History

AVM neurosurgery is a relatively recent innovation. Even though there have long been reports about the surgical management of this kind of lesion, only from 1960 onward has the veritable surgery become a reality. In the 1960s, endovascular treatments started to be performed as a way to assist neurosurgeons in the resection of such lesions. Since the first description of embolization by Luessenhop and Spence in 1960,¹ endovascular procedures have been improved, enabling a dramatic change in the direction of AVM treatments. In that same decade, with the appearance of the operating microscope, surgery took a 180-degree turn in everything regarding neurosurgery, including neurovascular surgery.¹ Permanent laboratory practice in microsurgery, better and continued training in neuroanatomy, and constant development of surgical instruments for microsurgery, together with the supraspecialization in vascular neurosurgery, have led neurosurgeons to acquire more practice in the resection of such lesions, with a consequent decrease in morbidity and mortality. Eventually, radiosurgery came up and started to become popular as an adjuvant treatment—or as the only treatment for some cerebral AVMs—in the 1980s.

Fifty years! Thanks to the appearance of the different procedures and technologies mentioned above, their improvement, and the combination of all of them, it can be said that such is the age of the AVM surgery.

Classification

As AVMs became known, particularly their characteristics and the difficulties they posed in each case, they started to be classified according to different factors. Such classifications sought and still seek to help neurosurgeons decide whether to treat a cerebral AVM and what therapeutic procedures would be most appropriate for each case.²

In 1977, Luessenhop and Gennarelli proposed a classification based on the arterial pedicles that supplied AVMs.^1,17 This classification did not take into account the draining veins or whether AVMs were located in eloquent areas. In 1984, Luessenhop, with Rosa’s collaboration, published a grading scale based on AVM size.^4,18 Even though it was an easily applicable classification, it was not practical for surgery because it did not consider the venous drainage or whether it was in an eloquent area. Sugita also classified AVMs located at the sylvian fissure, taking into account whether they were directly in the fissure, lateral, medial, or deep seated.³ In 1986, Shi and Chen published a classification that took into account size, location, depth, the feeding arteries, and the venous drainage.^4,18 It was a complex classification, one difficult to remember due to the various combinations therein (since it had seven grades), and the determination of eloquent or noneloquent areas was imprecise. But in that same year Spetzler and Martin’s classification was published, which took into account size, location, and venous drainage.⁴ This classification turned out to be the most practical one, and its diffusion and acceptance worldwide were fast and unanimous. As time went by, and with the application of this classification, it became evident that AVMs in grade III presented different difficulties in treatment and prognosis. This was motivated by the variability of the lesions included in this grade: for example, it included a small AVM with deep venous drainage located in an eloquent area, such as the brain stem, as well as a 6-cm-wide cortical lesion with superficial venous drainage located in a noneloquent area. This range led several authors to propose modifications to this grade of the classification. The most accepted ones are grades IIIA and IIIB, regarding location.^3–6

Even though modifications are still being proposed, the Spetzler-Martin classification is the most commonly used at present, and treatment criteria are mostly based on it. Regardless, with the existing research possibilities on brain function, it is becoming increasingly difficult to speak of eloquent and noneloquent areas. In one way or another, practically the whole brain is “eloquent.”

Based on my experience, as well as on the work that has been carried out in Uruguay for year, and the concepts given by Borovich in the 1980s,⁶ I would add to the Spetzler-Martin classification the arterial territories involved in AVMs. They are grouped according to whether one, two, or three vascular territories are involved. Most AVMs are included in the groups for one or two vascular territories, which correspond, to a great extent, to grades I to III in the Spetzler-Martin classification.⁶

Treatment

In a publication about the management of AVMs in 1979, almost 20 years after the beginning of the AVM era, Drake proposed five options neurosurgeons could use when dealing with such a lesion⁷:

What Drake proposed has never become obsolete. The five points suggested are still present in the neurosurgeons’ considerations when dealing with cerebral AVMs.

After complete surgical resection of an AVM, with no evidence of remaining lesions in the angiographic control, the patient can be deemed cured. Although this happens in the vast majority of the cases, AVM relapses have been reported.⁸ It is my opinion that in such cases the angiographic subtraction control study failed to reveal small pathologic pedicles, from which the lesion subsequently reproduced.

Better knowledge of the natural history of AVMs, their flow, the pressure at the nidus, and AVM functionality, as well as that of the surrounding brain, has permitted substantial improvement in the prognosis of such lesions in the last two decades. This has been achieved through the use of dynamic studies, such as supraselective digital angiographies, digital subtraction angiography combined with color scaling, diffusion and perfusion magnetic resonance imaging with flow study, magnetic resonance imaging tractography, magnetoencephalography, and positron emission tomography. All these techniques allow better knowledge of the malformation, its behavior, and the characteristics of the surrounding brain, thus enabling the surgical team to plan the best treatment or combination of treatments for each case.^9–16

Constant advances in neuroanesthesia with brain protection allow complex surgeries, sometimes with transitory interruption of blood flow through the arteries (transitory clipping), to become less harmful. The association of neurophysiologic intraoperative monitoring with intraoperative control of the areas where the surgeon is working, as well as control of the remaining brain functions during transitory clipping, further increases the possibility for success in such surgeries. In turn, they permit surgeries that, due to the topography or complexity of the lesion, would have been impossible to perform in the past.

Arteriovenous Malformations

Vascular malformations of the brain (excluding aneurysms) constitute a heterogeneous group of lesions that includes AVMs, pial malformations, cavernous angiomas, venous malformations, dural fistulae, and telangiectasias. Therefore, AVMs widely dominate the spectrum of lesions.¹⁷

Cerebral AVMs are lesions of likely congenital origin or predisposition constituted by a conglomerate of abnormal vessels (both arterial and venous) of variable size and number. There is no intermediate capillary network; there is no parenchyma among vessels. Usually, AVMs are surrounded by a thin layer of nonfunctioning brain tissue (reactional gliosis).^6,17

Independent of their classification, AVMs can be divided into one of two categories, superficial (also called “of the convexity,” “cortical,” or “corticosubcortical”) or deep, when planning for their treatment.^6,18 Given the variability of these lesions, in both their anatomy and their physiology, there is no universal consensus about how to treat them, even those AVMs that a priori could be similar. Thus, there have long been attempts to establish standards and guidelines for the treatment of AVMs. The problem lies in rapid changes to the proposed guidelines arising from development in diagnostic techniques and knowledge about AVMs and their treatments.

