CLINICAL STUDIES

ROLE OF RADIOSURGERY IN THE MANAGEMENT

OF CEREBRAL ARTERIOVENOUS MALFORMATIONS
IN THE PEDIATRIC AGE GROUP: DATA FROM A
100-PATIENT SERIES
Nicolas Reyns, M.D. OBJECTIVE: To assess the safety and efficacy of radiosurgery for the management of
Department of Neurosurgery, arteriovenous malformations (AVMs) in the pediatric age group.
University Hospital,
Lille, France METHODS: We reviewed data from 100 children (44 girls and 56 boys) presenting a total
of 103 AVMs treated by linear accelerator radiosurgery between December 1988 and
Serge Blond, M.D. May 2002. The median patient age was 12 years (range, 2–16 yr). Sixty-seven AVMs
Department of Neurosurgery, (65%) were in functional locations and 30% were inoperable. The mean AVM volume
University Hospital, was 2.8 cm3 (range, 0.9–21.3 cm3). The mean marginal dose was 23 Gy (range, 15–25
Lille, France
Gy) and required between one and four isocenters. Fifty patients received multimodal
treatments with embolization and/or surgery before and/or after radiosurgery. Given
Jean-Yves Gauvrit, M.D.
that 16 patients underwent two sessions of radiosurgery and one patient received three
Department of Neuroradiology,
University Hospital, sessions, a total of 119 radiosurgical treatments were delivered. We maintained our
Lille, France clinical and angiographic follow-up for at least 36 months after irradiation or until the
complete obliteration of the AVM was confirmed by angiography (our sole end point for
Gustavo Touzet, M.D. judging clinical efficacy). Univariate and multivariate analysis were performed to deter-
Department of Neurosurgery, mine predictive factors for obliteration.
University Hospital,
Lille, France RESULTS: Complete obliteration was achieved for 72 AVMs (70%). The permanent
neurological deficit rate was 5%. One patient died because of rebleeding. None of our
Bernard Coche, M.D. patients presented bleeding after an angiographically verified AVM obliteration. The
Department of Radiation Oncology, main predictive factors for obliteration were low AVM volume and no previous
Centre Oscar Lambret,
embolization. Moreover, the younger the patient, the more effective the radiosurgery
Lille, France
seemed to be.
Jean-Pierre Pruvo, M.D. CONCLUSION: Radiosurgery is a safe and effective treatment for AVMs in the pedi-
Department of Neuroradiology, atric age group. One criterion for success was the use of a prescription dose similar to
University Hospital, that used with adult populations.
Lille, France
KEY WORDS: Arteriovenous malformations, Children, Linear accelerator radiosurgery, Pediatric,
Patrick Dhellemmes, M.D. Radiosurgery
Department of Neurosurgery,
Neurosurgery 60:268–276, 2007 DOI: 10.1227/01.NEU.0000249277.72063.BD www.neurosurgery-online.com
University Hospital,
Lille, France

Reprint requests:

I
Nicolas Reyns, M.D.,
n view of the high hemorrhage rate associ- ding to a grim prognosis (15). Hence, a 13-year-
Department of Neurosurgery, ated with their natural history, pediatric old child with a newly detected AVM has a 33%
University Hospital, cerebral arteriovenous malformations cumulative risk of hemorrhage in the next
University of Lille, (AVMs) require curative treatment (18, 35). 20 years and a 55% risk in 40 years (24). Of
Avenue Emile Laine,
Indeed, the risk of hemorrhage for untreated course, the younger the patient, the more sub-
59037 Lille Cedex, France.
E-mail: n-reyns@chru-lille.fr cerebral AVMs has been estimated at 2 to 4% stantial the cumulative risk of hemorrhage,
per year, with an attendant mortality rate of 5 thus, increasing the potential benefits of early
Received, August 16, 2005. to 10% and a 50% risk of serious neurological intervention in children (43). At present, micro-
Accepted, September 18, 2006. morbidity associated with each hemorrhage surgery is still the “gold standard” treatment
(15, 48). Moreover, it seems that rebleeding may for accessible, low-grade AVMs, providing an
occur in almost 30% of these cases, correspon- immediate cure with generally accepted risks

