Acta Neurochir (Wien) (1996) 138:119-131

Acta Neurochirurgica
9 Springer-Verlag 1996
Printed in Austria

Complications After Multidisciplinary Treatment of Cerebral Arteriovenous
Malformations
R. Deruty ~, I. Pelissou-Guyotat ~, D. Amat 1, C. Mottolese 1, Y. Bascoulergue 2, F. Turjman 2, and J. P. Gerard 3
Department of Neurosurgery, 2 Department of Neuroradiology, Hopital Neurologique et Neurochirurgical, Lyon, and 3 Department of
Radiotherapy, Centre Hospitalier Lyon Sud, Lyon, France

Summary
Patients and techniques. A series of 67 patients treated for cerebral AVMs with a multidisciplinary approach is reported, with special attention for the complications due to treatment. The malformations were classified after the Spetzler Grading Scale, with 67%
low-grade and 33% highgrade AVMs. Three modes of treatment
were used: surgical resection, endovascular embolization, and radiosurgery (linear accelerator technique). The actual treatment was:
resection alone (25% of cases), embolization plus resection (24%),
embolization alone (2l %), and radiosurgery (30%), either alone or
after embolization or surgery. The following eradication rates were
obtained: overall 80%, after resection (with or without embolization) 91%, after embolization alone 13 %, after radiosurgery 87%.
Clinical outcome. The outcome was evaluated in terms of deterioration due to treatment. A deterioration after treatment occurred
in 19 patients (28%), and was a minor deterioration (19%), a neurological deficit (4%), or death (4%). As far as the mode of treatment is concerned, surgical resection was responsible for deterioration (minor) in 17% of all cases operated upon. Radiosurgery was
followed by a minor deterioration in 10% of irradiated cases.
Embolization gave a complication in 25% of all embolized cases
(minor or neurological deficit, or death). The mechanism of the
complications was: resection or manipulation of an eloquent area
during surgery, radionecrosis after radiosurgery, ischaemia and
haemorrhage (50% each) following embolization. In most cases of
haemorrhage due to embolization, occlusion of the main venous
drainage could be demonstrated.
Discussion. The haemodynamic disturbances tO AVMs and to
their treatment are reviewed in the literature. The main haemodynamic mechanisms admitted at the beginning of a complication
after treatment of cerebral AVMs are the normal perfusion pressure
breakthrough syndrome, the disturbances of the venous drainage
(venous overload or 'occlusive hyperaemia), and the retrograde
thrombosis of the feeding arteries.
Conclusions. According the authors' experience, the emphasis
of treatment for cerebral AVMs has now shifted from surgical
resection to endovaseular embolization. One of the explanations is
that endovascular techniques are now employed in the most diffi-

cult cases (high grade AVMs). As severe complications of endovascular embolization may also occur for low-grade malformations, the question arises whether surgery or radiosurgery should
not be used first for this low-grade group even if embolization is
feasible.
Keywords: Cerebral AVMs; endovascular embolization; radiosurgery.

Introduction
The treatment of cerebral arteriovenous malformations (AVMs) which had long being restricted to neurosurgery has lately become more and more multidisciplinary. Modern methods of treatment arose during
the past years: first endovascular embolization, and
then radiosurgery. These three modes of treatment
(surgical resection, endovascular embolization, radiosurgery) may be used either alone or in various combinations [11]. With this multidisciplinary approach,
are the complications of treatment still unpredictable
and catastrophic? During the period in which surgery
was the sole available method of treatment for cerebral AVMs, special interest focussed on complications based on haemodynamic modifications due to
the presence of AVM. The question arises nowadays
whether, with the new methods of treatment, these
haemodynamic complications still exist.
In this paper, we report a series of cerebral AVMs
treated by a multidisciplinary approach, with special
attention to the complications which occurred after
the treatment. Then we review the neurosurgical literature dealing with the haemodynamic disturbances
relating to the presence of cerebral AVMs or occurring
after the treatment.

120
Patient

R. Deruty et al.: Complications of Cerebral AVMs
and

Techniques

Distribution of Patients

Between 1989 and 1993, 67 patients were treated in our Department of Neurosurgery for an intracranial AVM (average 13 per
year). These AVMs were classified according to the Spetzler's
Grading Scale into five grades (Grade I: 4%, Grade II: 29%, Grade
III: 34%, Grade IV: 30% and grade V: 4%). Considering together
the low-grade AVMs (I, II and III) and the high-grade AVMs (IV
and V), the figures are respectively 67% and 33%. The distribution
according to gender was: female 33% and male 67%. According to
age, the distribution was: up to 30 years 30%, from 31 to 50 years
54%, and over 50 years 16% (maximum 65 years).
Methods of Treatment

Three modes of treatment were used: surgical resection (in 52%
of cases), endovascular embolization (in 60% of cases) and radiosurgery with the linear accelerator technique (30% of cases). These
three methods were used either alone or in various combinations,
giving four possibilities: resection alone (25% of cases), embolization + resection (24%), embolization alone (21%) and radiosurgery
(30%), alone or after surgery or after embolization.

Results

Angiographic Eradication

60 patients underwent post-treatment angiography,
and the following figures o f complete eradication
(calculated on those patients angiographically
checked) were obtained:
1.) overall: 80%,
2.) after surgical resection (with or without e m b o l ization): 91%,
3.) after embolization alone: 13%,
4.) Two years after radiosurgery: 87%.

Clinical O u t c o m e
Outcome evaluation. Clinical results are evaluated in terms of
deterioration following treatment, in five groups: 1) No deteriortion, 2) minor deterioration, 3) long-lasting neurological deficit, 4)
major deterioration, and 5) death. Minor deterioration means a
moderate neurological deficit, such as hemianopia, hemiparesis,
moderate speech disturbances. Major deterioration means severe
neurological deficit or severe intellectual impairment, or vegetative
survival, the patients being bed-ridden or dependent for self care.
The sub-group "long-lasting neurological deficit" includes patients
who present with hemiplegia or aphasia (and so cannot be classified in the "minor deterioration group") but these patients are independent in the daily life and sometimes are able to work (and so
they cannot be classified in the "major deterioration group").
In 20 patients, a deterioration occurred after treatment (30%). One o f these patients was admitted in a
deep c o m a due to spontaneous rupture o f the AVM,
and was submitted to a "last c h a n c e " or "rescue"
operation. She died, and this death cannot be ascribed