Regardless, the guidelines are recommendations formulated by groups of professionals dedicated to a particular pathology and must be taken as a base for the treatment of AVMs. They are flexible and can be modified according to the results of the different possible treatments and to the appearance of new therapeutics.^19,20

Through its chapter on vascular neurosurgery, the Latin American Federation of Neurosurgical Societies in 2003 formulated the first recommendations for the management of AVMs.¹⁷ In 2009, I led a group of neurosurgeons dedicated to vascular surgery who published an updated version of those recommendations under the (translated) title “Recommendations for the Management of Brain Arteriovenous Malformations.”¹⁷ This publication aims to offer the specialist a guide for managing such lesions, but by no means is strict compliance necessary.^17,19

In the last decade, several authors have published guides with the same objective (e.g., Vazquez and Larrea in 2000,²¹ Ogilvy et al. in 2001,¹⁹ and Starke et al. in 2009²²). These guides do not substantially differ in their recommendations, proposing surgery for grades I and II, combination of treatment for grade III, and conservative treatment for grades IV and V. Starke et al. conclude that surgery is the ideal and curative treatment. They advocate surgery, endovascular surgery, radiosurgery, or combination therapies depending on various factors: angioarchitecture of the AVM, topography, experience of the surgical team, and so on.^19–22

AVMs are lesions whose incidence ranges between 1.4% and 4.3% of the population.^23,24 The percentage of symptomatic AVMs is unknown, but it has been proved that they are “dynamic” lesions.^23,24 In other words, a cerebral AVM diagnosed but not treated can show in a high percentage of the cases, under a periodic follow-up, changes in its anatomy, size, and symptomatology.^23,24 Hence, the conservative (do nothing) treatment should not be considered in low-grade malformations or in high-grade AVMs that have shown important clinical aggressiveness, such as repeated bleeding and epilepsy that is difficult to control.

When a treatment for an AVM is proposed—which should be a curative one—it should have a lower morbidity and mortality than those accorded by the natural history of that particular malformation. The treatment or treatments to be carried out in each case issue from the discussion of a multidisciplinary group formed by neurosurgeons, neurointerventionists, neurophysiologists, neurointensivists, neuroanesthesiologists, and neurologists. With all the necessary studies that enable understanding of an AVM (computed tomography, magnetic resonance with functional components, digital angiography, etc.), the team can then decide whether to treat the lesion. If treatment is chosen, one or a combination of the previously mentioned options should be chosen. Every time the team decides to treat an AVM, a curative treatment should be sought, which means the elimination of the lesion. If the treatment was meant to be curative but a remnant of the lesion remained, the treatment is considered a failure. Frequently, palliative treatments are considered, for instance, when dealing with grade V AVMs with recurrent bleeding. In such cases, we know that the possibility of curing the patient is low, but a palliative treatment may significantly reduce the possibility of future bleeding.^25–28

When and How to Treat a Cerebral AVM

Cerebral AVMs manifest clinically through bleeding in 50% of the cases, followed by epilepsy. When an AVM has bled, it must be treated. When we face epilepsy refractory to the medical treatment, it must also be treated. When we face an AVM with no bleeding or significant clinical manifestations but with studies that show it presents intranidal aneurysms, stenosis of the afferent vessels, or significant venous stasis, it also must be treated. If AVMs are small and deep seated, they also must be treated, since it is accepted that they have a greater chance of developing complications. Regardless, studies have not been able to demonstrate—based on levels of evidence—a clear correlation between the characteristics of the lesion and its chances of bleeding.^26,27,29,30

Following the analysis of malformations according to the Spetzler-Martin grading scale, the following guidelines could be established.¹⁷

Grade I and II AVMs

Grade I AVMs should always, in principle, be treated. Due to their characteristics, they should not pose any difficulty for the surgeon. Direct surgery without prior endovascular treatment is the indicated, and curative, procedure. Morbidity and mortality rates are extremely low. If this malformation has not bled, the second option is radiosurgery. There is a period (perhaps up to 2 years) during which the malformation can bleed until its disappearance.

The endovascular option would be, for this group, in the third place. Even though they should not represent any difficulty for the endovascular surgeon, the angiographic disappearance of the lesions does not ensure they have been cured. Small, nonvisible pedicles may exist that may make the nidus reappear.^{2,5,6–19,21–23,25,31–37} Even though previous papers mentioned total and definitive occlusion with histoacryl glue,³⁸ recent publications about treatment with Onyx reinforce that grade I AVMs can be cured with endovascular treatment (Fig. 83-1).^39,40

Grade I AVM

1. Surgery

2. Radiosurgery

3. Endovascular therapy

4. Do nothing

FIGURE 83-1 Grade I AVM. Only one pedicle, small size.

In principle, grade II AVMs should be treated. The natural risks are higher than those consequent to the treatment. As with grade I AVMs, surgery is the first therapeutic option for this group. It ensures the complete cure and, due to the characteristics that determine lesions’ inclusion in this grade, they should not indicate a priori a major difficulty for a well-trained neurosurgeon. Endovascular therapy (endovascular neurosurgery) may be of help for the surgeon as treatment prior to surgery. However, it is not advisable as the sole therapeutic option. Although it is possible to achieve an “angiographic cure” by means of endovascular therapy, it is highly probable the presence of small pedicles not revealed in the studies that will determine the reappearance of the lesion, evidenced in early or later angiographic controls (1 or 2 years later) if it has not bled previously. Surgeons are well aware that when operating patients with this grade of AVM (with complete angiographic closure after the endovascular therapy), arterial open pedicles are always found and even some arterialized veins are seen. In surgery, as the nidus is being surrounded and these small pedicles are coagulated, the aspect of the lesion visibly changes.

Radiosurgery is not indicated as a stand-alone option in this group of lesions. It can be considered a second-line treatment after endovascular therapy has been carried out, but it has been proved (as in AVMs of higher volume) that the area to be radiated does not include the whole nidus and that areas treated by endovascular means may not be included, thus leading to lesion reproduction in the long term.^13,17,75 Even though the proposal to always treat grade I and II AVMs may be controversial, the specialized bibliography supports my strong recommendation for surgery (Fig. 83-2).^{2,5,6–19,21–23,25,29,31–37}

Grade II AVM

1. Surgery

2. Endovascular therapy + surgery

3. Endovascular therapy

4. Radiosurgery

5.Surgery + radiosurgery

FIGURE 83-2 Grade II AVM. Two pedicles, small size, eloquent area.

