arterial afferents. At the same time, Yasargil also classified AVMs as deep or superficial but, on the contrary, took into account the size of the nidus: up to 2 cm, between 2 and 4 cm, or more than 4 cm. ^,

In 1986, Shi and Chen published a grading system that considered the size, location and depth, arterial afferents, and venous drainage. This was a complex classification and was difficult to remember, as it posed several possible combinations (with seven grades total). However, in the same year, Spetzler and Martin developed a new grading system that took into account the size, location, and venous drainage of the AVM. This classification was easy to apply and corresponded well with surgical findings and prognoses. This rapidly became accepted worldwide. In fact, it was used without any modifications until recently. With time, some difficulties were encountered, particularly with grade III AVMs, not only in determining the appropriate treatment but also in establishing an outcome. This is explained by the considerable variability that can be found in this group. For instance, small AVMs with deep venous drainage in eloquent areas (i.e., basal ganglia, brain stem) and a large nidus of more than 5 cm, superficially located with superficial draining veins, both belong to group III. Hence, some authors have proposed modifications to this grading system. The most accepted one is the division of group III in A and B, depending on the superficial or deep location of the nidus. Another issue is that Spetzler and Martin’s classification is exclusively designed for supratentorial AVMs and is not applicable in cases of posterior fossa malformations. This classification has also been criticized because it does not consider the medical condition of the patient and the existence of previous bleeding episodes of a particular AVM. ^, Thus, in 2010, M. Lawton proposed a supplementary classification to better predict surgical outcomes. It adds five more points to the Spetzler-Martin classification, depending on the age of the patient, the presence of hemorrhage, and the characteristics of the nidus (compact or diffuse).

Arteriovenous Malformations

AVMs are clusters of abnormal arterial and venous vessels, without intermediate capillaries and no intervening normal parenchyma in between. They are usually surrounded by a thin nonfunctional layer of gliotic tissue that allows the surgeon to resect the malformation without lesioning the normal brain. ^,

Regardless of the different classifications mentioned previously, at the moment of surgical planning, it is important to analyze and understand both the superficial and deep afferent vessels of an AVM. This information, together with a detailed study of the anatomy, angioarchitecture, and blood flow of the lesion, will result in different treatment strategies and even different surgical techniques. This is why management guidelines are necessary. Such guidelines should always be dynamic and updated. Treatment guidelines are recommendations made by a group of neurosurgeons dedicated to vascular neurosurgery, and they should be considered as a general basis to start planning a treatment scheme for a particular case.

In the last 15 years, many authors have proposed treatment guidelines (e.g., Vazquez and Larrea in 2000, Ogilvy et al. in 2001, and Starke in 2009). ^{, , , , ,} The senior author of this chapter was part of the group that in 2009 published the “Management Guidelines of the Latin-American Federation of Neurosurgery for Brain AVMs.” These guidelines are still up to date and intended to help neurosurgeons make decisions regarding the treatment of an AVM. The treatment recommendations of the Latin-American Federation of Neurosurgery (FLANC) are similar to those elaborated by the other authors mentioned previously.

AVMs are dynamic lesions, and it is thus difficult to predict when they can present complications like hemorrhage or provoke seizures or neurologic deficits. This is why it is also very hard to calculate the number of symptomatic AVMs. It is estimated that between 2% and 5% of AVMs are diagnosed because of symptoms other than bleeding. Any diagnosed AVM, left untreated, depending on its characteristics and in a variable period of time, will probably develop changes in its anatomy and clinical manifestations that may lead to a hemorrhage. Hence, all low-grade AVMs (I or II) should be treated, as the morbidity and mortality of conservative management outweigh the risks of surgery. Nowadays, grade III AVMs should also be considered for treatment. In this group, the best option for treatment is somewhat debated, even though the senior author of this chapter considers surgery to be the ideal treatment with the best results. Depending on the case, combined treatments can be performed in an attempt to permanently eliminate the lesion. A curative treatment should always be attempted for grade III AVMs, and one should avoid speaking of palliative treatments.