to surgery. So the treatment alone was responsible for
the deterioration in 19 patients (28% of all patients).
D e g r e e o f deterioration. - 13 patients presented
with a minor deterioration (68% o f all deterioration
cases, and 19% o f all treated cases), either hemianopia (7 patients) or moderate aphasia (4p) or hemiparesis (2p). In each case the neurological impairment
c o r r e s p o n d e d to the location of the A V M in the corresponding eloquent area
- 3 patients presented with a neurological deficit,
also relating to the A V M location (16% o f all deterioration cases and 4% o f all treated cases),
- 3 patients died after treatment (16% of deterioration cases and 4% o f the total series).
M e t h o d o f t r e a t m e n t a n d deterioration. - Surgical
resection was responsible for 6 cases o f deterioration
(17% o f all operated cases). All were in the " m i n o r
deterioration" group. In each case, the deterioration
corresponded to the anatomical location o f the AVM.
- Radiosurgery was responsible for 2 cases of
minor deterioration, due to deferred radionecrosis
(10% of all irradiated cases).
- E n d o v a s c u l a r embolization was responsible for
10 cases o f deterioration (25% of all embolized cases). This deterioration was o f the minor type in 5 cases (13% o f embolized cases), o f the neurological deficit type in 2 cases (5%), and resulted in death in 3
cases (8% o f embolized cases).
- One patient is m o r e difficult to classify and could
be a complication o f the "embolization + resection"
type of m a n a g e m e n t (6% of all cases m a n a g e d with
this method). Surgery was planned and carried out 3
weeks after an incomplete embolization for this
patient. Under anaesthesia and during the opening of
the skull the A V M ruptured. Resection o f the A V M
could be performed and this patient survived with a
neurological deficit.
M e c h a n i s m o f deterioration. Various m e c h a n i s m s
were responsible for the deterioration in these 19
patients:
- Resection or manipulation of an eloquent area o f
the brain during surgery (6 patients), responsible for a
m i n o r deterioration in each case.
- Radionecrosis after radiation therapy (2 patients).
In these cases, the condition was at first severe (neurological deficit) and progressively i m p r o v e d after
some months. Eventually the patients suffered a
m i n o r deterioration.
- Cerebral ischaemia (5 patients). This ischaemia
occurred during the embolization procedure itself,

121

R. Deruty et al.: Complications of Cerebral AVMs
either through o c c l u s i o n of an A V M feeder (3 cases,
e n d i n g with a m i n o r deterioration) or by o c c l u s i o n of
the m a i n t r u n k of the M C A (2 cases, e n d i n g with a
n e u r o l o g i c a l deficit). T h e s e cases with i s c h a e m i a
i n c l u d e d 2 grade II, 2 grade III, and 1 grade IV AVM.
- H a e m o r r h a g e (6 patients). These cases i n c l u d e d
1 grade I, 1 grade III, and 4 grade IV AVMs. O n e
p a t i e n t (grade IV) bled at the b e g i n i n g of surgery 3
weeks after i n c o m p l e t e e m b o l i z a t i o n of the AVM.
W h e t h e r this e m b o l i z a t i o n is the cause of the A V M
rupture r e m a i n s hypothetical. In 5 patients h a e m o r rhage could be linked up with e n d o v a s c u l a r e m b o l i z a tion. In 3 of these 5 patients, the rupture of the A V M
occurred d u r i n g the e m b o l i z a t i o n procedure itself,
g i v i n g a deep c o m a with decerebrate rigidity. All 3
patients died, despite e v a c u a t i o n of the clot and resection of the A V M in two of t h e m (so called "rescueo p e r a t i o n " despite the desperate clinical status). In the
other 2 patients with h a e m o r r h a g e , the rupture of the
AVM occurred d u r i n g the first week f o l l o w i n g the
e m b o l i z a t i o n procedure (2 days and 5 days). Both
these patients were operated upon, their m a l f o r m a t i o n
was resected, and they s u r v i v e d with a m i n o r deterioration. O n the post e m b o l i z a t i o n a n g i o g r a m , occlusion of the m a i n v e n o u s d r a i n a g e could be d e m o n strated in 4 out of these 5 cases of p o s t - e m b o l i z a t i o n
haemorrhage.

Illustrative Cases
Case l: Male, 38 years old, admitted after seizure of the "grand
real" type. The AVM was located in the left temporal lobe (middle
part), with several feeders coming from the left MCA, and a steal
phenomenon through both the anterior and left posterior communicating arteries. The largest diameter of the nidus was 5 cm. The
venous drainage was of the deep type, towards the great vein of
Galen, with a large venous ectasia. This AVM was classified as a
Grade IV according to the Spetzler's scale. Intravascular embolization was undertaken (September 27, 1990). During the procedure,
the AVM ruptured with an intra-temporal haemorrhage. Clinically
the patient was in a deep coma with bilateral mydriasis. Despite this
desperate clinical status, he was immediately operated upon, the
AVM was resected, hut the brain remained turgescent and herniated and the patient died rapidly. When reviewing the angiographic
images, it was evident that the venous &ainage, and particularly the
venous ectasia, was partially occluded (Fig. 1).
Case 2: Male, 45 years old. The AVM was located in the right
temporal lobe (posterior part), the greatest diameter of the nidus
was 3 cm, two feeders came from the right MCA, the venous drainage was superficial (towards both the sagittaI and the right transverse sinuses). The AVM was classified as a Spetzler grade I. Intravascular embolization was performed (January 5, 1993), through
the right MCA, with a large but incomplete eradication of the malformation: persistance of a small part of the nidus, (filling slowly),
late opacification of the venous drainage, disappearing of the drain-