Grade I and II AVMs have with surgery (alone or combined) a cure rate close to 100%.^{17,31,32,36,37,41,42}

Grade III AVMs

Grade III AVMs must always be treated. They can become highly problematic due to the extensive number of variants that can appear within this group. AVMs that have a large nidus, are superficial (corticosubcortical), and involve one or two vascular territories must always be treated by means of combined techniques. Endovascular therapy is essential in these cases (in either one or two sessions) and is aimed at progressively occluding the afferent vessels (through which inversion of the flow is achieved) but not causing a sudden redistribution of it, which could be harmful. Endovascular therapy with Onyx offers excellent results in this group of AVMs as a treatment prior to surgery.^43,44 A prudent amount of time should elapse between this treatment and surgery (no less than 4 or 5 days). After the surgery, and by means of a control angiogram, the complete elimination of the nidus must be verified. If small remnants remain, the treatment can be complemented with radiosurgery.

Direct surgery can be performed in these AVMs, but a grade III malformation that was embolized beforehand clearly bleeds less, the surgery time diminishes significantly, and the possibility of sequelae lowers dramatically. Grade III AVMs, although small, are always deep seated. They pose technical and surgical difficulties, but not unsurpassable ones, as explained later. According to their topography, it is possible to plan the surgery with assistance (e.g., neuronavigation and stereotaxy).

Endovascular therapy (depending on the afferent pedicles) with radiosurgery and radiosurgery alone are valid options in these cases.

Most lesions within this grade, when treated, have a cure rate close to 100% but a morbidity rate close to 25%.^27,31,37 As with grades I and II, the preceding proposal may result in controversy, but the specialized bibliography recommends such treatment (Figs. 83-3 and 83-4).^{17,21–23,25,31–37,45}

Grade III AVM

1. Endovascular therapy + surgery

2. Surgery

3. Endovascular therapy + surgery + radiosurgery

4. Endovascular therapy + radiosurgery

5. Radiosurgery

FIGURE 83-3 Grade III AVM. Eloquent area, superficial location, and deep venous drainage.

FIGURE 83-4 Grade III AVM. Small size, deep localization, and deep venous drainage.

Grade IV AVMs

Grade IV AVMs that have not bled and do not have angiographic signs suggestive of complications (e.g., intranidal aneurysms) should be controlled with clinical and imagenologic evaluation.^19,22,33,34 The morbidity and mortality rates of the treatment are higher than those supplied by spontaneous course. If there was hemorrhage, these AVMs must be treated, since it is well known that if a lesion of such characteristics has bled and the patient survived, the AVMs will bleed again. In addition, it is highly likely that after hemorrhage the patient will present with neurologic sequelae, which diminishes the rate of sequelae occurrence due to treatment.

If these lesions are treated, the treatment should always be a combination of techniques, starting with endovascular therapy. The number of sessions depends on the characteristics of the lesion, the number of pedicles, the compromised vascular territories, and the results of every session of endovascular therapy. There are always at least two sessions. Once the endovascular treatment has been completed, a prudent amount of time must elapse; this period should be decided on by the multidisciplinary team in charge of the treatment, because surgery is the next step. A posteriori, and if the angiographic control shows any remnant, the treatment is completed with radiosurgery.

When these cerebral vascular malformations are treated (using a combination of options), the cure rate is close to 100%, but the possibility of sequelae rises to almost 50%.^{17,21,22,32–34}

Sometimes a palliative treatment will be proposed: endovascular surgery in association with radiosurgery or as a stand-alone option. With such treatment, the possibility of bleeding is reduced but the patient is not necessarily cured (Fig. 83-5).

Grade IV AVM

1. Do nothing

2. Endovascular therapy + surgery

3. Endovascular therapy + surgery + radiosurgery

4. Endovascular therapy + radiosurgery

5. Endovascular therapy

FIGURE 83-5 Grade IV AVM. A, Preembolization. B, Postembolization but before surgery.

Grade V AVMs

In principle, grade V AVMs are not treated.^17,19,37 Risks are high; cure is possible, but at the cost of sequelae. If they are aggressive lesions, in the sense of presenting recurrent bleeding, treatments must be tried. In my opinion, those treatments will always be palliative and will be dominated by endovascular surgery and the possibility of being completed with radiosurgery. Direct surgery should not be an option in malformations of this grade. Regardless, a grade V malformation can be operated on, but always with the prior endovascular therapy as an adjuvant. Surgery will always be difficult. The sequelae rate is high. In my opinion, surgery of grade V AVMs must be reserved exclusively for those cases in which there is recurrent bleeding and the patient, already with sequelae, has a life risk in case of new hemorrhage.^{17,19,33,35,45–48} Some authors propose radiosurgery as the first step of treatment and then endovascular therapy. In all cases, it will be a palliative treatment (Fig. 83-6).^{17,21,22,37,46–48}

Grade V AVM

1. Do nothing

2. Endovascular therapy + radiosurgery

3. Endovascular therapy

4. Endovascular therapy + surgery

5. Endovascular therapy + surgery + radiosurgery

6. Radiosurgery + endovascular therapy

FIGURE 83-6 Grade V AVM. A and B, Preembolization; C, postembolization before surgery.

A recent publication⁴⁹ studied the safety of surgery in the treatment of AVMs. The conclusion was that in grades I and II the surgical risk (morbidity) is under 1%, while in grade III to V AVMs it varies according to the involvement of eloquent areas; surgical morbidity and mortality range from 17% to 21%. The comments of Batjer and Hernesniemi on this study support surgery as the best option in the treatment of the AVMs.⁴⁹

The Surgeon

The specialist who is going to perform the surgical operation of an AVM must be a well-trained and experienced neurosurgeon. This surgeon must have training in microsurgery, which mean many hours of practice in a laboratory. The laboratory must be a continuous practice. Also, proper knowledge of anatomy is indispensable, with continual updates regarding cisterns, sulcus, circumvolutions, arteries, and veins.

Constantly updating knowledge of the brain’s anatomy, as well as its vascular tree, allows the surgeon to better interpret angiographic imaging, magnetic resonance imaging, and so on. The laboratory practice allows the surgeon to acquire more skill in the dissection of sulci and cisterns, indispensable points in any AVM surgery.