268 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com

 RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

(21, 23, 43). Moreover, immediate surgical management is indi-

cated for patients who present with intracerebral hemorrhage TABLE 1. Additional treatment modalities in 50 patientsa
and associated progressive neurological deficits. It is only in Embolization Embolization
the past decade that the role of radiosurgery for treating AVMs before before and No
in children has been more precisely defined; previously, there RS after RS embolization
was little information available regarding the use of this tech-
No surgery 38 2
nique in a pediatric population (21). However, encouraging
Surgery before RS 2 6
results in carefully selected patients have been reported in the
Surgery before and 1
literature during the past few years (1, 27, 29, 32, 33, 43, 45, 46).
after RS
Nowadays, there is no real reason to avoid radiosurgery as an
Surgery after RS 1
initial treatment for an inoperable nidus or after partial
a
embolization or partial surgical resection. Indeed, if the treat- RS, radiosurgery.
ment goal is to preserve life and neurological function while
curing the AVM and preserving normal cerebral circulation (21),
the potential use of radiosurgery should be considered for clin- TABLE 2. Clinical summary of 100 children with arteriovenous
ically stable patients who do not require emergency treatment. malformations (n ⴝ 103)a
Initial symptom No. of AVMs
PATIENTS AND METHODS Hemorrhage 69 (67%)
Seizures 10 (9.7%)
Patient Selection Headaches 11 (10.7%)
Between 1988 and 2002, 1050 patients presenting with cere- Asymptomatic 8 (7.7%)
bral AVMs were treated by linear accelerator radiosurgery in Others 5 (4.9%)
our institution. Of these, 100 individuals (9.5%) were younger a
AVMs, arteriovenous malformations.
than 16 years at the time of initial treatment (44 girls and
56 boys). All of these patients were included in the presently
reported series. At least 36 months of clinical and imaging location and angioarchitecture, afferent vessels and draining
follow-up data were available for this population, except when veins, and for grading all AVMs according to the Spetzler-Martin
earlier obliteration was observed. Twenty-six patients received system. AVM locations are presented in Table 3. AVMs were clas-
first-line care in our institution, with a multidisciplinary team sified as Spetzler-Martin Grade I (9.7%), Grade II (34%), Grade III
of neurosurgeons and interventional radiologists deciding (22, 3%), Grade IV (4, 9%), Grade V (0%), or Grade VI (29.1%
whether or not to apply previous embolization, surgery, inoperable, located in the brainstem and basal ganglia) (Table 4).
and/or radiosurgery as the therapeutic approach. Seventy-four In 67 patients, the AVM was in a functional location.
patients were referred to us for radiosurgical treatment by
Radiosurgical Technique
physicians in other institutions (either as the sole treatment or
after partial surgical resection or previous embolization). In Radiosurgical Device
fact, 50 patients were treated with multimodal strategies All radiosurgical procedures in this study were performed
(Table 1). However, all results in this series relate only to radio- according to Betti’s technique (4), with a Saturne 18-MeV linear
surgery performed on 103 residual nidi because the four accelerator (CGR, Paris, France) fitted with a stereotactic add-
patients who were treated by embolization and or surgery after on device for radiosurgery and a choice of additional collima-
radiosurgery were considered treatment failure vis-à-vis radio- tors (8, 16, 20 mm) for delivering x-ray minibeams. The move-
surgery. The median age at the time of treatment was 12 years ments of the stereotactic frame relative to those of the
(range, 2–16 yrs). The initial symptom was intracranial hemor- accelerator create a spherical sector, which represents the sur-
rhage in 69 out of 100 patients—a far higher proportion of face of possible beam entry portals, extending 160 degrees in
spontaneous hemorrhagic modes than in otherwise comparable the coronal plane and 120 degrees in the sagittal plane. Patients
adult populations. Ten children presented with seizures, 11 pre- were seated in a Betti armchair, with the head maintained in a
sented with headache, five presented with various other symp- Talairach frame (4).
toms (cardiac failure in two patients due to thalamic AVMs
with large draining vein of Galen, Rendu-Osler disease in two Treatment Procedure
patients, and macrocrania in one patient), and five were asymp- Attachment of the Talairach stereotactic frame and stereotactic
tomatic (two of the latter presented several AVMs in cases of teleangiography were performed with the patient under general
Rendu-Osler disease) (Table 2). anesthesia. Four transcranial, smooth-tipped screws were used to
attach the frame. Once the frame had been correctly positioned
Characteristics of the AVMs and adjusted, it was removed and the patient was discharged
For the purpose of this series, all pretreatment angiograms until the day of the radiosurgical operation. On readmission, the
were reviewed by a radiologist (JYG) for determining nidus frame was repositioned (with the same stereotactic coordinates)

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 269

REYNS ET AL.