age towards the transverse sinus. Only minimal drainage towards
the sagittal sinus was visible (Fig. 2). At the end of the embolization procedure, the patient complained of headache and fell into
deep coma. The CT scan showed a huge intra-temporal haematoma.
The patient was rapidly operated upon, the AVM resected, the clot
evacuated. At surgery the nidus proved to be almost completely
occluded, with only one vein, still arterialized. The patient
remained comatous and died two days later.
Case 3: Female, 41 years old, admitted after two seizures (Bravais-Jackson type). The AVM was located in the left parietal lobe,
supplied through 2 long and tortuous feeders (coming from the left
MCA and the left PCA). The diameter of the nidus was 6 cm. The
drainage was superficial, towards the sagittal sinus. The malformation was classified as a Spetzler grade IV. Intravascular embolization was undertaken (September 15, 1992). The major part of the
nidus was occluded, together with the feeder coming from the
MCA and the venous drainage. The nidus was still partially supplied through the PCA, but no venous drainage was visible coming
out of this residual nidus. Two days later, just prior to discharge,
the patient became suddenly comatose, with a right hemiplegia,
showing deep haemorrhage on the CT scan. She was taken to the
intensive care unit, and progressively recovered. An angiogram
performed 3 weeks after the haemorrhage failed to show the AVM,
which at the time was thought to be completely eradicated. Despite
this constatation, the patient was operated upon and the malformatin was resected. The nidus was found occluded, the venous drainage was thrombosed. However, several thin deep feeders were
found still patent. The post-operative course was satisfactory, the
right hemiplegia recovered, the patient remained with a moderate
right hemianopia (minor deterioration), and a post-operative anglogram confirmed the complete eradication of the malformation.
Case 4: Female, 42 years old, admitted after two seizures
("grand mal" type). The AVM was located in the right parietal area,
the greatest diameter of the nidus was 5 cm, the venous drainage
was superficial, through the sagittal sinus (Spetzler grade III).
Intra-vascular embolization was carried out (October 13, 1993), a
50% reduction of the volume of the nidus was obtained, with a late
filling of the venous drainage. A second embolization procedure
was planned one week later and the patient was discharged. She
was readmitted as an emergency 5 days later, stuporous, without
neurological deficit, the CT scan showing an intra-cerebral and
intraventricular haemorrhage. After four days in the intensive care
unit, she was operated upon. At surgery, the nidus was found
almost completely occluded, one superficial arterial feeder
remained patent, the venous drainage was completely thrombosed.
At the end of the resection, the deeper part of the nidus, next to the
ventricle, was thrombosed, but several small and thin arterial feeders were still patent. The patient recovered after surgery, presenting
with a moderate left hemianopia (minor deterioration), the eradication of the AVM proved to be complete on the post-operative
angiogram.

Discussion
I n the n e u r o s u r g i c a l literature, various h a e m o d y n a m i c disturbances have b e e n described, relating to a
cerebral AVM. These disturbances, caused b y the
existence of the m a l f o r m a t i o n itself, m a y also be
r e s p o n s i b l e for c o m p l i c a t i o n s after t r e a t m e n t of the
AVM.

Fig. 1. Case 1. (a) Left carotid angiogram before embolization (lateral view). Note the
deep venous drainage into the great vein of Galen and the straight sinus, with a venous
ectasia. There is an additional drainage, superficial, directly into the lateral sinus. Note the
intracranial steal around the malformation. (b) CT scan after onset of the haemorrhage,
during the embolization procedure (intracerebral, intraventricular and subarachnoid haemorrhage). (c)Post-haemorrhage angiogram (left carotid, larteral view). Note l) the
incomplete occlusion of the nidus, 2) the partial occlusion of the venous ectasia, with
embolic material inside, 3) the suppression of the deep venous drainage, with persistance
of the superficial drainage

r

3

O

o

,~

t~3

Fig. 2. Case 2. (a) Right carotid angiogram (lateral view). Note the venous drainage, towards the sagittal sinus and towards the lateral sinus. ( b ) C T scan after
onset of the haemo~Thage, demonstrating a huge right imra temporal haematoma.
(c) Right carotid angiogram (lateral view) after embolization. Note the incomplete
occlusion of the nidus, and the suppression of the venous drainage towards the
sagittal sinus. The drainage towards the lateral sinus is very slow and does not
reach the sinus itself

tad

V_,

>
<

g-

9

,7

124

Circulatory Modifications Due to the AVM
In the presence of an AVM, changes in the vessels
relating to the malformation have been documented.
The pressure in the feeding arteries is reduced, and
becomes much lower than the systemic pressure, as
demonstrated by means of intra-operative Doppler
sonography investigations [4, 12, 15, 26]. In the feeding arteries, Nornes [26] found a mean arterial pressure of 56 mm Hg. The decrease of this low arterial
pressure was proportional to the number and to the
length of these arteries. On the contrary, the pressure
in the draining veins is increased, and the figures are
far higher than in the normal veins: Nornes [26]
found a mean venous pressure of 15 mm Hg. As a
consequence of both a low arterial pressure and a high
venous pressure, the perfusion pressure of the surrounding brain is reduced. There may be some exceptions, and Pertuiset [29] described some cases in
which the venous pressure was found low: in these
cases, the venous drainage was unique and went into
the sagittal sinus via an enlarged vein of Galen.
In the feeding arteries, the flow velocity is
increased [12, 16]. By means of transcranial Doppler
sonography (TDS), Lindegaard [16] measured the
blood flow velocity in the basal arteries. In the arteries feeding the AVM, he found a high flow velocity
(mean 124 cm/sec instead of 65 cm/sec in normal
arteries), and a low pulsatility index (median 0.48
compared to 0.87 in normal arteries). The flow velocity was also found increased in the carotid arteries in
the neck [2, 28].
The vasomotor reactivity was found altered, in
vivo and in vitro. Hassler [12] using in vivo microvascular sonography measurements, found a poor vasomotor reactivity to CO2 changes in the AVM feeders.
Murasko [24] took isolated fragments of feeding
arteries of AVM and tested them in vitro with various
vaso-active substances; he compared the in vitro
results with the spontaneous vessel activity (spasm,
ectasia, postoperative course). He found that some of
the AVMs had non-reactive feeders in in vitro testing,
and that these vessels had also no spontaneous activity. He could not find any histological basis for the
lack of reactivity, and made the hypothesis that these
vessels segments might act as passive conduits in vivo.

The Steal Phenomenon
An intracranial steal may be induced in the surrounding brain by an AVM, as demonstrated by blood
flow studies and by COz reactivity measurements.