Preparation for Surgery

Unlike in other neurosurgical interventions, the preparation for AVM surgery requires the fulfillment of several indispensable steps that allow the surgical act to be performed with fewer risks. First, the surgical intervention of an AVM should always be scheduled. A patient with an AVM complicated by a hemorrhage may require an emergency operation—a case that supports such action. But the indicated action in these cases is to proceed exclusively to the hemorrhagic area evacuation and not seek the malformation in this urgent surgical act. A swollen brain, a distorted anatomy, vessels compressed by the hematoma, venous stasis, and associated ischemic phenomena due to several of these factors conspire in such conditions against the surgical attempt on the AVM, rendering it a surgical failure, not to mention affecting the vital and functional prognosis of the patient. The exception can lay in small malformations found in the wall of the hematoma. In such cases, the resection can be performed. Although a hemorrhagic lesion was evacuated and nothing was found on searching in its “walls,” the pathologist may later report that a small AVM was found among the clots (Fig. 83-7).

FIGURE 83-7 A, In an AVM treated 3 years prior solely with radiosurgery, hematoma is now seen (GCS 4). Angiography shows pre- and postembolization before surgery. Emergency surgery for evacuation of the hematoma is performed. B, The same case is shown on recovery after surgery (GCS 14) through magnetic resonance imaging. A, left: CT at admission; right: CT after emergency surgery. B, right: MRI postsurgery; middle: angiography preembolization; left: angiography postembolization.

Second, when a malformation is going to be operated on, the patient must be prepared for surgery. This implies complete studies and, if necessary, the performance of endovascular techniques. When occlusions of afferent pedicles and part of the malformation nidus have been carried out by endovascular means, surgery must be performed no fewer than 4 or 5 days after the treatment was carried out.^50,51 Apart from this requirement, the timing for the surgery depends on the characteristics of the AVM, its flow, its size, the afferents and efferents, and the grade of occlusion achieved. Most importantly, it depends on the patient’s clinical condition after the procedure. An asymptomatic patient can be operated on prematurely, but those who start to present a neurologic deficit and whose imaging studies reveal an ischemic lesion must be differed until clinical stabilization is achieved. Therefore, the surgical team, taking into consideration these factors, should determine the best moment for surgery in each case. The existence of an AVM provokes an alteration in the brain’s blood flow. The occlusion determines a new redistribution of the brain’s blood flow, regardless of the percentage occluded by endovascular means. Such modification, which has been progressive with the endovascular therapy, experiences an abrupt change with surgery. Thus, allowing time for the changes in blood flow to be induced by the therapy before surgery takes place reduces the possibility of associated morbidity.⁵¹

Third, every case must be discussed and surgery must be planned by the team that will work with the surgeon, that is, the assistant vascular surgeons, endovascular therapists, neuroanesthesiologists, neurologists, and neurophysiologists.

Operating Room

The surgery begins with the anesthetic act performed by a neuroanesthesiologist, who, knowledgeable about the pathology and well informed about the particular case, plans the appropriate anesthesia for that case. Due to the ongoing advances in anesthetic techniques and drugs, it is impossible to impose one anesthetic protocol. The use of a purely intravenous anesthesia or one in combination with inhalation agents depends on the particular case. It is necessary to count on a central venous line, as well as large peripheral venous catheters in case further access is needed. Monitoring the arterial pressure with an arterial line is indispensable, not to mention the classic monitoring of oxygen saturation, capnography, entropy, and so on. Next, the neurophysiologist places the electrodes for the neurophysiologic control of the cortical activity, peripheral motor response, cranial nerves, and so on. If necessary because of the lesion’s topography, it is possible to then proceed to localization by means of a stereotactic frame or to start with neuronavigation to localize lesions in real time. The next step is fixing and positioning the head by means of a cranial fixation headrest, according to the lesion topography.

An operating microscope is essential for surgery. Regarding the surgical instruments, leaving aside all necessary implements to perform the craniotomy, little is needed for the AVM resection:

Of these, the essential “tools” that surgeons use in most malformation dissections are the aspirator and the bipolar coagulator (Fig. 83-8).

FIGURE 83-8 Surgical instruments for AVM resection.

Surgical Technique

Superficial Malformations

Regardless of whether the AVM to be resected was previously treated, either by endovascular treatment or by radiosurgery, the technique does not vary. The skin flap, as well as the bone flap, is determined by the lesion characteristics. The surgeon must always be focused on the malformation. The position of the head follows the same criteria as those previously described. It must always be fixed with a cranial fixation headrest, and care must be taken not to lateralize or flex the head significantly so as not to compress the veins at the neck level, since this could provoke a limitation in the venous return. Such a limitation can lead to surgical complications, with a tense brain and venous pressure in a malformation, which increases the risks of intra- and postoperative complications.

In interhemispheric lesions, which represent a significant amount of the supratentorial malformations, it is possible to choose the half-sitting position, with the head held by a cranial fixation headrest. It is possible to place the head in a medial position or slightly laterally toward the side of the malformation. When adhesions and so on fixing the brain to the midline are released, this permits the brain to droop slightly (i.e., to separate spontaneously), enhancing—without the use of retractors—the interhemispheric sight.

In superficial (corticosubcortical) AVMs, a large flap must be performed centered on the malformation. Even in lesions only a few centimeters wide, the best option is for the flap to include not only the lesion but also the surrounding area, giving the surgeon a view of the normal surrounding brain and thus allowing identification of the sulcus, the different circumvolutions, and the eloquent areas. This also enables the surgeon to assess the normal and not-so-normal vessels that supply and leave the malformation. Thus, the surgeon ensures control of the afferent arterial vessels and of veins, as well as the capacity to identify the arterialized veins. Once structures have been visualized and recognized, the anatomy is compared with the imaging, which must be carefully studied before and during the surgical act (Fig. 83-9).

FIGURE 83-9 Big flap with the AVM in the center.

Before starting the resection, the neurophysiologist has to place electrodes for corticography with identification of areas in the zones surrounding the lesion, which allows the surgeon to work more comfortably.