was recorded) via clinical evaluation, MRI scans, and annual

TABLE 3. Anatomic location of arteriovenous malformations angiography. MRI scanning was performed at 6 months after
(n ⴝ 103)a irradiation and annually thereafter. Referring physicians were
Location No. of AVMs also contacted by mail to optimize patient follow-up.
The sole efficacy end point for complete obliteration was the
Frontal lobe 16 (15.5%)
angiographic result according to Linquist and Steiner’s criteria
Parietal lobe 12 (11.65%)
(normal circulation, no nidus, and no draining vein) (44).
Occipital lobe 8 (7.8%)
Temporal lobe 11 (10.7%) Statistical Analysis
Rolandic 15 (14.6%)
Univariate analysis of the factors affecting nidus obliteration
Thalamus/basal ganglia 18/7 (17.5/7%)
was performed. For categorical variables, a χ2 test or a Fischer
Intraventricular 1 (0.9%)
exact test was used, depending on the number of variables per
Corpus callosum 2 (1.9%)
class. For quantitative variables, a nonparametric Mann-Whitney
Peduncular and pons 5 (4.9%)
test or a t test was performed, again depending on the number
Cerebellum 8 (7.8%)
of variables per class. We studied the following parameters: age,
Functional location 67 (65%)
sex, nidus location and depth, nidus characteristics (compact,
a
AVMs, arteriovenous malformations. plexiform, and presence or absence of fistulae), previous
embolization, afferent arterial vessels, type of venous drainage,
size and volume of the marginal isodose, and dose prescribed to
marginal isodose. A similar statistical methodology was applied
TABLE 4. Correlation between Spetzler-Martin Grade and
to factors that could have been correlated with radiosurgery side
obliteration ratea
effects. Multivariate analysis was performed using logistic
Grade No. of AVMs Obliteration rate regression. The statistical significance threshold was set at 0.05.
I 10 (9.7%) 7 (70%)
II 35 (33.9%) 33 (94%)
III 23 (22.4%) 15 (65%)
RESULTS
IV 5 (4.9%) 2 (40%)
Neuroimaging Follow-up and Obliteration Rate
V 0
VI 30 (29.1%) 14 (46.5%) The median time from first-stage radiosurgery to the last imag-
a
ing follow-up examination was 26 months (range, 11–126 mo).
AVMs, arteriovenous malformations.
Sixty-four patients with 67 AVMs (out of 100 patients with 103
AVMs) were cured (i.e., 65%), with a mean time to obliteration of
with the patient under general anesthesia, and the patient was 25.5 months. Sixteen patients underwent second-stage radio-
then transferred to the irradiation site. The stereotactic frame was surgery, with five (31%) of these exhibiting complete obliteration
removed immediately after the radiosurgical procedure and the (mean time delay, 18.4 mo after the second intervention) at the
patient was usually discharged on the following day. end of a median follow-up period of 25 months (range, 12–51
mo). Two children underwent a third session of radiosurgical
Target Delineation and Treatment Planning treatment; one patient was lost to follow-up and the other was
All targets were delineated by reference to biplanar anteropos- recently treated in our institution using gamma knife radio-
terior and lateral teleangiographic data. The goal of the targeting surgery. Obviously, this procedure was not included in the pres-
was to match the marginal isodose (70%) to the nidus (while ent study.
sparing afferent vessels and draining veins), thus obtaining con- At the time of the last follow-up examination, 69 patients
formal coverage of the entire AVM nidus and the proximal sec- with 72 AVMs exhibited angiographically confirmed, complete
tions of the draining veins. Dynamic angiographic imaging pro- AVM obliteration, corresponding to an overall obliteration rate
vided us with identification on the nidus borders. The exact of 70% (Table 5).
location of the AVM and its relationships with functional struc- Obliteration rates as a function of target volume are pre-
tures were documented using magnetic resonance imaging sented in Table 6. In Group 1 (AVM volume ⱕ1 cm3), 30 (85.7%)
(MRI) scanning. Most targets required a single isocenter treat- out of the 35 AVMs were obliterated. In Group 2 (AVM volume
ment procedure. The dose delivered to the marginal isodose was ⬎1 to ⱕ3 cm3), 25 (69.5%) out of the 36 AVMs were obliterated.
adjusted according to the volume enclosed by the prescription In Group 3 (AVM volume ⬎3 to ⱕ10 cm3) 17 (61%) out of the
isodose and the distance to any nearby functional structures. 28 AVMs were obliterated. Lastly, in Group 4 (AVM volume
The mean marginal dose was 23 Gy (range, 15–25 Gy). ⬎10 cm3), none of the four AVMs was cured (0%).
Obliteration rates as a function of Spetzler-Martin grade are
Follow-up Period presented in Table 4. Angiographically documented oblitera-
All patients underwent at least 36 months of follow-up after tion was observed for seven (70%) out of the 10 Grade I AVMs,
radiosurgery (unless earlier, definitive evidence of obliteration 33 (94%) out of the 35 Grade II AVMs, 15 (65%) out of the 23

270 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com

 RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

grade (P ⫽ 0.001), on the other. Obliteration was not correlated

TABLE 5. Overall obliteration rate after first and second stage of with the number of isocenters (P ⫽ 0.3), sex (P ⫽ 0.19), type of
radiosurgerya venous drainage (P ⫽ 0.34), nidus characteristics (P ⫽ 0.23),
No. of AVMs Obliteration rate nidus deepness (P ⫽ 0.748), or dose to marginal isodose
(P ⫽ 0.77). Postradiosurgery morbidity was correlated with the
First stage 103 67 (65%)
male sex (P ⫽ 0.009), left hemisphere location (P ⫽ 0.04), and
Second stage 16 5 (31%)
nidus volume (P ⫽ 0.014). Morbidity was not correlated with
Total 103 72 (70%)
either AVM location or previous radiosurgery.
a
AVMs, arteriovenous malformations. A multivariate analysis did not reveal any significant associa-
tions between parameter combinations and the obliteration rate.

