CHAPTER 388 Microsurgical Management of Arteriovenous Malformations
This chapter is intended to address aspects that are found to be most relevant when attempting to treat cerebral arteriovenous malformations (AVMs) microsurgically. Although other chapters in this volume are devoted to nonmicrosurgical AVM treatment modalities, in this chapter we provide a brief discussion of the aspects that we have found most important in the treatment of AVMs by embolization and radiosurgery. We know that different treatment modalities for AVMs are frequently presented as “alternatives” to the patient, but we try to make the point that for each patient there is only one “best” alternative, although that alternative may include more than one modality (e.g., embolization followed by microsurgery). The different modalities of treatment and treatment paradigms should not be presented as interchangeable.
In this chapter we concentrate on decision making, as well as on the operative strategies and technical considerations that the senior author (R.C.H.) has found to be most useful.
General Considerations
Although much information on etiology, pathophysiology, natural history, and clinical findings is covered in other chapters, a brief review of this topic, from our perspective, seems necessary before directly discussing microsurgical treatment.
In principle, AVMs are arteriovenous shunts composed of feeding arteries that connect into an abnormal vascular nest (nidus), which in turn is drained by abnormal veins. Consequently, there is no or only an abnormal capillary bed. The so-called nidus is the network of channels that is interposed between the feeding arteries and draining veins. A compact nidus is characterized by a well-circumscribed vascular network, whereas a diffuse nidus refers to an abnormal vascular network that is more widespread within normal brain parenchyma.1 In both cases the abnormal vessel conglomerate is intermingled with gliotic brain tissue. This pathology needs to be differentiated from a dural AVM, or fistula, which has no nidus but a more direct communication between enlarged arteries and veins. AVMs are thought to be congenital in nature, but they frequently grow during childhood, adolescence, and young adulthood. It is rare for them to grow significantly in more mature adults. Therefore, it is important to stress that these lesions are not neoplastic and thus the term “angioma” should be avoided because it is inappropriate. Yet it is becoming more and more evident that these lesions also exhibit dynamic features that make them change over time, a fact that could influence even “successfully treated” AVMs.
It is obvious that a good understanding of the natural history of a disease is essential for the process of decision making if one is to weigh the risks and potential benefits of different treatment modalities against the risk of doing nothing. Fortunately, we already have several very good studies in the literature that give us robust data about the natural history of both AVMs that are initially manifested as hemorrhage and those that have never bled.
Table 388-E1 
isa compilation of the natural history studies that we consider most important.2–7 In summary, patients with cerebral AVMs bleed at a rate of approximately 2% to 4% per year. For decision making and timing of potential procedures, it is crucial to understand that this rate of bleeding is similar in most studies regardless of whether the patient has ever bled, although patients initially seen with hemorrhage have about twice the risk for rebleeding during the first year. It appears clear, then, that unruptured cerebral AVMs have a much greater annual rate of bleeding than unruptured aneurysms do, although of course the consequences of hemorrhage from an aneurysm are worse than those of hemorrhage from an AVM; AVM hemorrhages result in significant morbidity in about 25% to 30% of patients and death in approximately 10%.8
Spontaneous obliteration of an AVM is rare, but does occur, and has been reported in single cases.9–13 Several factors appear to be associated with spontaneous occlusion: a single draining vein, a solitary arterial feeder, and nidus size smaller than 3 cm.14,15
In the process of decision making, it is important to try to understand what factors may influence the rate of bleeding, which undoubtedly is not identical for every patient with an AVM. Several anatomic and hemodynamic factors seem to increase the risk for bleeding. Apart from associated aneurysms on the feeding artery (see later), the significance of nidus size has engendered controversy, with some suggesting that small AVMs bleed more frequently.16,17 Deep location has been suggested to increase the risk for hemorrhage (basal ganglia, periventricular or intraventricular space, posterior fossa),18,19 as has deep venous drainage,20 feeding by perforators, and location in the vertebrobasilar system.21 We believe that an intraventricular or periventricular location has a higher risk for hemorrhage, although we cannot substantiate this opinion with numbers. The role of feeding artery and draining vein pressure in this regard has been investigated. Norbash and coworkers saw no correlation in a group of 32 patients.22 In a large series of 340 reviewed patients, Duong and colleagues identified “high arterial input pressure” and “venous outflow restriction (exclusively deep venous drainage)” as the most powerful risk predictors for a hemorrhagic AVM manifestation.23 We also believe that venous outlet obstruction potentiates the risk for hemorrhage. Spetzler and colleagues associated AVM size smaller than 3 cm with higher feeding artery pressure and larger hemorrhage.16 Graf and collaborators calculated a 5-year risk for hemorrhage of 10% for AVMs larger than 3 cm and 52% for AVMs smaller than 3 cm.2 Whether smaller size really results in a higher risk for bleeding is still subject to controversy.