A technical detail in the approach, both for superficial AVMs and for subcortical ones, is the opening of the dura mater, but with venous drainage toward the surface. There is always the risk of vessels, especially venous ones, that have adhered to the dura mater. The dura mater must be opened slowly and with the proper care; otherwise, a vessel may be torn in the process, with consequent bleeding that in cortical AVMs may correspond to a hemorrhage of the core of the lesion. Both in these AVMs and in those that drain to the surface, a vein can be torn—even the main vein exiting the AVM. Thus, a previously planned surgery of a priori low risk can turn into a catastrophe. Therefore, the opening of the dura mater must be performed slowly, with the surgeon continually checking for adhesions to release them with extreme care. The use of magnification is even necessary in many cases at this stage, by means of magnifiers or operating microscopes (preferably the latter), so that the surgeon can adapt to the surgical field under the microscope after opening of the dura mater (i.e., before seeing the malformation), which avoids distractions.

It is well known that a significant portion of cerebral AVMs are superficial but with an interhemispheric component. In these cases, the bone flap must surpass the midline, and the opening of the dura mater must expose the interhemispheric fissure to have at sight (i.e., to be able to manage) the longitudinal sinus. These maneuvers can take several minutes and even delay the resection of the nidus, but we must take into consideration that good exposure and good control over the structures later results in a less complicated malformation surgery.

In addition to the previously described care that must be used at the opening of the dura mater, if the sinus and the interhemispheric fissure have to be exposed, the dural opening must be done slowly and cautiously. In a brain with no vascular lesions, like the ones dealt with in this chapter, there are usually thick veins and adhesions in the vicinity of the sinus. When there is an AVM, such situations become more frequent; hence, the dissection and separation of veins and dural adhesions must be performed carefully.

Venous bleeding in the vicinity of the midline is not rare in such patients. Thus, it is recommended in these cases that the small dural vessels be coagulated and cut against the arachnoid and not against the dura. In case of tearing or bleeding of large-caliber vessels, they must not be coagulated, but a hemostasis must be performed by means of hemostatic materials and later compression with cottons. Trying to perform hemostasis with coagulation of large-caliber venous vessels is a technical mistake and will not solve the problem; on the contrary, the bleeding will increase and will lead the surgery to failure (Fig. 83-10).

FIGURE 83-10 Relationship between veins and dura mater.

Once the dura mater has been opened and the surgeon can see the malformation in the superficial lesions or its afferent vessels and draining veins—many of them arterialized in the subcortical lesions—it is necessary to identify the surrounding areas by means of direct cortical stimulation.

Afterward, and always with the corresponding magnification by means of an operating microscope, the area to be operated on is delimited. Arteries are studied as the surgeon tries to identify whether they are vessels heading to the malformation or en passage vessels; the process is repeated with veins. Usually, the veins that exit the nidus have an abnormal coloration, looking more like an artery, hence their designation as “arterialized veins” (Fig. 83-11).

FIGURE 83-11 Arterialized veins, indicating the approach to the AVM.

Such veins can be mistaken sometimes for arteries. These veins must be looked after; the surgeon must try neither to injure them nor to coagulate them until the final stages of the nidus resection. The main draining vein is the last to be eliminated. Its early elimination brings about an important increase in the pressure inside the nidus, with resultant bleeding that, in this kind of situation, becomes difficult to control. One way of knowing whether the resection is being performed correctly and the arterial pedicles are being eliminated (either in previously treated or in untreated malformations) is that, during resection and coagulation of the arterial pedicles, veins change their tension and coloration, becoming darker. It is often difficult—even for an experienced surgeon—to identify which arteries correspond to the malformation and which do not.

The resection of a malformation is always performed from the outside inward. The nidus is surrounded but never entered. The opening of the arachnoid is started, usually from the sulcus delimiting the lesion. This maneuver is continued until the whole AVM is practically surrounded. In general, in superficial malformations, this maneuver stops when the thick draining vein is reached. Like most superficial malformations, the corticosubcortical ones are roughly conical with the apex at the deepest point, and the dissection follows this contour. It could be said that it is being surrounded and deepened step by step.

An essential step is to identify the plane between the nidus and the surrounding tissue. Because of the chronic ischemia that determines the malformation on the neighboring parenchyma, there is always at least a small layer of nonfunctioning tissue (gliotic or cicatricial) that surrounds the lesion. This tissue is the margin within which the surgeon has to work without provoking lesions for compression and/or coagulation. In each step, it is important to seek the pedicles and coagulate them. The thickest can be coagulated or clipped. The use of clips is reserved for thick vessels, where coagulation or placement of hemoclips is controversial regarding their efficiency. If they are used, they must always be small or miniature clips. The placement of transitory clips is reserved for those vessels that the surgeon wants to preserve, because it is uncertain whether they are en passage arteries, go to the malformation, or supply the nidus. However, clipping distal to the malformation is preferred to enable the surgeon to manage it better, identify the branches that come to the nidus, and eliminate them. The new clips for AVMs are useful at this point in the surgery.

The aspirator, resting on cotton, separates the nidus, and the bipolar coagulator performs the double function of dissection and coagulation. This must always be performed following a circumferential pattern. Every section left behind is covered with cotton, which allows not only hemostasis but also separation.

As this type of malformation is being deepened, the vessels become smaller. Close to the apex of the lesion, it becomes increasingly difficult to coagulate them since they are thin walled and they recoil from this maneuver. In such cases, it is recommended to not insist on coagulation but to persist with the placement of hemostatic materials covered with cotton. Only in exceptional cases do these maneuvers not lead to hemostasis. Usually, during the resection of a malformation, it is not considered good practice to reach the subcortical region or the apex of the malformation to start coagulating all small vessels of altered walls that appear in the white matter. This would only lead to increased bleeding and to entry of healthy zones of the brain that can thus be damaged.

During a dissection of AVMs performed in a circumferential pattern, it is possible to mistakenly enter the nidus, often because we have not been able to identify its boundaries. Such entry is manifested through bleeding, often profuse. In such cases, it is recommended to apply hemostatic material, cover the area with cottons, and compress with the second aspirator. This area should be left alone, and the plane in another zone of the lesion must be sought to continue the circumferential dissection step by step. As the surgeon comes back to the zone that caused bleeding problems, it is necessary to seek the plane from the outer part of the area where work had been undertaken. Once it is found, surgery proceeds in the customary way. Any time bleeding is difficult to stop, it is because we are inside the nidus and not in the plane between the nidus and the surrounding brain.