On the contrary, high-grade AVMs (IV and V), if uncomplicated, should be observed, as the risks of treatment outweigh those of the natural history of the lesion. When these lesions bleed or provoke symptoms (neurologic deterioration, medically refractory seizures), then the risks change and aggressive management is advocated. Management in these cases involves a combination of treatments.

Treatment

Curative treatment of an AVM is undoubtedly one of the greatest challenges for a neurosurgeon. The choice of treatment modality is still controversial sometimes, as continuous progress in therapies alternative to conventional surgery provides noninvasive therapeutic options with good long-term results, according to some publications. Endovascular embolization and radiosurgery are alternative therapies. ^{, , ,} However, in many cases, these treatments should not be considered as alternative but complementary. Endovascular therapy, followed by surgery and complementary radiosurgery, might sometimes provide a solution for a particular patient. ^{, ,}

In any case, it is clear that what we seek in AVM treatment is to eliminate the risk of bleeding and to achieve a cure for the patient without causing neurological damage. Despite the great progress in endovascular therapy and in radiosurgery, complete microsurgical resection of an AVM is still the only treatment that can guarantee a cure.

In his 1979 publication, C. Drake proposed five options for the neurosurgeon to take into account when deciding the best course of action for an AVM :

• 1.

  Conservative management (no surgery or any other treatment except for pharmacological symptomatic treatment; e.g., antiepileptic drugs).

• 2.

  Surgery

• 3.

  Endovascular therapy

• 4.

  Radiotherapy (now radiosurgery)

• 5.

  A combination of the above

Drake’s proposal never became obsolete. These are the five options that all neurosurgeons consider nowadays when treating an AVM.

After a complete surgical resection of an AVM, certified by an angiographic study that shows no remnants, the patient can be considered totally cured, without any risks of recurrence. Even if this is what happens in most cases, there are some reports of AVM recurrence after an apparent complete resection. ^, In the authors’ opinion, in these cases, the postoperative angiography failed to demonstrate the presence of a small remnant that recruited more vessels and eventually reproduced the AVM. In fact, the possibility of a recurrence is almost inexistent after a complete microsurgical resection with a normal postoperative angiography. On the other hand, a patient exclusively treated with endovascular therapy, with 100% occlusion of the AVM in the angiographic control, should not be deemed cured, as small pedicles or many pial vessels not seen in the angiography might persist and reproduce the lesion in the future.

The authors agree with M. Lawton’s statement (published in 2015) : “Surgical resection is still the gold standard for the treatment of an AVM, providing immediate angiographic cure and protection against future ruptures. The use of radiosurgery as an alternative treatment is increasing, mostly for small AVMs that have not bled, knowing that the risk of hemorrhage persists several years after treatment. Last, endovascular embolization is generally not used to cure an AVM, but as techniques and selection criteria improve, obliteration rates of the nidus are getting better.”

We agree that a patient exclusively treated with endovascular embolization should not be considered cured. The literature on this matter shows complete angiographic occlusion rates, with 5 years of follow-up, that vary between 13%, 45%, and 64%. These rates are far from the 99% complete elimination of AVMs treated with complete surgical resection and negative postoperative angiographic control.

Better knowledge of the natural history of an AVM, its blood flow, pressure within the nidus, functionality of the AVM, and characteristics of the surrounding brain have led to significant improvements in treatment outcomes in the last 2 decades. Neuroimaging studies have made great contributions to our understanding of AVMs, with advances such as supraselective digital angiography, functional magnetic resonance sequences (diffusion, perfusion, and diffusion tensor imaging), magnetoencephalography, and PET scans. Thanks to all the information provided by these studies, surgical planning is now more accurate, and the need for endovascular preoperative embolization or postoperative radiosurgery can be better defined. ^{, ,}

Other factors that have changed the outcome of patients undergoing AVM surgery are modern neuroanesthesia and intraoperative neurophysiologic monitoring. Anesthetic neuroprotection allows transitory clipping of important arterial vessels with very low risk of brain damage. Neurophysiologic monitoring has become indispensable for many neurosurgeries, particularly vascular neurosurgery, not only when operating near eloquent areas but also in ensuring and controlling cerebral protection when arterial clipping is needed.