Pregnancy does not seem to increase the likelihood of hemorrhage.24,25 Horton and colleagues reviewed the cases of 451 women with AVMs who had 540 pregnancies, 17 of which were complicated by hemorrhage. None of the hemorrhages occurred during labor, vaginal delivery, or cesarean section. The authors concluded that the method of delivery should be based on obstetric consideration alone and the decision to operate after intracranial hemorrhage “should be based on neurosurgical principles.”25
Because of their size, giant AVMs differ somewhat in their typical findings from smaller lesions. Very large malformations are frequently manifested not only as hemorrhage or seizures (or both) but also as transient and progressive neurological dysfunction because of steal phenomena; the large shunt volume can initiate relative hypoperfusion in the surrounding functional parenchyma that sometimes results in ischemia.
Microsurgical Treatment
Indications for Microsurgical Treatment
Patient-Related Factors
When evaluating a patient for treatment of an AVM, we must consider factors such as age, general health, neurological status, history of recent or past hemorrhage, symptoms from the AVM, occupation and hobbies, and psychological makeup. The patient’s age is most important in determining the cumulative risk for AVM rupture during the remainder of the patient’s life expectancy. Assuming an annual hemorrhage rate of 2% to 4% and an average life expectancy of 70 years, the cumulative risk (in percentage) for AVM rupture may be estimated by the following formula: 105 minus the patient’s age in years.26,27 Younger age may thus justify a more aggressive approach because the cumulative risk is so high. At the same time, of course, it is easier for the young to tolerate a prolonged operation, and the chances for satisfactory recovery from any neurological deficit that might occur are better. Likewise, the general health of the patient needs to be taken into account. The current overall and neurological condition of a patient may dictate the timing, and severe comorbid conditions may preclude surgery. The occupation and lifestyle of a patient are also important aspects to consider. For a schoolteacher or lawyer it may be intolerable to have a speech deficit, but a mild visual field deficit might not hinder them from practicing their profession, whereas the latter would be intolerable for a pilot or truck driver. Some patients simply cannot live with the thought of harboring a brain lesion that has a relatively high potential to bleed at some point, whereas others live perfectly well with this knowledge and might prefer to decline treatment.
Arteriovenous Malformation–Related Factors
Location, size, configuration (compact versus diffuse), location and pattern of arterial supply, presence of deep perforators, location and pattern of venous drainage, evidence of outflow obstruction, associated aneurysms, blood flow to the AVM, evidence of “steal,” and fresh or old hematoma from the AVM are all factors that must be considered in determining the feasibility of surgery and the involved surgical risk.28 To help in this process, several classifications have been developed,29 including the one that is used most frequently today—the Spetzler-Martin grading scheme.30 Although such a classification is very helpful, especially in terms of reporting and comparing results, no classification could possibly take all the necessary variables into account that an experienced surgeon should consider when estimating surgical risk. Such factors as the presence of deep perforator arterial supply,31 location of the venous drainage, nidus configuration, and others would be difficult to account for in a classification that is still of practical use.32 Other classifications that have attempted to specifically address the risk for hemorrhage from an AVM (in contrast to surgical risk) never gained popularity in daily use (e.g., that of Nataf and coworkers20).
Surgeon-Related Factors
Finally, experience, availability of embolization and radiosurgery, familiarity with the results and morbidity of each treatment modality, supporting facilities, appropriate intensive care unit, anesthesia, intraoperative angiography, and the availability of experienced surgical assistance must be considered.