By the end of the resection, the surgeon may be weary, especially if dealing with large lesions. In such cases, more care must be taken. Sometimes weariness leads to incorrect maneuvers, such as thinking the whole nidus has been surrounded and there are no pedicles left in the depth and then proceeding to try to withdraw it, which can lead to arterial or venous tearing. From the beginning to the end of the resection, the same level of attention must be paid, and a meticulous inspection of the bed must be performed before withdrawing the nidus.

As soon as the dissection of the whole AVM is complete, the coagulation and the cutting of the main draining vein are undertaken. Only then can the nidus be removed. Often, the bed where the lesion lies presents small and multiple bleedings. In such a case, placing hemostatic material and covering it with cottons is the indicated procedure. Later, the surgeons wash with saline and wait for spontaneous hemostasis to take place.

Occasionally, there is a zone of major bleeding that cannot be controlled with these maneuvers. This usually indicates that remnants of the malformation were not removed. In such cases, and by means of higher magnification, the zone must be explored. This small, remaining nest must be surrounded and then resected in the same way as with the main nidus.^{6,29,37,52–56}

Deep-Seated Malformations

As their name indicates, deep-seated lesions are not visualized in the cortical inspection. At most, an arterialized vein can be seen “going out” of a sulcus, which would be a guide for dissection. But in most cases, and following the criterion that surgically treated deep-seated AVMs are usually small, help is needed to know the precise location of such lesions and their corresponding resection.

It is necessary to differentiate those malformations that bled from those that did not. If there are deep hematomas (e.g., in the basal ganglia or the insular region), a high percentage of these patients are already presenting a neurologic deficit determined by this bleeding. In such cases, the surgeon works in a relatively secure way, since the hematoma dramatically diminishes the possibility of adding morbidity. But malformations that did not bleed pose the problem that, due to their topography, will always be in an eloquent area.

Whatever the therapeutic maneuver, it entails life risk and especially a considerable morbidity risk. Therefore, in those lesions that did not bleed, abstention is probably the best choice. However, it is well known that deep-seated lesions have higher possibilities of bleeding than those located in other areas of the brain. Morbidity of these lesions has diminished significantly in recent years,^18,23,24,37 basically because of advances in neuroanesthesia, the practice of neurosurgeons in microsurgery of sulcus and cisterns, and the strict knowledge of gyri (Fig. 83-12).

FIGURE 83-12 A, Deep grade III AVM preembolization. B, The same case after embolization and surgery.

Following Spetzler and Martin’s classification, deep-seated AVMs with indication for surgery would be grades II and III. Rarely would surgery be performed on a grade IV AVM, not to mention a deep-seated grade V AVM. In such cases, both surgical morbidity and surgical mortality are very high, even higher than those of the natural evolution of the disease. But grade II and III lesions that undergo surgery would be, in most cases, of small nidus, with venous drainage to the depth exclusively or associated superficial drainage.

This subgroup of malformations poses an important challenge to neurosurgeons, and every case raises a dilemma about how to reach the final and curative solution. Prior to describing surgery of deep-seated AVMs, I must mention the usual topographies of this group of lesions. At the supratentorial level, they are localized in the region of the basal ganglia and the insula, periventricular and intraventricular. At an infratentorial level, it is possible to find them along the brain stem, in the IV ventricle, or deep at the level of the vermis or cerebellar hemispheres.

Solomon and Stein⁵⁷ divide these malformations into three groups: the basal ganglia ones, especially the thalamocaudate; those of the brain stem; and those neighboring the tentorium. Among the first group, he includes those neighboring the atrium of the ventricle, and among the last group are those placed at the quadrigeminal cistern, mesencephalon, and cerebellar vermis.

Surgery

When lesions of small nidus are considered, which are generally irrigated by arteries of small caliber, it is possible to proceed directly to surgery, However, in those whose pedicles are thick and the nidus is not so small, preoperative endovascular treatment will be of great value to the surgeon. Depending on the lesion’s topography, it is necessary to seek the proper approach, always following the criterion for dissection by cisterns, sulcus, or fissures.

For lesions of the hippocampus or of the insula, the transcisternal approach is ideal. In intraventricular or periventricular lesions, the interhemispheric callosal approach can be the best choice.

For external lesions of the basal ganglia, it is necessary to seek some sulcus that, due to its proximity, leads the surgeon to the lesion. It is always necessary to avoid injuring the parenchyma, so correct dissection of a cistern or sulcus permits the surgeon to make enough room to reach the malformation, with minimal or no brain compression. In this case, the head must be placed laterally toward the side opposite the lesion, always avoiding compressions of the neck. It is fixed with a cranial headrest (Fig. 83-13).

FIGURE 83-13 Deep AVM. Surgery by a sulcus.

In deep-seated lesions of the basal ganglia, centered on the thalamus or the so-called thalamocaudate, an interhemispheric transcallous approach might be the most indicated. In these cases, the patient can be placed in a half-sitting position, with the head fixed with a cranial fixation headrest. The same position can be adopted for the intraventricular or subependymal lesions. This position has my preference. However, the patient could also be placed in lateral decubitus.

For those lesions neighboring the tentorium, it is possible to place the patient in lateral decubitus and use a subtemporal approach or to place the patient in a half-sitting position and use a supracerebellar, supratentorial, or infratentorial approach, depending on the lesion and its topography.

In lateral and subtemporal approaches, it is necessary to use special care in identifying and preserving the Labbé vein, independently of whether it is linked to the malformation, since the function of this vein in venous drainage is of the utmost importance, and damaging it not only can provoke swelling of the neighboring brain but also can lead to lesions by venous ischemia that endanger the patient’s life or determine disabling sequelae.

For the interhemispheric approaches and the tentorial ones, it is possible to section the fold of the dura mater, either the falx cerebri or the tentorium, to have better control over the afferent branches, as well as the draining veins. Until recently, to perform such approaches, it was necessary to drain cerebrospinal fluid to permit a better retraction of the structures. At present, with neuroanesthesia performed in good conditions by a well-trained neuroanesthesiologist, a compliant brain is achieved without this maneuver. In subtemporal approaches, as well as in those of the posterior fossa, the liquid of the cistern can be extracted, which allows the surgeon to work without compression of the structures.

In malformations of the cerebellum, my preference is the half-sitting position with a cranial fixation headrest. This position requires correct monitoring by means of thoracic Doppler to search for embolisms and act in consequence. Other authors prefer to position the patient in ventral decubitus. Though this position is safer as concerns avoiding embolisms, there is a higher accumulation of blood in the bed as the surgeon goes forward through the different planes. Regardless, well-trained surgeons should never find difficulties in this position. I consider the ideal position to be the one to which the surgeon is accustomed and for which it is most comfortable for that surgeon to operate (Fig. 83-14).