TABLE 6. Correlation between size and obliteration ratea DISCUSSION

3
Volume (cm ) No. of AVMs Obliteration rate
Multimodal Strategies
⬍1 35 30 (85.7%)
The choice between microsurgery, radiosurgery, or emboliza-
1–3 36 25 (69.5%)
tion as the first-line treatment for each child treated initially in
3–10 28 17 (61%)
our institution was based on a multidisciplinary approach. In
⬎10 4 0
view of our experience and that of other groups (20), it seems
a
AVMs, arteriovenous malformations. that pediatric AVM management strategies often require a com-
bination of several techniques, as has been observed in adults (9,
10). Hence, 50 patients in our series were treated using a combi-
Grade III AVMs, two (40%) out of the four Grade IV AVMs, and nation of surgery and embolization before and/or after radio-
14 (46.5%) out of the 30 Grade VI AVMs. Of the 67 AVMs in surgery (Table 1). Surgery was attempted as the definitive treat-
functional locations, 45 were obliterated (67%). ment whenever the size and location of the AVM enabled
surgical resection under safe conditions. Radiosurgery was
Functional Outcomes and Complications selected as the first-line treatment whenever the nidus was
Ninety-five patients returned to their previous activity level deeply located or close to eloquent areas or in the event of a
after radiosurgery. There were no treatment-related deaths. residual nidus after primary surgical resection and/or emboliza-
Seven patients experienced complications with a time lag after tion. A cure is seldom obtained with embolization alone (14, 25,
surgery (Table 7). Six of these exhibited a permanent neurolog- 32, 49), even though some authors consider that the hemody-
ical deficit at the end of the follow-up period: two cases of namic modifications after embolization and the obliteration of
slight motor palsy (a Grade II rolandic AVM and a Grade VI dangerous portions of the AVM are sufficient to protect patients
thalamic AVM), one case of monocular blindness (a frontal against bleeding (25). In our population, embolization was reg-
Grade III AVM close to the optic nerve), one case of hemianop- ularly performed 1) to permanently decrease the AVM volume
sia (a Grade III temporomesial AVM), and two cases of diplopia and flow and 2) to allow more effective radiosurgery (17, 31, 34,
(a Grade VI thalamic AVM and a Grade VI tectal plate AVM). 47); with smaller AVMs, a higher and more effective radiation
Two patients experienced radiation-induced epilepsy, one of dose can be administered with a lower risk of side-effects.
whom (a patient with a Grade III parietal lobe AVM) resisted However, in most instances, embolization achieves a reduction
complete medical control. Hence, in view of the total number of in flow but not size. Nevertheless, we think flow reduction is
radiosurgical procedures (n=119), the overall morbidity rate useful for boosting the efficiency of radiosurgery (5) even when
was 6.7% (eight out of 119) and the permanent neurological the radiosurgical target volume remains unchanged, but we do
deficit rate was 5% (six out of 119). Two patients (both with not have sufficient data to prove this assertion.
basal ganglia AVMs) experienced delayed intracranial hemor-
rhaging. One died as a result (16 mo after radiosurgery) and the Follow-up and Obliteration Criteria
other experienced repeated intracranial hemorrhaging in the Because of its delayed efficiency, radiosurgery requires as
absence of neurological deficits during the 30-month postrad- rigorous a follow-up as possible. Moreover, in view of a child’s
iosurgery latency interval. None of our patients with angio- long life expectancy, this follow-up should continue for as long
graphically defined AVM obliteration experienced any bleeding as possible because of potential AVM regrowth, even in cases of
whatsoever. angiographically documented obliteration (28, 42). Indeed, sev-
eral authors involved in pediatric surgery have described
Statistical Analysis recurrent AVM rebleeding after negative postoperative arteri-
Univariate analysis demonstrated a correlation between the ography (3, 21, 22, 51). It is postulated that AVM regrowth may
obliteration rate on one hand, and nidus volume (P ⫽ 0.001) occur in children and could be a consequence of their relatively
and size (P ⫽ 0.001), no previous embolization (P ⫽0.001), age immature cerebral vasculature (21, 41). This potential for
younger than 10 years (P ⫽ 0.05), and low Spetzler-Martin regrowth seems to persist even after angiographically proven

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 271

REYNS ET AL.

TABLE 7. Review of literature on six major and recent pediatric series regarding radiosurgery for treatment of arteriovenous
malformationsa
OR, % (no.)
No. of Median/ Median/ Overall
Median/ Second Bleeding Permanent
Series patients mean mean ⬍1 cm3
mean stage after RS, sequelae,
(ref. no.) with volume marginal 1–3 cm3
age radiosurgery no. (%) no. (%)
follow-up (cm3) dose (Gy) 3–10 cm3
⬎10 cm3

Tanaka et
11.5c
al., 1996 23 4.8c 20.5c 0 87 (20/23) 0 0
(2–15)
(46) LGK

Shin et
15b
al., 2002 82 1.8b 20a 0 86.6 (71/82) 4 (4) 2 (2)
(4–19)
(42) LGK

73.5 (39/53)
Levy et al., na
12b 10
2000 (27) 53 1.7b 20b 80 4 (7.5) 1 (2)
(2–17) (18.9%)
LGK 64.7
0

35 (11/31)
Smyth et na
11.2b 1.6b/ 12
al., 2002 31 16.7c 53 8 (25) 2 (6)
(3.4–17.5) 5.4c (38.7%)
(43) LGK 14
0

61.2 (30/49)
Nataf et al., b c b
100
12 /11 3.5 /
2003 (32) 49 28.3c 0 73 (1–4 cm3) 4 (8.2) 0
(7–15) 3c
LINAC 40 (4–10 cm3)
100% (1 pt)

Nicolato et
11.7c
al., 2005 47 3.8c 21.6c 0 80 (31/39) 0 (0) 1 (2)
(5–16)
(33) LGK

70 (72/103)
85.7
This report 12b 1.7b/ 16
100 23.3c 69.5 2 (1.7) 6 (5)
LINAC (2–16) 2.8c (16%)
61
0
a
OR, obliteration rate; RS, radiosurgery; LGK, Leksell gamma knife; na, not available; LINAC, linear accelerator-based radiosurgery
b
Median.
c
Mean.