R. Derutyset al.: Complicationsof Cerebral AVMs

The regional cerebral blood flow was often found
decreased in cases of AVM, especially in sites near to
the malformation [4, 6, 15, 27, 41]. Batjer [6] found a
decrease of the blood flow ipsilaterally and contro
laterally to the AVM. He found a more severe steal in
older patients and in patients with progressive deficit.
Lo [17] using a biomathematical analysis of haemodynamic alterations in AVMs demonstrated that the
vascular steal is inversely proportional to the haemodynamic resistance of the AVM: an AVM with
lower resistance will shunt more blood away from the
surrounding vessels. The relationship between the
size of the AVM and the severity of the steal remains
controversial: Batjer [6] found no relationship. On the
contrary, for Leblanc [15] the steal is more severe for
large AVMs. Contrary to these studies in which a
reduced CBF was found, Young [42, 43] found a normal CBK around the AVM. However, A1 Rhodan [1]
recently stated that these CBF studies, performed
with different techniques, make direct comparison
inaccurate; moreover in the cases of Xenon measurements, areas of hypoperfusion may not be revealed
due to impairment in the delivery of the indicator in
the region of interest.
The C02 reactivity is modified in the brain surrounding the AVM. Normally, the administration of
Acetazolamide produces changes in the tissue CO2
which provoke vessel dilation, and increase the rCBF
by 35%. In cases of AVM, this CO2 reactivity has
been frequently found reduced. This fact is in favour
of a decreased perfusion pressure and thereby proves
the presence of a steal [4, 15, 31]. But these findings
have been criticized. The CO2 reactivity was sometimes found normal [42, 43], which indicates a dissociation between pressure autoregulation (lost) and
CO2 reactivity (maintained). Last of all, Batjer [5]
reported a patient with an AVM who developed a fatal
hyperaemic complication and in whom the CO2 reactivity was increased (more than normally) before any
treatment. This constatation indicated that the vessels
surrounding the AVM were still capable of further
dilation after Acetazolamide delivery. Then Batjer
studied [81 a series of patients with rCBF and Acetazolamide administration. An increase of vasodilation
was noted in the following conditions: 1) patients
with perforator recruitment, 2) patients with anglographic steal phenomenon, 3) patients who developed
ischaemic complications, 4) patients with poor neurological outcome.
Steal demonstration using other methods. Apart

R. Deruty et aI.: Complications of Cerebra. AVMs

from the angiographic demonstration of steal, which
is well known, Leblanc [15] using intra-operative fluorescein angiography described an early filling of
feeders, nidus and veins, and a late filling of the arteries in the surrounding cortex. He also found, in PET
studies, a fall in glucose metabolism proportional to
the size of the AVM.

Mechanism of the Clinical Manifestations
The changes described above may be responsible
for the onset of spontaneous clinical manifestations.
Various mechanisms have been advocated.
(1) The size of Al/-M and the pressure in the feeding
arteries. Spetzler [32] studied the relationship
between the size of an AVM and its propensity to haemorrhage. He measured the pressure in the arteries
feeding the AVMs, and found that smaller AVMs had
higher feeding artery pressures than larger ones. He
concluded that, it perfusion pressure increases as
AVM size decreases, then smaller AVMs with higher
feeding pressure would be more likely to haemorrhage than large AVMs under low pressure. And in
patients whose AVM had ruptured, he actually found
a significantly higher pressure in the feeding arteries.
This disparity in feeding artery pressure could also
explain the fluctuating neurological deficit with large
AVMs (the steat symptoms). The low feeding artery
pressure would be likey to provide a low perfusion
pressure to the surrounding cortex, thereby producing
relative ischaemia.
(2) The strength of the shunt walls. This parameter
was studied by Lo [17] with his biomathematical
model: He demonstrated that haemorrhage occurs
when the transmural stress exceeds the strength of the
shunt walls. The transmural stress is proportional to
the vessel radius and inversely proportional to the
wall thickness. Therefore, large thin-walled shunts
would be more likely to rupture.
(3) The steal phenomenon. Manchola [18] also
tried to relate AVM haemodynamics with clinical
findings, and calculated the flow volume to the AVM,
by means of transcranial Doppler and angiographic
data. He found that patients with haemorrhage had
significantly lower flow volume than patients having
other symptoms and signs. Conversely, patients presenting with neurological deficit had higher flow volumes than other patients (which favour the hypothesis
of a steal).
(4) The role of the venous drainage. Taneda [39]
studied the response of the CBF to hypotension in the

125
presence of an AVM. Before the resection of the
AVM, he found regions in the brain where CBF
increased in response to hypotension. After total
resection of the AVM, this increase in the CBF was no
longer obtained. This phenomenon could be
explained by the increased pressure on the venous
side of the AVM circulation: As a consequence of the
shunt, there is an increased back-pressure and so a
reduction in the perfusion pressure in the territory of
the veins draining the AVM. Therefore, decrease in
systemic arterial blood pressure, transmitted to this
venous system, could reduce the elevated venous
blood pressure and increase the CBF in the corresponding tissue around the AVM.
(5) The regional cerebral atrophy. Tanaka [38]
described on MRI morphological changes related to
the haemodynami c dysfunction in areas remote from
the AVM. He compared these changes to the "diaschisis phenomenon" (which correspond to reversible
changes occurring after acute stroke): In cases of
AVM, hypoperfusion can end in secondary atrophy in
the basal ganglia. Tanaka proposed to call these irreversible morphological changes the "diaschisis-like
phenomenon" caused by metabolic depression due to
the decrease in CBF, being different from the steal
phenomenon and responsible for clinical manifestations.

Haemodynamic Modifications Occurring After Resection of the AVM
Excision of the AVM produces immediate haemodynamic changes in the surrounding brain [4, 5, 7,
12, 15, 25, 26, 30, 31, 35, 43] and these changes convert a shunt flow into a perfusion flow [15]. The arterial pressure increases and becomes normal. Nornes
[26] found an increase from 56 to 76 mm Hg (average). Hassler [12] also found an increase of arterial
pressure by 60% to normal values after occlusion of
the feeding vessels. The venous pressure decreases
and becomes normal (from 15 to 0 mm Hg, average
for Nornes). These modifications result in an acute
increase in the arteriovenous pressure gradient [43]
and so the cerebral perfusion pressure increases and
becomes normal. The flow velocity decreases and
becomes normal [12]. The cerebral blood flow
increases in the ischaemic rim around the AVM [15].
This increase may become exaggerated: the rCBF
may rise over the normal values, up to two times the
pre-excision values [25, 35] and may reach the remote
areas, from 2 to 4 cm from the resection site [4]. In

126

most cases, the C O 2 reactivity is restored to normal,
and sometimes may show an hyperresponsiveness [5,
8].