All of these adjuncts increase the rates of success, with lower functional risks for the patient. Modern neurosurgical techniques, neuroanesthesia, and neurophysiology have made it possible to operate on lesions in eloquent areas or in deep-seated areas that would have been considered inoperable years ago. Nonetheless, the treatment of AVMs is and will continue to be challenging for neurosurgeons. The decision of whether to treat or not, along with when and how to treat, is fundamental. This responsibility is even more pressing after the 2014 publication of ARUBA. ^{, ,} This trial assessed the benefits of treating unruptured AVMs. The authors concluded that treatment does not improve the outcome compared to natural history, mainly for small AVMs without significant bleeding risk factors. However, the bleeding risk is never 0%. In the best cases, it reaches 0.7%. In our opinion, this is very dangerous, because no matter how low the risk is, the malformation can still bleed, particularly in young patients. Thus, depending on the expertise and results of the attending surgeon, the elimination of AVMs with certain characteristics should always be attempted. The authors’ opinion is that the risk of surgery (combined with other treatments or not) is significantly lower than that of the natural history of the disease.

When and How to Treat an Arteriovenous Malformation

The answer to these questions should be guided by the symptoms that the patient presents with. If an AVM has bled, then it should always be treated. This is absolute for grades I to III. For grades IV and V, after a hemorrhage, a multidisciplinary team should discuss each particular case and offer the patient the less-risky option. Sometimes, this is conservative treatment. However, if multiple hemorrhages occur, aggressive treatment is warranted. ^{, , , , , , , , , ,}

If the symptom is epilepsy, for grades I to III, treatment is also the first option. Epilepsy reflects irritation of the surrounding brain. AVMs that provoke epilepsy can be difficult to control, and there is always the risk of hemorrhage, so treatment is consistently recommended for AVMs grades I to III with epilepsy. For grades IV and V, adjustment of medical treatment with antiepileptic drugs is recommended, because the risk of directly treating the AVM is always higher. ^{, , ,}

When the symptoms are focal deficits or those other than epilepsy, the same rationale applies. Grades I to III should be treated. For grades IV and V, with cognitive impairment or progressive neurological deterioration, therapeutic options should be proposed in an attempt to control the symptoms. Palliative treatments are generally suggested, as we will explore later on.

Finally, we previously noted the value of angiography, not only for the diagnosis of an AVM but also for a detailed anatomic and hemodynamic knowledge of the lesion. When deciding on treatment, one should take into account the existence of venous aneurysms, venous stasis, arterial flow aneurysms, tortuosity, and the caliber of draining veins. If angiographic risk factors that predict bleeding are encountered, then surgery is recommended. For example, radiosurgery could achieve a cure for a small AVM, but this could take up to 3 years. During this period, the risk of hemorrhage persists, even more so if angiographic risk factors for bleeding are detected. So, in these cases, indicating radiosurgery would be a mistake ( Fig. 57.1 )

Grade I Arteriovenous Malformations

Grade I AVMs should always be treated. The anatomical and angiographic characteristics of the AVMs in this group make them relatively easy to resect with conventional surgery.

Direct surgery without any other previous treatments is indicated and curative. Morbidity and mortality of surgery for grade I AVMs are extremely low. Direct surgical resection is by far the first therapeutic option to consider for grade I AVMs. ^{, , , ,} If the lesion did not bleed, the second option would be radiosurgery. One should always remember that radiosurgical treatments have a latency that might require 3 years to secure the disappearance of the malformation. The risk of bleeding persists during this period, and even though it is low, it is not 0%. Morbidity is also very low with this technique. Endovascular embolization is not discarded. The problem is that, as we said before, angiographic disappearance does not ensure a cure, as small pedicles that are impossible to detect in angiography might persist and reproduce the nidus. This may require re-treatments or could eventually provoke a hemorrhage. However, there are publications that report complete endovascular occlusion of AVMs with long-term follow-up and no evidence of recurrence. This treatment poses some risk of morbidity and mortality, but the rates are extremely low ( Fig. 57.2 ).