resection or obliteration (3, 21, 22, 28). Hence, a number of nidus. Moreover, one of the three patients presenting with
these authors recommend performing an angiogram long after Rendu-Osler disease declared a new AVM location after radio-
a child’s AVM has been resected (3, 22). In our opinion, this rec- surgical obliteration of three AVMs. These observations illus-
ommendation is equally applicable to the follow-up of AVMs trate the possible angiogenesis phenomena evoked in cases of
cured by radiosurgery, although we did not notice the phe- AVM recurrence in pediatric populations. Regarding the
nomenon of nidus regrowth after obliteration in the present follow-up, we (similar to others [13, 32]) are convinced of the
series. However, in one case, we did notice a new nidus com- absolute necessity for definitive, angiographic validation of
ponent appearing outside the marginal isodose of a radiosurgi- AVM obliteration. At present, MRI scanning is not sensitive
cal procedure that had successfully obliterated the original enough (even with new techniques, such as magnetic reso-

272 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com

 RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

volume is a fully independent,

A B C predictive factor for oblitera-
tion. In relation to this, Levy
et al. (27) found that volume
was significantly correlated
with obliteration rate only
when multivariate logistic
regression was applied. These
a u t h o r s re p o r t e d a m u c h
higher obliteration rate for an
AV M s e r i e s w i t h a s m a l l
E median volume of 1.7 cm 3 .
Hence, differences in oblitera-
F tion rates reported for litera-
ture series do not seem to be
correlated with the mean or
D median AVM volume (Table 7).
However, the doses applied
are clearly dissimilar. In an
attempt to minimize potential
radiation toxicity in their pa-
tients’ developing brains,
FIGURE 1. A, preradiosurgery, T2- Smyth et al. (43) adopted a
weighted, cerebral MRI scan for a conservative dose prescription
14-year-old girl, showing a Rolandic policy and reduced the periph-
AVM that had manifested itself eral dose, with a relatively low
through seizures. B and C, lateral obliteration rate of 35%. These
and anterior angiograms documenting the angioarchitecture of this AVM. D, authors reported a significant
13 months after radiosurgery, the AVM seemed to have been cured: this T2-
correlation between dose pre-
weighted cerebral MRI scan shows the absence of a draining vein and no resid-
ual nidus. E, lateral angiogram performed 13 months after radiosurgery show-
scription and obliteration rate
ing an early-stage Rolandic draining vein. F, anterior angiogram showing a (P ⫽ 0.025) but only observed
tiny nidus at the beginning of the Rolandic draining vein. This residual AVM a (nonsignificant) trend toward
was treated by a second radiosurgery procedure. a correlation between AVM
size and obliteration rate.
nance angiography and digital subtraction angiography) to However, some (larger) lesions received adequate prescription
detect small nidi or residual microshunts with bleeding poten- doses, whereas others did not, possibly explaining why a sta-
tial (Fig. 1). Unfortunately, the overall obliteration rates tistically significant correlation between response and AVM
reported for several literature series include patients with size was not observed. Smyth et al. (43) also noted that when a
AVMs considered to be obliterated on the basis of MRI scan- marginal dose of at least 18 Gy was delivered to the target vol-
ning results (27, 38, 43). We think such a procedure could over- ume, obliteration rates were nearly 10-fold higher than those
estimate the true obliteration rate. At present, we use magnetic for patients receiving lower doses, suggesting that this dose is
resonance angiography digital subtraction angiography as a at or near the threshold for AVM involution, regardless of AVM
routine examination during intermediate follow-up until an volume. These authors suggested that the poor obliteration rate
AVM seems to have been obliterated (26). The apparent oblit- in larger AVMs may not be related so much to nidus size but
eration is always confirmed by angiography. Moreover, follow- rather to the limitation in dosing a larger nidus volume within
up should also assess the neuropsychological behavior of chil- a larger volume of surrounding brain tissue while also striving
dren who received ionizing radiation. To date, there are few to reduce the risk of radiation-induced central nervous system
data available on the long-term neuropsychological outcome toxicity. They (and Yamamoto et al. [50]) argued that the vessel-
for pediatric radiosurgery patients (40). level radiobiological response to treatment is likely to be the
same in large and small AVMs and that a favorable response to
Predictive Factors for Obliteration radiosurgery should be expected as long as a sufficient dose is
The large number of patients in the present series offers the reached, regardless of overall AVM size. Unfortunately, our
advantage of statistical significance. One of the major predic- methodology did not permit us to determine potentially signif-
tive factors for obliteration rate was low AVM volume. icant correlations between dose prescription and obliteration
Obviously, this is not a new finding (6, 8, 12, 13, 27, 32, 33, 37, rate because we treated almost all patients identically. Indeed,
42), although our statistical analysis does suggest that AVM we adopted the same prescription dose policy in children as in

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 273

REYNS ET AL.