Haemodynamic Complications Following the Eradication of AVM
Complications may follow the eradication of the
AVM. These complications have been described
mostly after surgical excision, but also more recently
after endovascular occlusion of the AVM. They may
occur after complete or partial eradication of the malformation, both intra- and postoperatively and consist
of brain oedema and/or haemorrhage in the normal
surrounding parenchyma. These complications have
been related to haemodynamic mechanism, but this
precise mechanism is far from being clearly understood. Several theories have been developed so far.
(1) The normal perfusion pressure breakthrough
(NPPB) theory was proposed by Spetzler in 1977
[31]. In the presence of a large AVM, the relating vessels offer no resistance to blood flow. The consequence is an intracranial steal and the surrounding
hemisphere is ischaemic. On the contrary, the small
surrounding vessels (not relating to the AVM) offer a
high resistance to blood flow and must remain maximally dilated to correct the chronic steal and the
chronic ischaemia. Thus, the conjunction of a chronic
ischaemia and of a chronic vessel dilation induces a
state of vasomotor paralysis which results in a loss of
autoregulation in the brain surrounding the AVM.
After occlusion of the malformation, the flow is redirected to the surrounding hemisphere, and this
ischaemic hemisphere returns to a normal perfusion
pressure. But, due to the loss of autoregulation, the
paralysed vessels supplying the functional brain are
unable to constrict in response to this suddenly elevated perfusion pressure. So there is an impossibility to
increase the resistance of the arterial network (which
has been chronically dilated) to regulate this new
pressure, and hence an impossibility to protect the
capillaries. Therefore the conjunction of occlusion of
the AVM and of loss of autoregulation results in disruption of the capillary bed: the capillary breakthrough, responsible for oedema and haemorrhage in
the adjacent brain.
This NPPB theory which is supposed to occur only
in large AVMs (that is not very frequently), has had
supporters and detractors. The supporters [4, 10, 15,
28] advocate that this breakthrough complication may
well be explained by the pre-operative decrease of the

R. Deruty, et al.: Complications of Cerebral AVMs

cortical arterial pressure, of the CBF, and of the CO2
reactivity, and by the postoperative normalization of
these parameters. And Young [43] proposes a scheme
with four possible haemodynamic responses to AVM
occlusion: 1) No change in AV gradient with shunt
occlusion. 2) Increased pressure gradient with appropriate autoregulatory vaso-constriction and no CBF
increase. 3) Increased pressure gradient with impaired
autoregulatory vasomotion and resultant CBF
increase. 4)Same as 3) but with perfusion related
complications.
However, several criticisms have been directed
against the NPPB theory. Hassler [12] for example
does not agree with this hypothesis, for he found no
modification of pressure and flow velocity (neither
before nor after AVM resection) in the remote brain
arteries, but only in the feeders of the AVM. For Morgan and Sundt [21] this denomination (NPPB) is not
appropriate, and this complication should rather be
called normal pressure hyperperfusion breakthrough.
Their hypothesis is that surgery, and especially multistaged surgery, could end into an "acute-on-chronic
hypoperfusion", with the following mechanism: in
the presence of both chronic hypoperfusion and loss
of autoregulation, an intra-operative deep hypotension could increase both the surrounding ischaemia
and the capillaries lesions, and could favour the
occurrence of micro infarcts. So, when the arterial
pressure returns to normal, the occurrence of oedema
and haemorrhage could be facilitated.
Recent haemodynamic studies exploring the CBF,
ended also in criticisms against the NPPB theory. To
test the mechanism of NPPB phenomenon, Young
[44] measured the CBF before and after resection of
the AVM, and again after increasing the mean arterial
pressure by 20 mm Hg. Immediately after resection,
he found a significant increase in CBF, but no further
increase after elevation of the systemic arterial pressure. This experiment demonstrated that the vessels
around the AVM were still able to constrict in reaction to an increased perfusion pressure, and that the
autoregulation was preserved. So, even if there are
probably certain cases in whom complications occur
as a result of impaired autoregulation, there is also a
general hyperperfusion response that does not involve
pressure autoregulatory mechanisms. Young and his
colleagues [45] conducted a second study using
induced hypertension during embolization procedures. In the patient studied, rCBF did not change
after induced hypertension. The authors concluded

R. Deruty et al.: Complicationsof CerebrN AVMs
that these patients were not in a "pressure-passive"
state, but that there could exist in response to chronic
hypotension a downward shift of their autoregulatory
curve to the left, thereby adaptating the lower limit of
autoregulation in affected vascular territories. The
role of the CO2 reactivity was also criticized by Batjer
[8] after his study using CBF and administration of
Acetazolamide: patients who developed hyperaemic
complications actually demonstrated an increased
vasodilation (greater than in normal brain tissue), and
this fact does not support the NPPB theory. For Batjer
these findings could suggest smooth muscle derangements in cerebral vasculature in some AVM patients.
(2) Disturbances of the AVM drainage have also
been advanced as a source of post treatment complications. Miyasaka [20] pointed out that postoperative
haemorrhagic infarction caused by thrombosis in the
draining veins should be considered as a distinct complication after removal of an AVM. In his experience
several arguments were in favour of the venous origin
of such a complication: no evidence of residual AVM,
delayed onset of the haemorrhage (inconsistent with
the time course of a NPPB or with insufficient haemostasis), good correlation between the location of
the haemorrhage and the occlusion of the draining
vein, and multifocal haemorrhage affecting both the
gray and the white matter. More recently, two theories, very similar to each other, have been proposed
regarding the venous mechanism of post-treatment
complications.
a) The theory of "venous overload" was proposed
by Wilson [40]: For this author, interference with
venous drainage represents the single factor responsible for almost all clinically significant complications
of AVMs following treatment. A spontaneous thrombosis of a major draining vein may cause occlusive
hyperaemia and abrupt neurological deterioration.
This venous complication could be related with the
time. Early deterioration (first day or so) could reflect
differing degrees of venous restriction, and late deterioration (after several days) could be attributed to
delayed thrombosis of one or more critical draining
veins. In Wilson's opinion, the NPPB theory exists,
but is neither common nor a universal explanation for
complications related to the AVM eradication. In
some cases, a combination of both NPPB and hyperaemia due to venous occlusion could be responsible
for these complications.
b) The theol 7 of "occlusive hyperaemia" was proposed by A1-Rhodan, Sundt, and co-workers [1].