As will be emphasized, a good understanding by the surgeon of how to select the most appropriate treatment for the patient seems to be as important as mere technical skills at surgery. Besides considerable experience with all the surgical aspects, good understanding of the indications, efficacy, and complications of other treatment modalities is essential. We have emphasized repeatedly that it is the ethical responsibility of the concerned surgeon to give an answer-seeking patient a clear opinion of what in his or her view would be the best treatment of the patient’s AVM. This means that after weighing the pros and cons of the different modalities and their combination, the surgeon should supply the patient with one straightforward and understandable treatment recommendation. If not possible, it is prudent to refer the patient to a more experienced colleague. Of course, it is always the patient’s prerogative to accept or reject the recommendation, but an expert recommendation should never have the form of a potpourri of treatment modalities from which to choose. To this effect, the microsurgeon must have a thorough understanding of other treatment modalities that may be used in preference to or in combination with microsurgery. These other treatment modalities (embolization and radiosurgery) are covered in detail in other chapters but are discussed here from the perspective of microsurgery.
Embolization
Embolization is used with curative intention, as a palliative maneuver, or before surgical excision33,34 or radiosurgery. Partial embolization has its indications for palliation and presurgical treatment, but even in experienced hands it carries substantial risk. As can be seen from the review in
Table 388-E2,35–53 embolization in all its forms has morbidity and mortality rates that are not negligible, with morbidity ranging from 2% to 27% and mortality from 0% to 8%. Preoperative embolization has made it possible to operate on cerebral AVMs that before this technique became available could not undergo surgery without risking substantial morbidity. Frequent use has also been made of pre-radiosurgical embolization. The rationale to “downsize” large inoperable AVMs in an attempt to reach a so-called critical size for effective Gamma Knife surgery is very weak in our opinion. In many of these cases radiosurgery fails to completely obliterate the nidus, and in fact it has been shown to reduce the obliteration rate of linear accelerator radiosurgical treatment.54 Previous embolization had been identified as a negative predictor of successful AVM radiosurgery.55 After it became evident that particle embolization before radiosurgery resulted in a 15% to 20% recanalization rate, it was generally abandoned and liquid embolic agents are now preferred.55–57 However, with these agents the problem is inadvertent embolization of the feeding artery, thus resulting in proximal occlusion with only minimal or no nidal penetration.33 Inadvertent, premature closure of draining veins is a specific significant risk with all liquid agents. With a partially “open” nidus, creation of new arteriovenous shunts can be anticipated. In addition, there is no evidence that pre-radiosurgical embolization reduces the risk for hemorrhage during the critical period after radiosurgery until the induced devascularization and fibrosis lead to complete obliteration of the nidus.33 This is why in general we see only rare indications for preradiosurgical embolization.
Complete cure of AVMs with embolization alone is usually very difficult to achieve, although in very experienced hands success rates have reached 40% 
(Table388-E2). Complete cure in this context should be defined as disappearance of the nidus and early venous drainage. Generally, only relatively “simple” AVMs with readily accessible feeders are potential candidates for a curative endovascular approach. However, in many instances these AVMs are also the ones that are easy to remove surgically with minimal morbidity but with a higher certainty of complete resection. Therefore, we would recommend attempts at endovascular curative obliteration only for lesions that in addition to accessible feeders have a deep, critical location, where the risk associated with surgery would exceed the risk related to embolization, or in patients with advanced age or significant comorbid conditions.
In general, we believe that there are very limited indications for palliative embolization because the procedure changes flow dynamics and the actual bleeding risk can be increased with this form of partial treatment. The best indications for palliative embolization are “steal” symptoms (progressive neurological deficits) and chronic venous hypertension as a result of high-flow lesions with limited drainage capacity.42,58,59 Other indications can be associated aneurysms and large inoperable AVMs with inherent angiographic “risk factors” that could be addressed endovascularly. Patients with “inoperable” AVMs sometimes undergo palliative embolization after they have already experienced multiple hemorrhages to reduce the flow in an attempt to lower the risk for further rupture. Thus far, however, there is no evidence to support this empirical method. Other indications are intractable headaches and seizures.