adults (as do other researchers [2, 27, 32, 46]) by applying a Furthermore, it is important to compare our obliteration and
marginal dose of 21 to 25 Gy for 95% of patients (on the 70% morbidity data with those from other pediatric AVM radio-
isodose in our study, according to the linear accelerator’s dose surgery series (Table 7) rather than surgical series. Indeed, pop-
distribution). Only a few patients were treated with lower ulations having undergone either radiosurgery or conventional
doses, based on their AVM volume and/or location. With the surgery should not be compared because radiosurgery is
exception of Smyth et al. (43), the aforementioned authors mainly suitable for inoperable or deeply located AVMs or those
report obliteration rates ranging from 61 to 95% (Table 7). In our located in functional areas. We wish to emphasize that approx-
opinion, these data constitute a strong argument for a correla- imately 70% of patients in this series presented an AVM in a
tion between dose prescription and obliteration rate. Hence, functional location and that 30% of the population had a
Smyth et al. (43) have begun to use adult dose prescriptions in Spetzler-Martin Grade VI AVM. Only 5% of the patients pre-
treating similar AVMs in children to increase the chances of sented with a permanent neurological deficit. Hence, the
obliteration. Spetzler-Martin grading system clearly seems to be an unsuit-
Another interesting predictive factor of obliteration seems able prognostic grading system for radiosurgery (36). We are
to be young age. In our univariate analysis, an age of 10 years convinced that a child with an AVM should be treated by sur-
or younger was significantly associated with a better oblitera- gery whenever the nidus can be resected under safe condi-
tion rate (P ⫽ 0.05). Indeed, the significance value was even tions, in view of the immediate protection against bleeding that
lower for an age of 8 years or younger (P ⫽ 0.03). These find- surgery offers (5, 11, 19, 21). On one hand, the postradiosurgery
ings confirm previous studies indicating young age as a predic- risk of hemorrhage during the latency interval persists until
tive factor for obliteration (37, 42). obliteration is achieved. On the other hand, the less-invasive
Our study highlights embolization as a negative predictive nature of the radiosurgical approach (combined with the
factor for obliteration. This had already been reported in an brevity or absence of hospitalization) contributes to the
adult population (37). Obviously, staged embolizations have a patient’s successful recovery from physical, mental, and emo-
potentially valuable preradiosurgery role in managing large tional standpoints (16).
lesions (17, 25, 31, 34, 47, 49) as long as complete elimination of
a peripheral portion of the nidus is achieved (17). However, in CONCLUSION
our experience, fragmenting of a residual nidus by emboliza-
tion could make a potential radiosurgical target in a heteroge- The present study confirms that radiosurgery is a safe and
neous environment less favorable for radiosurgery. This could effective treatment for AVMs in the pediatric age group. One
be one explanation for the fact that preradiosurgery emboliza- criterion for success was the use of a prescription dose similar
tion was reported in the present study as a negative predictive to that used with adult populations. Nidus volume seems to be
factor for radiosurgical AVM obliteration. Another explanation the other major variable affecting successful obliteration.
might be partial, postradiosurgery revascularization of the pre- Moreover, the younger the patient, the more effective the radio-
embolized portion (17, 39). This would mean that radiosurgery surgery seemed to be. There are now good evidence-based rea-
should be performed alone (i.e., in the absence of embolization) sons for considering this technique as an alternative to con-
whenever the nidus volume permits. ventional surgery in cases of deeply located AVMs or those
Our statistical analysis did not confirm single draining veins close to functional areas. Radiosurgery can be effective as a
(6, 37), patient sex (32), or nidus location as predictive factors sole treatment or as a part of a multimodal strategy when com-
for obliteration (30, 37). bined with other techniques. Hence, as in adult populations,
the key point in managing pediatric AVMs seems to be the use
Outcome and Morbidity of a multidisciplinary approach for deciding on the correct
technique for the patient. We are now extending our experience
With an overall obliteration rate of approximately 70% and a with gamma knife radiosurgery.
permanent neurological deficit rate of 5%, the present series
confirms the safety and efficacy of radiosurgery for the man-
REFERENCES
agement of AVMs in the pediatric age group. It could be argued
that concern for potential late consequences of high doses of 1. Altschuler EM, Lunsford LD, Coffey RJ, Bissonette DJ, Flickinger JC: Gamma
ionizing radiation can explain the general reluctance to use this knife radiosurgery for intracranial arteriovenous malformations in childhood
technique on young patients. Nevertheless, there are increasing and adolescence. Pediatr Neurosci 15:53–61, 1989.
2. Amendola BE, Wolf A, Coy SR: Radiosurgery for intracranial arteriovenous
numbers of well-documented pediatric series (2, 7, 27, 32, 42, malformations in children. J Radiosurg 2:159–164, 2000.
45, 46). To date, there is no real argument against using radio- 3. Andaluz N, Myseros JS, Sathi S, Crone KR, Tew JM Jr: Recurrence of cerebral
surgery as a first-line treatment for inoperable nidi or as a com- arteriovenous malformations in children: Report of two cases and review of
plementary treatment for a residual nidus after embolization or the literature. Surg Neurol 62:324–331, 2004.
surgery in a pediatric population. In particular, we are not 4. Betti OO, Munari C, Rosler R: Stereotactic radiosurgery with the linear accel-
erator: Treatment of arteriovenous malformations. Neurosurgery 24:311–321,
aware of any reported cases in the literature demonstrating 1989.
that stereotactic radiosurgery for the treatment of pediatric 5. Blond S, Coche-Dequeant B, Castelain B: Stereotactically guided radiosurgery
AVMs is a causative agent for new tumor formation. using the linear accelerator. Acta Neurochir (Wien) 124:40–43, 1993.