127
These authors emphasized the role of the veins that
drain the AVM as well as the surrounding brain parenchyma. Inadvertent embolization or surgical ligation
of these veins is likely to precipitate passive hyperaemia and engorgement with subsequent haemorrhage
and brain oedema, and a secondary reduction in arterial perfusion. The conclusion of Al-Rhodan's study
is that two separate but interrelated mechanisms may
be responsible for the complications of haemorrhage
and oedema following the resection of high flow
AVMs: 1) Stagnant arterial flow in former AVM feeders and in their parenchymal branches, with subsequent worsening of the existing hypoperfusion and
ischaemia, and subsequent haemorrhage and oedema.
2) Obstruction of the venous outflow system of brain
adjacent to the AVM, with subsequent passive hyperaemia, engorgement, and further arterial stagnation.
In addition, disturbed autoregulation at the microcirculation level may play a role.
(3) Retrograde thrombosis of the feeding arteries.
Nornes [26] pointed out that the length of the feeders
is a very important factor: in his experience, no brain
swelling was observed in cases with short feeders. In
the contrary, there is a high risk in cases with feeders
longer than 8 cm. Hassler [12] agrees on this risk due
to the length of the feeders. Miyasaka [9] also
described this complication, which was found in 6.6%
of all his cases, and in 14% of medium-sized and large
AVMs. He suggested that the development of retrograde thrombosis could be due partly to slow flow in
the former feeding arteries that was observed immediately after AVM removal and partly to pathological
changes in these vessels due to long-standing haemodynamic stresses.

Therapeutic Implications
Therapeutic measures are proposed on the basis of
the theories described above~
(1) The prevention of the NPPB. The principle of
the method for the prevention, as established by
Spetzler [31], is to gradually increase perfusion to the
ischaemic hemisphere by graded occlusion of the
AVM. This aim may be obtained through staged ligation or staged embolization of feeding arteries [23,
26, 31], or through multistaged resection [28, 31].
Postoperative hypotension is recommended [4, 10,
15, 25, 31]. Intra-operative hypotension is more controversial: sometimes recommended [25, 26, 28], but
sometimes not [21, 46], for lowering the blood pressure during surgery could increase the pre-existing

128

ischaemia and the hyperaemic complications. Some
authors have proposed partial temporary clamping of
the cervical carotid artery rather than lowering the
systemic arterial pressure. This arterial clamping may
be performed either intra-operatively [9, 28] or postoperatively [35, 36] or both, lasting for several days
and then gradually tapered off. Several additional
measures have been proposed during the post-operative period [15]: intensive care unit during at least 24
hours, reduced fluid intake, administration of diuretics and osmotherapy, and if ICP raises, barbiturate
coma.
(2) Prevention of occlusive hyperaemia (or of the
venous overload). In this hypothesis [1] in which
occlusive hyperaemia is a greater risk than NPPB, the
staged surgical resection is not recommended (except
if technically necessary). Whether the AVM resection
is accomplished in one or multiple stages, with or
without embolization, several precautions must be
taken to reduce the risk of occlusive hyperaemia:
avoiding injury of veins that drain both the AVM and
the surrounding brain parenchyma; clipping of the
terminal arteries as close to the malformation as possible, and protection of the vessels that supply normal
brain beyond the malformation; avoidance of excessive intravascular hypotension that could increase
cerebral ischaemia and predispose to sludging,
aggregation and clumping of formed elements (platelets and red blood cells) in the draining veins. Wilson
[40] also thinks that reducing the fluid intake during
the postoperative period and using the various means
of reducing oedema might increase the risk of venous
thrombosis. He recommends to assure an adequate
blood volume, to avoid haemoconcentration, and
even to give aspirin and low-dose heparin therapy
immediately after surgery.
(3) The role of selective embolization. Intravascular embolization has been advocated either as sole
treatment of AVMs, or more often as a pre-operative
method in the combined management of AVMs [11].
Several questions arise as far as the efficacy and the
innocuousness of embolization is concerned. Is there
a benefit after embolization on the haemodynamic
disturbances due to the malformation? The angiographic suppression of the intracranial steal due to the
AVM is a common finding even after partial eradication. Sugita [33] reported a case demonstrating the
correction of the steal phenomenon and the restoration to normal of a visual field defect after embolization. Takeuchi [37] investigating the effects of embol-

R. Deruty et al.: Complications of Cerebral AVMs

ization on cerebral haemodynamics in AVMs by
means of the SPECT technique, confirmed that there
is an acute flow redistribution immediately after
embolization of AVMs. This improvement of the
cerebral perfusion was found particularly marked in
patients with no history of haemorrhage, or when the
embolization had caused no new infarction in the surrounding normal brain, or when the nidus itself was
occluded (and particularly with liquid emboli such as
IBCA). After partial embolization, the improved
cerebral circulation did not last. The second question
is: are there specific haemodynamic complications
due to the embolization procedure itself? Immediate
and delayed complications of the embolization have
been reported, very similar to that described above in
our experience. Spetzler [32] reported several AVM
ruptures occurring several days to weeks after transfemoral embolization that converted a large AVM into
a smaller one. He explained this complication by the
AVM feeding artery pressure: following embolization
of a large AVM, the resistance to flow increases and
the feeding artery pressure in the remaining lesion is
elevated. Spetzler's conclusion is that the usefulness
of partial embolization as the sole treatment for
AVMs should be seriously questioned, and the goal of
treatment for AVMs should be complete obliteration.
Jafar [14] observed a transient ischaemic event in one
patient (5%) and a haemorrhagic complication in 3
(15%) requiring emergency surgery with subsequent
recovery: The haemorrhage occurred during the
embolization procedure in one case, and within 24 to
72 hours in two other cases. The third question is:
Does the embolization reduce the rate of complications occurring after the resection of AVMs? The fact
that embolization may favour the development of the
smaller deep feeding vessels, which could increase
the difficulty of surgery, seems well accepted [14,
22]. In our series, we observed no haemodynamic
complications after surgery, with or without embolization. Various opinions are expressed in the recent
literature. Morgan [22], discussing the complications
of surgery for AVMs, concluded that in his experience
no NPPB or intra-operative haemorrhage occurred
since the introduction of selective embolization.
Yound [45] had the same experience with a limited
group of patients who went to surgery after embolization. Jafar [14] compared a series of patients treated
by pre-operative embolization and surgery, and a
series of patients treated by surgery alone. In patients
operated on after embolization, he observed a 15%

R. Deruty et al.: Complicationsof Cerebgal AVMs
rate of postoperative haemorrhage (and overall a 3%
death rate). Nevertheless he concluded that transfemoral embolization with NBCA is a safe and beneficial adjunct therapy to the surgical resection of
AVMs. Embolization makes larger-sized, highergrade AVMs surgically equivalent to smaller-sized,
lower-grade lesions, by reducing operative time and
blood loss. Takeuchi [37] observed in his series one
case with hyperperfusion on SPECT two weeks after
embolization, together with a NPPB phenomenon
occurring at the end of the operation which was performed three weeks after embolization. He concluded
that the hyperaemic state due to embolization probably continues for two or three weeks after embolization. Awad [3] in a series of AVM patients treated by
staged embolization followed by surgical excision
concluded that this mode of treatment did not com~
pletely eliminate postoperative intracranial hypertension, which was observed in 28% of the cases, and
more frequently in AVMs greater than 6 cm in maximum diameter with no mortality.