FIGURE 83-14 Half-sitting position (A) and park bench position (B), both for an interhemispheric approach.

In no case do I consider transcortical approaches proper. Anatomy should be preserved; the natural pathways that cisterns, fissures, and sulcus offer permit the surgeon to properly approach all brain sectors.

For these surgeries, the assistance of equipment that permits to the surgeon determine the topography of the lesion and to search (as explained earlier) for the sulcus or sector of the cistern or fissure through which to approach the malformation is indispensable.

At a minimum, localization by means of skull computed tomography–guided marking should be performed. This can be performed in the immediate preoperative stage or in those health centers with operating rooms equipped with tomography devices during operations. In such cases, it is possible to perform marking first and then, as dissection is started, to check whether the direction is correct.

The assistance with marking by means of stereotaxy is of much value. Surgery starts with the placement of the stereotactic frame. The corresponding localization is performed, and once the dura mater is opened, the surgeon checks the entry point and the direction with the help of the frame needle. This is highly valued help in the surgery of deep-seated AVMs. It has the advantage of permitting the surgeon to confirm the approach site, relate it to some sulcus or sector of the cistern or fissure, and know the depth at which the lesion should be.

Neuronavigation techniques, either with equipment that permits surgical team to locate and assist in real time or with the latest with Doppler technology, are of great value, not only to indicate where the lesion is but also to assist the surgeon in determining whether the resection was complete. Magnification by means of an operating microscope is also useful.

Finally, for this type of lesion, the assistance of an endoscope can be of great help. Its use may be necessary at different stages of the dissection. For instance, for intraventricular lesions or those in the ventricular wall, once the sinus of the ventricle is approached, direct visualization by means of an endoscope can help in locating the lesion. It is even possible to use it as an assistant, both for lighting and for magnification, during resection of AVMs. In other cases, after the surgical dissection has been completed under microscope—especially in small and deep-seated lesions—the inspection of the bed through the sulcus and cistern by means of a neuroendoscope can rule out the existence of any remaining malformation and help check the correct hemostasis.

Unlike what has been explained regarding superficial lesions (where wide flaps are my preference), in these cases and with corresponding assistance by means of the aforementioned equipment, small lesion–centered flaps can be performed.

Malformation dissections do not significantly differ from the previously described superficial lesion dissections. Unlike those dissections, malformation dissections do not have a conical aspect; rather, they stretch on a plane like a jellyfish. A direct dissection of the nidus should never be performed. It must be surrounded, while searching for pedicles that reach the lesion. Small afferents predominate, which often contrasts with a thick draining vein, especially in those of high flow.

The bipolar coagulation and the aspirator are again the surgeon’s essential tools. The bed that is released must be covered with small cottons. The nidus is separated with the aspirator but always covered with cotton. Small, thin-walled vessels in the bed of the lesion are frequent. The hemostatic materials are of great help. It is necessary to avoid indiscriminate coagulation of the bed and of these small vessels, which, as has been described concerning the superficial ones, recoil to coagulation. The difference is the depth at which the surgeon is working and the small surgical field. Thus, it is necessary to be careful when coagulating and cutting the afferents that are clearly visualized; in the remaining ones, the surgeon must perform “indirect” hemostasis, that is, with hemostatic materials. The main draining vein is always left for the end. Once the lesion is dissected and liberated of all afferents, the surgeon proceeds to coagulation and cutting of the draining vein. Later, the bed is explored, either under microscope magnification or by means of an endoscope.

Eventually, the walls of the sulcus and/or of the cistern (through which the approach was performed) are covered with hemostatic materials to avoid bleeding of these walls, usually rich in vessels and congestive since it presents near a malformation and due to the neurosurgeon’s work.^{3,18,24,52,53,55–62}

Endovascular Surgery

For the treatment of an AVM, endovascular surgery, endovascular therapy, or neurointerventionism, a well-trained specialist in the management of such lesions is required. Before the treatment, correct planning of the surgery must be carried out by means of a throughout analysis of the imaging, widely dominated by the digital angiography of brain vessels. At present, it is also fundamental to have functional magnetic resonance, which permits the surgeon to delimit the areas of work.^9,11,12,14

This method seeks the greatest occlusion of the nidus, as well as of its afferents and of the origin of the draining vein, without causing damage to the patient. Many of those who work in the area of endovascular surgery ponder the possibility of reaching complete endovascular occlusion of an AVM with this method. The permanent and complete occlusion of the nidus is posed.^63,64 I think that sometimes it is possible to achieve the complete radiologic occlusion, but this does not mean the complete disappearance of the AVM. In those cases in which complete closure of the lesion is ensured—and that later undergoes surgery—there may be small vessels that did not refill in the endovascular procedure and that eventually would make the lesion reappear if surgery was not performed. In the cases of apparent complete endovascular occlusion with no surgery, it occasionally happens that in successive angiographic controls, which can be performed after 3, 6, or 9 months—or even later—pathologic vessels and draining veins reappear.⁴⁴ This is controversial, because endovascular surgeons consider complete and definitive occlusion to be achievable, especially in small AVMs and when Onyx is used.^{39,40,43,44,64–66} In recent publications, Weber et al.^39,66 demonstrated that Onyx penetrates small vessels, achieving occlusion. Even as these authors stated that complete angiographic occlusion can be achieved, they remarked that some cases of recanalization have been found in long-term controls. However, in dural fistulae, complete occlusion was reported in another study, and at 5-year follow-up no recanalizations were found.⁶⁷

Weber et al.³⁹ referred to the benefit of treatment with Onyx, achieving high occlusion rates, especially in cortical AVMs. Van Rooij et al.,⁴⁰ in a 5-year treatment series, referred to complete occlusion with Onyx in small AVMs. The authors recommended that several years of use of this method pass before the endovascular therapists’ opinion is accepted as true. Up to now, there have been no papers with enough evidence to sustain that position.

Every case is different. It is necessary to have a complete notion of the angioarchitecture of the malformation, its flow; the pedicles that come to an end in the AVM; those that go to it, give branches to the lesion, and continue to the normal parenchyma; and the branches that are exclusively en passage. Special care must be taken by the endovascular neurosurgeon in ruling out possible intranidal aneurysms, since their occlusion is fundamental to diminishing the risks of bleeding. Another element to assess in the analysis of the angioarchitecture of an AVM is the presence of intranidal fistulas, since their elimination requires a special treatment.