Probably the most important role for embolization is preoperative reduction of arterial inflow, occlusion of deep feeders (Fig. 388-1), and management of prenidal aneurysms. The goal is to enable or facilitate surgery. In our opinion, preoperative embolization should only be performed when in the estimate of an experienced neurosurgeon and endovascular surgeon it will reduce the overall risk—in other words, when it can be estimated that the risk of preoperative embolization added to the risk of surgical excision after embolization would be less than the risk of surgical excision without embolization. Additionally, preoperative embolization should be guided by the surgeon who intends to excise the AVM and knows which feeding pedicles are easy to access at surgery and therefore need not be embolized and which are the deep inaccessible feeders that would be helpful to have occluded by embolization before surgery.
FIGURE 388-1 Small inferolateral thalamic arteriovenous malformation. A, Axial magnetic resonance image. B, Anteroposterior (AP) vertebral injection showing a large thalamic perforator. C, AP vertebral injection after successful embolization of the thalamic perforator. D, Intraoperative photograph of the subtemporal approach to the malformation, which was successfully resected after embolization of the large single thalamic perforator.
Radiosurgery
Radiosurgery was developed and to date is mostly applied by neurosurgeons trained in this treatment modality. Its impact on the treatment of AVMs has been huge. In all its different techniques it is still most effective for smaller AVMs that are 3 cm or less in diameter. Obviously, the great advantages of radiosurgery are its minimal invasiveness and its potential to treat lesions that would otherwise not be amenable to surgical excision because of deep or critical location, or both.60–64 Other indications are patients who as a result of age and general condition are not considered good candidates for surgery. Achievable obliteration rates are related to size and configuration (diffuse versus compact) and range between 60% and 80% over a 2- to 3-year period (Table 388-E3)
.44,55,65–73 Obliteration rates are clearly higher for smaller AVMs.70
TABLE 388-E3 Outcomes of Arteriovenous Malformations after Radiosurgery
| SERIES | OBLITERATION RATES | MORBIDITY/MORTALITY RATES |
|---|---|---|
| Steinberg et al.,65 1991 | 100% (<4 cc); 70% (>3.7 cm in diameter) by angiography | 0.9% permanent morbidity; 11% hemorrhage |
| Friedman,66 1997* | 79% (<10 cc); 47% (>10 cc) | N/A |
| Pollock et al.,55 1998 | 61% overall; 83% (<4 cc) by angiography | N/A |
| Nozaki et al.,44 2000 | N/A | 36% morbidity from rehemorrhage; 0% mortality |
| Stieg et al.,67 2000* | 76% at 3 yr | N/A |
| Inoue and Ohye,68 2002 | 81.3% by angiography | N/A |
| Pollock et al.,69 2003 | 78.0% by angiography | 11.9% morbidity; 4.2% mortality |
| Shin et al.,70 2004 | 87.1% by angiography | 1.9% annual hemorrhage rate; 1.5% permanent morbidity |
| Pollock et al.,71 2004† | 75.4% by angiography; 22.8% by MRI | 12% morbidity from hemorrhage; 17.6% morbidity from radiation; 9% mortality |
| Maruyama et al.,72 2004 | 66% by angiography | 1.7% latency-interval hemorrhage rate per year for 1st 3 yr, then 0% |
| Fractionated Radiation Therapy for Large Arteriovenous Malformations | ||
| Karlsson et al.,73 2005 | 8% obliterated | 6% annual hemorrhage rate after radiation; 18% died of hemorrhage; overall 25% mortality (7 deaths, 2 of unknown cause) |
N/A, not available.
* It is not clear that the obliteration rates in these two articles were confirmed by angiography or magnetic resonance imaging and angiography.