Conclusion

The review of the literature demonstrates that the
complications occurring after treatment of cerebral
AVMs may be either not based on haemodynamics, or
based on haemodynamics [21]. Not based on haemodynamics are the resection of eloquent brain, the postretraction oedema and haemorrhage, the rupture of
occult retained AVM, and the new onset of seizures.
Based on haemodynamics are the NPPB phenomenon, the venous and arterial occlusion of vessels of
normal parenchyma, the retrograde thrombosis of
feeding arteries, and the venous overload or occlusive
hyperaemia phenomenon.
In our experience:
- Complications of resection (with or without preoperative embolization) remained minor and only due
to the resection or to the manipulation of the corresponding brain area. No haemodynamic complication
occurred in this group.
- Complications of radiosurgery (alone, or after
embolization, or after incomplete resection) remained
so far infrequent and of the minor type, relating to the
anatomical site of the AVM.
Complications of endovascular embolization
were the most frequent and the most severe ones,
responsible for all the deaths due to treatment, They
were all of haemodynamic origin, either with ischae-

129
mia in 50% of cases (no mortalities) or with haemorrhage in 50% of cases (with all the fatal cases). In
most of cases with haemorrhage the mechanism was
occlusion of the venous drainage. These complications of embolization occurred mainly in high-grade
AVMs, but also, to a lesser extent, in low-grade cases.
So, in our experience, the emphasis of the treatment
of AVMs, which once was devoted to surgical resection, has now shifted to endovascular embolization.
As far as high-grade AVMs are concerned, this shift
of treatment (surgery or embolization) is quite understandable and acceptable, for these malformations
remain the most difficult and the most dangerous to
treat, even with embolization, We will continue to
submit these cases to pre-operative or pre-irradiation
embolization. In any case, the patient and the family
should be warned that the embolization technique
remains a procedure of risk, contrary to the idea taking root in the mind of the population at large. But
several complications after embolization occurred
also in cases with low-grade AVMs, which are the
best cases for surgery (or in some cases for radiosurgery, depending on the greatest diameter). After this
experience, we tend to select these low-grade cases
more and more frequently for direct resection (or
radiosurgery) and to refrain from proposing embolization even if technically feasible. Even if the risk of
embolization in this sub-group is infrequent, the
severity of the complication outstrips tile risk of surgery or radiosurgery.
References

1. A1 Rhodan NRF, Sundt ThM, Jr, Piepgras DG, Nichols DA,
Rufenacht D, StevensLN (t993) Occlusivehyperemia: a theory for the hemodynamiccomplications following resection of
intracerebral AVMs. J Neurosurg 78:167-175
2. Ancri D, Pertuiset B (1985) Mesures des vitesses sanguines
instantan6es darts les art~res carotides internes par Doppler
puls6 darts les MAV cfr6brales. Neurochirurgie31 : 1-6
3. Awad JA, Magdinec M, Schubert A (1994) Intracranial hyperh
tension after resection of cerebral arteriovenous malformations. Predisposing factors and management strategy. Stroke
25:611-620
4. Barnett GH, Little JR, Ebrahim ZY, Jones SC, Friel HT (1987)
Cerebral circulationduring AVM operation. Neurosurgery20:
836-842
5. Batjer HM, Devous MD, Meyer YJ, Purdy PD, Samson DS
(1988) Cerebrovascularhemodynamics in AVM complicated
by normal perfusion pressure breakthrough. Neurosurgery 22:
503-509
6. Batjer HM, Devous MD, SR, Seibert GB, Purdy PD, Ajmani
AK, Delarosa M, Bonte J (1988) Intracranial AVMs. Relationship between clinical and radiographic factors and ipsilateral
steal severity. Neurosurgery23:322-328

130
7. Batjer HM, Purdy PD, Giller CA, Samson DS (1989) Evidence
of redistribution of cerebral blood flow during treatment for an
intracranial AVM. Neurosurgery 25:599-605
8. Batjer HM, Devous MD (1992)The use of acetazolamine.
Enhanced regional cerebral blood flow measurement to predict
risk to AVM patients. Neurosurgery 31: 213-218
9. Bormal J, Born JD, Hans P (1985) One-stage excision of highflow arterio-venous malformation. J Neurosurg 62:128-131
i0. Day A L , Friedman WA, Sypert GW, Mickl JP (1982) Successful treatment of the normal perfusion pressure breakthrough
syndrome. Neurosurgery 11 : 625-630
11. Deruty R, Pelissou-Guyotat I, Mottolese C, Bascoulergue Y,
Amat D (1993) The combined management of cerebral arteriovenous malformations (experience with 100 cases and review
of the literature). Acta Neurochir (Wien) 123:101 - 112
12. Hassler W, Steimetz H(1987)Cerebral hemodynamics in
angioma patients: an intra operative study. J Neurosurg 67:
822-831
13. Heros RC (1987)Comments on: cerebral circulation during
arteriovenous malformation operation, by Barnett GH, Little
Jr, Ebrahim ZY, Jones SC, Friel HT. Neurosurgery 20:841
14. Jafar JJ, Davis AJ, Berenstein A, Cboi IS, Kupersmith MJ
(1993) The effect of embolization with N-butylcyanoacrylate
prior to surgical resection of cerebral AVMs. J Neurosurg 78:
60-69
15. Leblanc R, Little JR (1988) Hemodynamics of arteriovenous
malformations. Clin Neurosurg 36:299-317
16. Lindegaard KF, Grolimund P, Aaslid R, Nornes H (1986)
Evaluation of cerebra/AVMs using transcranial Doppler ultrasound. J Neurosurg 65:335-344
17. Lo EH (1993) A haemodynamic analysis of intracranial arteriovenous malformations. Neurolog Res 15:51-55
18. Manchola JF, De Salles AF, Foo TK, Ackerman AH, Candia
GT, Kjelberg RN (1993) AVM hemodynamics: a transcranial
Doppler study. Neurosurgery 33:556-562
19. Miyasaka H, Yada K, Ohwada T, Kitahara T, Endoh M, Saito
M, Kurata A, Ohtaka H (1990) Retrograde thrombosis of feeding arteries after removal or arterial AVMs. J Neurosurg 72:
540-545
20. Miyasaka H, Yada K, Ohwada T, Kurata A, Tokiwa K, Suwa
T, Yamada M, Oka H (1991) Hemorrhagic venous infarction
after excision of an AVM. Neurosurgery 29: 265-26821. Morgan MK, Sundt ThM, Jr (1989) The case against staged
operative resection of cerebral AVMs. Ncurosurgery 25:
429-435
22. Morgan MK, Johnston IH, Hallinan JM, Weber NC (1993)
Complications of surgery of AVMs of the brain. J Neurosurg
78:176-182
23. Mullah S, Brown FD, Patronas NJ (1979)Hyperthermic and
ischemic problems of surgical treatment of arteriovenous malformations. J Neurosurg 51:757-764
24. Musaszko K, Wang HH, Pelton G, Stein BM (1990) A study of
the reactivity of feeding vessels to arteriovenous malformations: correlation with clinical outcome. Neurosurgery 26:
190-200
25. Nagao S, Ueta K, Mino S, Fuhiwara T, Yonma Y, Ito T, Ohmoto T, Ohkawa M (1989) Monitoring of cortical blood flow during excision of AVM by thermal diffusion method. Surg Neurol 32:137-143
26. Nornes H, Grip A (1980) Hemodynamic aspects of cerebral
arteriovenous malformations. J Neurosurg 53:456-464