In large lesions, neuromonitoring is fundamental, as well as the arterial occlusion tests. When the interventionist knows that navigating up to the origin of the nidus will not be possible, a test to assess the possible occlusion of proximal arterial branches or the stem, itself an important artery, is necessary. By means of administration of amobarbital sodium, Wada’s test can be performed in arteries distant from the nidus, as well as in the distal vessels against the nidus. This is a highly valued test to determine the tolerance of a certain vessel to occlusion.^6,65

The particles and histoacrylic glue were long the materials of choice for closure of the vessels of a malformation. Later, surgeons began to use some coils in some cases. Finally, the appearance of Onyx has been of great value in the treatment of such lesions. This material is administered in a liquid form that hardens when it comes into contact with blood, with scant risk of migration. At present, histoacrylic glue (n-butyl cyanoacrylate) and particularly Onyx are the materials of choice for the endovascular treatment of AVMs.^{39,40,43,44,65,66,68–71}

The treatment must be always performed under general anesthesia. This allows the endovascular surgeon to work more comfortably and, if complications occur, enables the anesthesiologist to act rapidly, correcting parameters, improving oxygenation, controlling the capnography, provoking arterial hypertension if necessary, enabling delivery of volume and inotropes, and so on.

The femoral artery is the chosen route for catheterization. If difficulties or anatomic problems arise that prevent its use, it is possible to catheterize through the humeral artery. At present, direct carotid puncture is rare.

The procedure begins with the placement of a guiding catheter up to the carotid artery or the vertebral artery. Navigation continues by means of a microcatheter up to the nidus or its vicinity. The farther the microcatheter can be introduced into the nidus or into the arterial branches that feed it, the better the results. It is always necessary to try injection of the material to occlude (histoacrylic glue or Onyx) inside the nidus or in the clearly identified afferents that end in it. However, it may not be possible for the injection to reach the nidus due to difficulties with the catheterization, especially when afferent vessels are thin and tortuous. In such cases, those can be occluded after the patient has undergone a tolerance test.

If the injection is intranidal, there is practically no risk of migration of the embolization material. However, there is the slight possibility of such migration when the occlusion occurs in the vessels outside the nidus. It is important to identify the AVM flow and to plan the amount and intensity of the chosen embolic material to be injected so that it reaches as far as possible into the nidus and even to the draining vein or veins.

The percentage of the AVM to be occluded depends on its size, the flow, the afferents, and the functional status of the surrounding brain. In small malformations, it is possible to perform the total planned occlusion in just one session, but in larger malformations (grade III or higher), treatments must take at least two sessions. In these cases, it is important not to cause an inversion of the flow due to an abrupt, massive occlusion, since this may provoke brain lesions, along with hemorrhages, ischemia, and intracranial hypertension, all of which can lead to sequelae—often permanent—and in some cases to the patient’s death.

Never to proceed to surgery immediately after embolization; rather, allow time for the flow redistribution to be determined by the endovascular therapy. In this period, the patient can also recover from neurologic deficits that may have appeared due to the treatment. Even though endovascular therapies conducted by a well-trained surgeon usually exhibit low morbidity, if morbidity takes place, the surgeon can let several days or weeks to elapse before advising surgery. On the opposite end of the spectrum, there is the risk of adding morbidity to the surgery (Figs. 83-15 and 83-16).^72,73

FIGURE 83-15 A, Interhemispheric AVM before treatment. B, The same case after endovascular therapy. C, The same case with an interhemispheric surgical approach. D, The same case shown on a computed tomography scan after surgery.

FIGURE 83-16 A, Deep AVM before endovascular therapy. B, The same case postembolization but before surgery.

Radiosurgery

The therapeutic procedure of radiosurgery has been routinely used for the treatment of AVMs for more than 30 years. Technically, it consists of the administration of a high-radiation dose in only one session, locating the point of application where the lesion is concentrated with scant or no diffusion to the surrounding areas.

Radiosurgery has a widely spread use in neurosurgery, and the AVM is one of the pathologies to which it is frequently applied. In this application, the nidal thrombosis is sought. This is achieved through progressive hyperplasia of the tunica intima up to occlusion of the vessels. Although it is usually well tolerated, positive results are never immediate.^74,75 On average, 2 years must elapse until complete sclerosis; thus, the AVM occlusion can be verified. Therefore, during the degenerative process that leads to closure of the nidus, the risk of bleeding persists. This is why radiosurgical therapy has some precautions. Such therapy is clearly indicated for some cases, whereas for other cases other options should be discussed; it is mostly indicated for AVMs that did not bleed and that have a volume below 8 cc. If we take the AVM as the integral dose (dose/volume relationship, expressed in millijoules) it would be 170 mJ. This is comparable with the 8-cc volume.¹⁷ If the patient presented with bleeding due to an AVM, it is best to wait no less than 6 months before proceeding to treatment with radiation.

Deep-seated lesions, and especially those that compromise the brain stem, have a priori a higher indication for radiosurgery. In these cases, attention must be paid so that the lesion does not compromise more than 30% of the brain stem. Special care also must be taken when the AVM in question is near structures that can suffer the effects of the treatment, such as a malformation close to the optical chiasm.

The treatment must be planned by a well-trained physicist in these procedures. It is necessary to correctly define the target and the optimal dose to apply in each case. For planning, angioresonance, angiotomography, or angiography can be used, but planning always concludes with stereotaxy. Hence, it is preferable to call it “stereotactic radiosurgery.”

Recently published works refer to the benefits of radiosurgery as the only treatment for some AVMs but also refer to its value as an adjuvant to the surgery. It has been proposed that a malformation subjected to stereotactic radiosurgery can have better surgical management and fewer risks of intraoperative blood loss. However, it is also accepted that AVMs that underwent stereotactic radiosurgery but 2 years later have not yet disappeared should be operated on if are accessible to the surgeon. In all cases, when surgery has been performed after the radiant treatment, morbidity has been low.⁷⁴

In 2008, a scale for radiosurgery and AVMs was proposed, with the aim of predicting the prognosis after the radiant treatment.⁷⁵ It needs to be used for some years before its value can be assessed.^76–89