† A series of grade IIIB and IV arteriovenous malformations (AVMs) in the basal ganglia, thalamus, and brainstem. Grade IIIB is defined as small Spetzler-Martin grade III AVMs located in areas where surgical resection is either too difficult or prohibitive.20
The main disadvantage of radiosurgery is its uncertainty of cure in terms of complete AVM obliteration and the fact that this process takes between 1 and 3 and sometimes 4 years to occur.61,74–76 It has been shown that during this “latent period” of incomplete obliteration the risk for hemorrhage is similar to the risk with untreated AVMs (3% to 4% per year),77–81 although there are also reports claiming lower rates72,80,82 and some series suggest that it may be higher (5.3% to 10%).83,84 In addition, there is a small but significant risk for neurological injury and other types of complications from radiation damage (3% to 10%, depending on location).84–87 There are a few reports of late hemorrhage after angiographically proven complete obliteration, and some reports describe histologically proven patency of some of the components of the AVM after angiographic obliteration. However, in view of the large number of AVMs treated by radiosurgery within the past 20 years, it can be concluded that the risk for future hemorrhage is extremely low and practically negligible once angiographic obliteration has been demonstrated. Clinically significant complications that can be directly attributed to radiosurgery, such as symptomatic radiation necrosis, appear in 3% to 6% of treated patients 
(Table388-E3). The morbidity of treating AVMs in very critical regions such as the brainstem and internal capsule is most likely substantially higher.71
Coexisting Intracranial Aneurysms
Thompson, Deruty, and their associates found coexisting intracranial aneurysms in 45 of 600 patients with AVMs (7.5%),88,89 and Cunha e Sa and colleagues’ analysis of 400 AVMs identified aneurysms in 39 patients (10%).90 They can be proximal to the lesion (prenidal on a feeding artery), within the lesion (intranidal), or in a remote location. The risk of these patients having an intracranial hemorrhage is higher than the risk in patients without associated aneurysms and is estimated at 7% per year91; another study found a higher rebleeding rate in AVMs associated with intranidal aneurysms.92 Patients with coexisting aneurysms tend to be older and are more frequently initially evaluated for epilepsy and neurological events.93 Half of these patients have multiple aneurysms. The majority (85%) of aneurysms are prenidal on a feeding artery or on a major arterial trunk that is participating in the supply. If the aneurysms are postnidal and remote from the AVM, their risk for rupture seems to be lower. In consideration of these numbers, we think that aneurysms need to be addressed during the initial evaluation and early in care if the location of the hemorrhage suggests aneurysmal rupture. Some authors also recommend initial treatment of the aneurysm, even when the AVM has bled, or in cases in which the source is not clear.
It has been proposed that hemodynamic stress from high flow or high arteriovenous shunting in feeding arteries is an important factor for the development of coexisting aneurysms; however, they can also develop in low-flow shunt situations. In high-flow situations, coexisting prenidal aneurysms may diminish in size and disappear when the shunt is obliterated.93
Timing of Surgery
In general, AVM resection should be elective surgery. Patients who sustain neurological deficits from AVM hemorrhage tend to improve over time, especially young patients. The occasional patient with a substantial intracerebral hemorrhage that requires evacuation as a lifesaving measure is an exception. Even in these cases, however, we recommend that the hematoma be evacuated in a very conservative manner in an effort to avoid having to deal with the AVM if it bleeds. The AVM in such a case should be treated at a later time under more favorable conditions. A few weeks after the hemorrhage, a repeat angiogram not infrequently reveals different angioarchitecture. The exception, of course, would be a very superficial, easily identifiable, and safely resectable AVM.
General Principles of Microsurgical Treatment
Positioning and Craniotomy
Positioning should help enable a rather perpendicular approach to convexity lesions and at the same time use gravity to minimize brain retraction. For this reason, the surface of convexity lesions is usually placed parallel to the floor—in other words, with the convexity representation of the lesion uppermost in the field.
For deep-seated AVMs, frameless stereotactic guidance can be a very helpful adjunct to allow a relatively small craniotomy and to tailor an optimized approach trajectory through noncritical areas of the brain—again, with the point of cortical entry uppermost in the field. Large superficial lesions, however, are approached through larger than necessary craniotomies to be able to map the vascular surface anatomy. Feeding arteries can be identified easier on the surface before they plunge deeply into a sulcus as they approach the AVM. One of the most important reasons for a larger craniotomy is to allow thorough mapping of the anatomy of the superficial draining veins. In the event of intraoperative hemorrhage, a craniotomy well beyond the confines of the lesion is highly appreciated when the operative field suddenly needs to be extended to remove the hematoma.