274 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com

 RADIOSURGERY FOR PEDIATRIC ARTERIOVENOUS MALFORMATIONS

6. Chang JH, Chang JW, Park YG, Chung SS: Factors related to complete occlu- previously proven to be obliterated after gamma knife radiosurgery.
sion of arteriovenous malformations after gamma knife radiosurgery. J Neurosurgery 46:803–810, 2000.
Neurosurg 93 [Suppl 3]:96–101, 2000. 29. Maity A, Shu HK, Tan JE, Ruffer J, Sutton LN, Tochner Z, Lustig R: Treatment
7. Colombo F: Linear accelerator radiosurgery of cerebral arteriovenous malfor- of pediatric intracranial arteriovenous malformations with linear-accelerator-
mations in the pediatric age group, in Raimondi AJ, Choux M, Di Rocco C based stereotactic radiosurgery: The University of Pennsylvania experience.
(eds): Cerebrovascular Diseases in Children. New York, Springer, 1991, pp Pediatr Neurosurg 40:207–214, 2004.
159–167. 30. Meder JF, Oppenheim C, Blustajn J, Nataf F, Merienne L, Lefkoupolos D,
8. Colombo F, Benedetti A, Pozza F, Marchetti C, Chierego G: Linear accelera- Laurent A, Merland JJ, Schlienger M, Fredy D: Cerebral arteriovenous malfor-
tor radiosurgery of cerebral arteriovenous malformations. Neurosurgery mations: The value of radiologic parameters in predicting response to radio-
24:833–840, 1989. surgery. AJNR Am J Neuroradiol 18:1473–1483, 1997.
9. Deruty R, Pelissou-Guyotat I, Morel C, Bascoulergue Y, Turjman F: Reflections 31. Millar C, Bissonnette B, Humphreys RP: Cerebral arteriovenous malforma-
on the management of cerebral arteriovenous malformations. Surg Neurol tions in children. Can J Anaesth 41:321–331, 1994.
50:245–256, 1998. 32. Nataf F, Schlienger M, Lefkopoulos D, Merienne L, Ghossoub M, Foulquier
10. Deruty R, Pelissou-Guyotat I, Mottolese C, Bascoulergue Y, Amat D: The JN, Deniaud-Alexandre E, Mammar H, Meder JF, Turak B, Huart J, Touboul
combined management of cerebral arteriovenous malformations: Experience E, Roux FX: Radiosurgery of cerebral arteriovenous malformations in chil-
with 100 cases and review of the literature. Acta Neurochir (Wien) dren: A series of 57 cases. Int J Radiat Oncol Biol Phys 57:184–195, 2003.
123:101–112, 1993. 33. Nicolato A, Foroni R, Seghedoni A, Martines V, Lupidi F, Zampieri P, Sandri
11. Di Rocco C, Tamburrini G, Rollo M: Cerebral arteriovenous malformations in MF, Ricci U, Mazza C, Beltramello A, Gerosa M, Bricolo A: Leksell gamma
children. Acta Neurochir (Wien) 142:145–158, 2000. knife radiosurgery for cerebral arteriovenous malformations in pediatric
12. Ellis TL, Friedman WA, Bova FJ, Kubilis PS, Buatti JM: Analysis of treatment patients. Childs Nerv Syst 21:301–308, 2005.
failure after radiosurgery for arteriovenous malformations. J Neurosurg 34. Perini S, Zampieri P, Rosta L, Piovan E, Micheli ED, Nicolato A, Mazza C, Benati
89:104–110, 1998. A, Ruatti P: Endovascular treatment of pial AVMs: Technical options, indications
13. Friedman WA, Bova FJ: Linear accelerator radiosurgery for arteriovenous and limits in pediatric age patients. J Neurosurg Sci 41:325–330, 1997.
malformations. J Neurosurg 77:832–841, 1992. 35. Perret G, Nishioka H: Report on the cooperative study of intracranial
14. Frizzel RT, Fisher WS 3rd: Cure, morbidity, and mortality associated with aneurysms and subarachnoid hemorrhage. Section VI. Arteriovenous malfor-
embolization of brain artetriovenous malformations: A review of 1246 mations. An analysis of 545 cases of craniocerebral arteriovenous malforma-
patients in 32 series over a 35-year period. Neurosurgery 37:1031–1040, 1995.
tions and fistulae reported to the cooperative study. J Neurosurg 25:467–490,
15. Gerosa M, Cappellotto P, Licata C, Iraci G, Pardatscher K, Fiore DL: Cerebral
1966.
arteriovenous malformations in children (56 cases). Childs Brain 8:356–371,
36. Pollock BE, Flickinger JC: A proposed radiosurgery-based grading system for
1981.
arteriovenous malformations. J Neurosurg 96:79–85, 2002.
16. Gerszten PC, Adelson PD, Kondziolka D, Flickinger JC, Lunsford LD: Seizure
37. Pollock BE, Gorman DA, Coffey RJ: Patient outcomes after arteriovenous
outcome in children treated for arteriovenous malformations using gamma
malformation radiosurgical management: Results based on a 5- to 14-year
knife radiosurgery. Pediatr Neurosurg 24:139–144, 1996.
follow-up study. Neurosurgery 52:1291–1297, 2003.