Treatment of grade I AVMs:

• 1.

  Surgery

• 2.

  Radiosurgery

• 3.

  Endovascular embolization

• 4.

  Do nothing

Grade II Arteriovenous Malformations

Grade II AVMs should also be treated. The risks of the different therapeutic options (i.e., the morbidity and mortality rates) are by far lower than those associated with the natural history of the disease. As in grade I AVMs, conventional surgery is the first option to ensure that the patient is cured. Trained vascular neurosurgeons should not experience any difficulty completely resecting the lesion without causing secondary damage.

Some AVMs in this group might benefit from endovascular embolization in order to occlude afferent vessels in the preoperative period. However, this cannot be considered a sufficient treatment per se, even if complete angiographic obliteration of the AVM is achieved. By no means is this as equally effective as eliminating the lesion. There are always small pedicles not seen in angiography that, if left untreated, sooner or later will lead to a recurrence of the AVM. During surgery of grade II AVMs previously treated with endovascular embolization and with 100% angiographic occlusion, it is actually frequent to find small, and not so small, arterial pedicles feeding the nidus. This is even more evident when, under microscopic magnification, the surgeon can identify arterialized veins. This means that there must be patent arterial afferents. The senior author of this chapter has repeatedly verified this in all AVMs previously treated with endovascular therapy, presenting with apparent angiographic cure. During microsurgical resection, while the surgeon dissects all around the AVM, coagulating these small arterial pedicles, the arterialized veins (which are red in color) begin to take on a blueish color. This is a good way to prove during surgery that all afferents have been eliminated and that the main draining vein can be cut to completely remove the nidus.

Radiosurgery is never a first-line treatment for grade II AVMs. There might be considerable morbidity if the nidus is not small, and as noted previously, there is a latency of several years to achieve significant obliteration. The patient is then exposed to the main complication, which is bleeding. ^{, , , ,}

Combined treatments for grade II AVMs are not the ideal. Preoperative endovascular embolization can be done in an attempt to occlude arterial afferents and reduce the risk of intraoperative hemorrhage. Such risk is low for a conventional grade II AVM surgery. It has long been said that endovascular embolization transforms the nidus into a hard mass that can afterward be removed as a tumor. This statement is somewhat deceiving. A firm, rigid nidus is hard to mobilize, which makes it difficult to properly visualize and coagulate small arterial afferents. Manipulation of a rigid nidus might end up in traction and lesion of draining veins with unwanted bleedings. On the contrary, a non-embolized nidus is more easily dominated; there are no difficulties in its mobilization and the surgeon can control how it changes when afferents are coagulated. These changes involve loss of turgor, volume reduction, and color change in the draining veins.

A combination of surgery and radiosurgery can sometimes be useful for a few grade II AVMs where the nidus is large and diffuse. Surgery carries a risk of leaving a remnant in these cases. Combination therapy can be applied to AVMs that have not bled, in cases where the remnant of the nidus is very small. Leaving a remnant after surgery for a grade II AVM is very unlikely but not impossible. In recent years, with the use of intraoperative angiography and new microscopes with vascular study technology, this is even less likely. Intraoperative imaging allows for the identification of these small remnants, and the rate of complete resection is increased.

Finally, a combination of nonsurgical therapies, like endovascular embolization and radiosurgery, should not be considered for grade II AVMs, especially if the AVM has bled. There is a high risk of hemorrhage during the waiting period required for the lesion to disappear with the effect of both treatments. If the AVM has not bled, this combination of treatments is not justified, considering that there is another curative therapeutic option with low risks, like direct surgery.

Endovascular embolization has a low but considerable morbidity, and it does not guarantee complete obliteration. Radiosurgery also has a very low morbidity, but the effects take years. In conclusion, a combination of endovascular therapy and radiosurgery is generally not an option for grade II AVMs.

Treatment of grade II AVMs:

• 1.

  Surgery

• 2.

  Endovascular + Surgery

• 3.

  Surgery + Radiosurgery

• 4.

  Radiosurgery

• 5.

  Endovascular + Radiosurgery