For intraoperative angiography, a radiolucent head frame can be used; however, carbon frames do not usually offer the same degree of positional freedom. With modern portable angiographic units, we have found the standard Mayfield clamp to work adequately. It is important to emphasize that in AVM surgery there are no “routine positions”; rather, they need to be tailored to the lesion and patient and are therefore highly individual.
Identification of the Malformation
After opening the dura carefully in an attempt to not injure potentially attached veins, the superficial anatomy needs to be very well defined, understood, and correlated with the angiographic image to identify superficial feeding arteries and at the same time distinguish them from arterialized draining veins. It is essential to correlate the preoperative angiographic findings with the surgical surface anatomy.
Elimination of Superficial Feeding Arteries
With convexity AVMs, the next step is to open all the sulci around the AVM under microscopic magnification in a search for superficial feeders, which are then coagulated and divided as they enter the AVM. Every effort has to be made to take feeders right at the point where they enter the nidus. It can be difficult at times to differentiate an arterialized vein from a feeding artery. If inspection under microscopic magnification is not sufficient for this process, use of a temporary clip greatly helps in differentiation: after clip application, an arterialized vein will tumesce between the clip and nidus but collapse or become softer and bluer distal to the clip; a feeding artery will collapse between the clip and nidus and continue to pulsate proximal to the clip.
In critical areas of the brain, it is of utmost importance to ensure that a particular artery actually ends in the AVM rather than being an “en passage” vessel that supplies brain tissue distal to the AVM. These en passage arteries feed the AVM via small lateral branches and are a particular problem with AVMs in the sylvian fissure and with pericallosal AVMs. Such arteries need to be very carefully dissected to be able to just take the small AVM-feeding side branches and preserve the main branch.
If larger feeding arteries cannot be coagulated completely because of the high flow, temporary aneurysm clips can be placed before transection. Usually, more proximal coagulation is then possible, and the clip can be taken off at a later time or left. Sometimes the flow before transection is so high that the closing pressure of a temporary clip is not high enough, in which case it needs to be exchanged for a permanent clip of similar size or a hemoclip. Generally, we prefer to use a permanent clip proximally (usually a hemoclip) in all feeders larger than about 1 mm.
An attempt is made to preserve all arterialized veins, but occasionally one or more superficial veins are sacrificed to facilitate dissection, provided that the major venous drainage is left intact and undisturbed.
Circumferential Dissection of the Nidus
In AVMs with a surface representation, the nidus is “developed” by circumferential corticectomy around the AVM after all superficial feeders have been identified and disconnected by opening the adjacent sulci as a first step. It is very helpful to carry this perpendicular dissection and corticectomy around the nidus to a depth of 2.5 to 3 cm before proceeding with deeper dissection. We have found empirically that a perpendicular corticectomy to such a depth will have disconnected all the superficial arterial supply. Once this has been accomplished, the dissection is continued in a more vertical plane around the AVM, usually in a “spiral” fashion until its deepest aspect is reached. During this deeper, spiraling dissection phase it is very important to not mistake loops and eccentric lobules of the AVM that project into normal brain for feeding arteries or draining veins. Coagulation and interruption of such loops can lead to significant and hard-to-control bleeding from the AVM. Great care should be taken to avoid coagulation of the AVM itself during the early phase, when internal AVM pressure and flow are still very high, because any decrease in nidus volume without a decrease in inflow will increase internal pressure and thus the risk for rupture. After the major arterial feeding pedicles have been taken, the turgor of the nidus will substantially diminish. Only then can coagulation be used to stroke some of the AVM loops in an effort to shrink them away from eloquent brain areas. It is not advisable at all to use such a maneuver in the early portion of the procedure when the AVM is still under high inflow pressure and thus more prone to rupture with manipulation.94 As the dissection approaches deeper areas, it usually becomes more tedious because more persistent bleeding will frequently be encountered as a result of the smaller and more fragile arterial supply from deep perforators or subependymal choroidal branches as a ventricle is approached. Once inadvertently lacerated or cut, bleeding from these deep vessels is extremely difficult to control because the vessel diameter is small and there is an element of retraction into parenchyma once cut. The late Dr. Thoralf Sundt designed microclips specifically for placement on these tiny fragile vessels. Sundt microclips are a far better option than creating an ever-deeper tunnel into normal brain in an attempt to reach the retracted and still patent bleeder with bipolar coagulation. Bleeding from the AVM itself can usually be stopped by placing a cottonoid pad, gently retracting on the bleeding point, and moving to a different plane of dissection. If one uses such a maneuver, it is important to remember to never pack when bleeding in a direction away from the AVM because this will risk significant parenchymal or intraventricular hemorrhage. In the case of AVMs that reach the ventricle, oftentimes the bleeding will not stop until the ependyma of the ventricle has been reached and small ependymal feeders to the AVM are controlled.