R. Deruty et al.: Complications of Cerebral AVMs
27. Okabe T, Meyer JS, Okayasu H, Harper R, Rose J, Grossman
RG, Centeno R, Tachibana H, Lee YY (1983) Xenon-enhanced
CT CBF measurements in cerebral AVMs before and after
excision. Contribution to pathogenesis and treatment. J Neurosurg 59: 21-3l
28. Pertuiset B, Ancri D, Sichez JP, Chauvin M, Guilly E, Metzger
J, Gardeur D, Basset JY (1983)Radical surgery in cerebral
AVMs. Tactical procedure based upon haemodynamic factors.
In: Krayenbuchl H etal (eds)Advances and technical standards in neurosurgery, vol 10. Springer, Wien New York,
pp 81-146
29. Pertuiset B, Ancri D, Mahdi M, Nakano H, Arthuis F, BagnatGuilly E (1990) A new haemodynamic factor in cerberal
AVMs: aspiration from the venous system demonstrated in two
cases of pedonculo-galen AVM successfully cured by occlusion of the superior longitudinal sinus. Acta Neurochir (Wien)
104:136-142
30. Rosenblum BR, Bonner RF, Oldfield EH (1987) Intra operative measurement of cortical blood flow adjacent to cerebral
AVM using laser Doppler velocimetry. J Neurosurg 66:
396-399
31. Spetzler RF, Wilson CB, Weinstein PH, Mehdom M, Townsend JJ, Telles D (1977)Normal perfusion pressure breakthrough theory. Clin Neurosurg 25:651-672
32. Spetzler RF, Hargraves RW, Mc Cormick PW, Zabramski JM,
Flora RA, Zimmerman RS (1992) Relationship of perfusion
pressure and size to risk of hemorrhage from AVMs. J Neurosurg 76:918-923
33. Sugita M, Takahashi A, Ogawa A, Yoshimoto T (1993)
Improvement of CBF and clinical symptoms associated with
embolization of a large AVM. Neurosurgery 33:748-752
34. Sundt ThM, Jr, Piepgras DG, Stevens LN (1989) Surgery for
supratentorial AVMs. Clin Neurosurg 37:49-115
35. Tamaki N, Lin T, Asada M, Fujita K, Tominaga S, Kimura M,
Ehara K, Matsumoto S (1990) Modulation of blood flow following excision of a high-flow cerebral AVM. J Neurosurg 72:
509-512
36. Tamaki N, Ehara K, Fujita K, Shirakuni T, Asada M, Yamashita H (1993) Cerebral hyperfusion during surgical resection of
high-flow AVMs. Surg Neurol 40:10-15
37. Takeuchi S, Abe H, Nishimaki K, Minakawa T, Koike T,
Kameyama S, Tanaka R (1994) Cerebral haemodynamic
changes after endovascular treatment of arteriovenous malformations: evaluation by Single-Photon. Emission CT. Acta
Neurochir (Wien) 127:142-150
38. Tanaka K, Yonekawa Y, Kaku Y, Kazekawa K (1993) Arteriovenous malformations and diaschisis. Acta Neurochir (Wien)
120:26-32
39. Taneda M, Mayakawa T (1993) The paradoxical blood pressure-flow relationship in the brain with an AVM. Surg Neurol
40:390-394
40. Wilson CB, Hieshima G (1993) Occlusive hyperemia: a new
way to think about an old problem. J Neurosurg 78:165-166
41. Yamada S (1982) Arteriovenous malformations in the functional areas: surgical treatment and regional cerebral blood
flow. Neurol Res 4:283-322
42. Young WL, Solomon RA, Prohovnik I, Ornstein E, Weinstein
J, Stein BJ(1988)133Xe Blood monitoring during AVM
resection. A case of intra operative hyperperfusion with subsequent brain swelling. Neurosurgery 22:765-769

R. Deruty et al.: Complications of Ceret ..... AVMs
43. Young WL, Prohovnik I, Ornstein E, Ostapkovich N, Sisti MB,
Solomon RA, Stein BM (1990). The effect of AVM resection
on cerebrovascular reactivity to carbondioxide. Neurosurgery
27:257-267
44. Young WL, Kader E, Prohovnik I, Omstein E, Fleischer LH,
Ostapkovitch N, Jackson LD, Stein BM (1993) Pressure autoregulation is intact after AVM resection. Neurosurgery 32:
491-497
45. Young WL, Pile-Spellman J, Prohovnik I, Kader A, Stein BM
(1994) Evidence for adaptive autoregulatory displacement in

131
hypotensive cortical territories adjacent to arteriovenous malformations. Neurosurgery 34:601-611
46. Zabramski JM, Spetzler RF (1989)Comments on: the case
against staged operative resection of cerebral arteriovenous
malformations. In: Morgan MK, Sundt ThM Jr (eds) Neurosurgery 25:436

Correspondence: R. Deruty, M.D., Hopital Neurologique, 59
boulevard Pinel, F-69394 Lyon, France.