Transection of Deep Venous Drainage and Removal of the Lesion
If an important draining vein prematurely bleeds, it is usually advisable to attempt hemostasis with hemostatic agents and gentle pressure instead of coagulation and sacrifice. When the major arterial pedicles have been disconnected, the color of the large draining veins should become darker and finally change from red to blue because the drained blood will be progressively less arterialized. After inflow to the nidus has been completely eliminated, the nidus should deflate. The draining veins can be taken at that point unless they still have an arterialized element, which is frequently the case with the last large draining vein. Quite often there is one hidden arterial feeder left, in close proximity or underneath the vein, that needs to be identified and disconnected. After this final step the nidus can be removed. Complete resection of the malformation should be confirmed by intraoperative angiography.95,96 Promising new techniques are intraoperative ICG angiography (near-infrared indocyanine green fluorescence angiography),97–101 intraoperative computed tomographic (CT) angiography,102 and three-dimensional intraoperative ultrasound.103,104 It can be foreseen that the intraoperative application of magnetic resonance imaging (MRI) will not remain limited to tumor surgery.105,106
Hemostasis
The mainstay of AVM dissection is proper use of bipolar coagulation. Novices will instantly notice that because of the high flow it is a lot more difficult to coagulate an AVM feeder than a normal artery of the same diameter. We prefer intermittent coagulation for 1 or 2 seconds under constant irrigation. It is essential that the bipolar forceps tips not be brought together to avoid “sticking” of the tips. For these dissection steps the surgeon has to be very diligent in general and specifically with regard to instruments, and the bipolar tips need to be maintained absolutely clean at all times. This necessitates very frequent cleaning by the operating room nurse, and it is therefore more efficient to have an identical set of bipolar forceps available that can be alternated. If one of the aforementioned deep fragile AVM vessels “sticks” to the bipolar tip, avulsion of this vessel can lead to serious problems at any point in the operation but particularly during the end phase of AVM resection. Frequently, a refractory source of bleeding in the depth of the nidus resection cavity is a small satellite AVM lobule that is still hidden under a thin layer of parenchyma after having mistaken the AVM to have a single feeder.
Once the lesion has been removed, the resection cavity needs to be checked meticulously for potential points of breakthrough hemorrhage. After thorough irrigation and inspection, systolic blood pressure is raised by 15 to 20 mm Hg above normal mean arterial pressure for that patient for 10 to 15 minutes. If bleeding can be provoked, usually a residual portion of the AVM is left. By rubbing gently with a small cottonoid against the resection wall, suspicious areas can be tested. If no bleeding can be provoked, it is time to line the cavity with the usual hemostatic agent. From then on the patient’s blood pressure is kept at or slightly below the normal mean for 24 hours.
Specific Surgical Considerations for Arteriovenous Malformations in Different Locations
Convexity Arteriovenous Malformations
Resection of convexity AVMs follows the principles outlined earlier. In patients with additional arterial supply from external carotid branches, enhanced alertness is necessary when the craniotomy flap is turned to avoid serious bleeding in the very early proceedings of the surgery. Multiple bur holes and very careful stripping and opening of the dura can help avoid injury to meningeal vascular pedicles. Elimination of all or most of the external supply by preoperative embolization is most helpful in these cases. Generally, we have noted exuberant external supply only in AVMs that have bled or have been embolized.
Deep Parasagittal Arteriovenous Malformations
Anterior frontal parasagittal lesions