17. Gobin YP, Laurent A, Merienne L, Schlienger M, Aymard A, Houdart E,
38. Pollock BE, Kondziolka D, Flickinger JC, Patel AK, Bissonette DJ, Lunsford LD:
Casasco A, Lefkopoulos D, George B, Merland JJ: Treatment of brain arteri-
Magnetic resonance imaging: An accurate method to evaluate arteriovenous
ovenous malformations by embolization and radiosurgery. J Neurosurg
malformations after stereotactic radiosurgery. J Neurosurg 85:1044–1049, 1996.
85:19–28, 1996.
39. Pollock BE, Kondziolka D, Lunsford LD, Bissonette D, Flickinger JC: Repeat
18. Guidetti B, Delitala A: Intracranial arteriovenous malformations. Conserva-
stereotactic radiosurgery of arteriovenous malformations: Factors associated
tive and surgical treatment. J Neurosurg 53:149–152, 1980.
with incomplete obliteration. Neurosurgery 38:318–324, 1996.
19. Hladky JP, Lejeune JP, Blond S, Pruvo JP, Dhellemmes P: Cerebral arteriove-
40. Riva D, Pantaleoni C, Devoti M, Lindquist C, Steiner L, Giorgi C: Radio-
nous malformations in children: Report on 62 cases. Childs Nerv Syst
surgery for cerebral AVMs in children and adolescents: The neurobehavioral
10:328–333, 1994.
20. Hoh BL, Ogilvy CS, Butler WE, Loeffler JS, Putman CM, Chapman PH: outcome. J Neurosurg 86:207–210, 1997.
Multimodality treatment of nongalenic arteriovenous malformations in pedi- 41. Schmit BP, Burrows PE, Kuban K, Goumnerova L, Scott RM: Acquired cere-
atric patients. Neurosurgery 47:346–358, 2000. bral arteriovenous malformation in a child with moyamoya disease. Case
21. Humphreys RP, Hoffman HJ, Drake JM, Rutka JT: Choices in the 1990s for the report. J Neurosurg 84:677–680, 1996.
management of pediatric cerebral arteriovenous malformations. Pediatr 42. Shin M, Kawamoto S, Kurita H, Tago M, Sasaki T, Morita A, Ueki K, Kirino
Neurosurg 25:277–285, 1996. T: Retrospective analysis of a 10-year experience of stereotactic radio surgery
22. Kader A, Goodrich JT, Sonstein WJ, Stein BM, Carmel PW, Michelsen WJ: for arteriovenous malformations in children and adolescents. J Neurosurg
Recurrent cerebral arteriovenous malformations after negative postoperative 97:779–784, 2002.
angiograms. J Neurosurg 85:14–18, 1996. 43. Smyth MD, Sneed PK, Ciricillo SF, Edwards MS, Wara WM, Larson DA,
23. Kiris T, Sencer A, Sahinbas M, Sencer S, Imer M, Izgi N: Surgical results in Lawton MT, Gutin PH, McDermott MW: Stereotactic radiosurgery for pedi-
pediatric Spetzler-Martin grades I-III intracranial arteriovenous malforma- atric intracranial arteriovenous malformations: The University of California
tions. Childs Nerv Syst 21:69–76, 2005. at San Francisco experience. J Neurosurg 97:48–55, 2002.
24. Kondziolka D, Lunsford LD: Re: Current status of radiosurgery for arteriove- 44. Steiner L, Lindquist C, Adler JR, Torner JC, Alves W, Steiner M: Clinical out-
nous malformations. Can J Neurol Sci 19:514–515, 1992. come of radiosurgery for cerebral arteriovenous malformations. J Neurosurg
25. Lasjaunias P, Hui F, Zerah M, Garcia-Monaco R, Malherbe V, Rodesch G, 77:1–8, 1992.
Tanaka A, Alvarez H: Cerebral arteriovenous malformations in children. 45. Steiner L, Lindquist C, Steiner M: Stereotactic radiosurgery: Part 1—Radio-
Management of 179 consecutive cases and review of the literature. Childs surgery with focused gamma-beam irradiation in children, in Edwards MS,
Nerv Syst 11:66–79, 1995. Hoffman HJ (eds): Cerebral Vascular Disease in Children and Adolescents.
26. Leclerc X, Gauvrit JY, Trystram D, Reyns N, Pruvo JP, Meder JF: Cerebral arte- Baltimore, Williams & Wilkins, 1989, pp 367–388.
riovenous malformations: Value of the non invasive vascular imaging tech- 46. Tanaka T, Kobayashi T, Kida Y, Oyama H, Niwa M: Comparison between
niques [in French]. J Neuroradiol 31:349–358, 2004. adult and pediatric arteriovenous malformations treated by Gamma Knife
27. Levy EI, Niranjan A, Thompson TP, Scarrow AM, Kondziolka D, Flickinger radiosurgery. Stereotact Funct Neurosurg 66 [Suppl 1]:288–295, 1996.
JC, Lunsford LD: Radiosurgery for childhood intracranial arteriovenous mal- 47. terBrugge KG: Neurointerventional procedures in the pediatric age group.
formations. Neurosurgery 47:834–842, 2000. Childs Nerv Syst 15:751–754, 1999.
28. Lindqvist M, Karlsson B, Guo WY, Kihlstrom L, Lippitz B, Yamamoto M: 48. Wilkins RH: Natural history of intracranial vascular malformations: A review.
Angiographic long-term follow-up data for arteriovenous malformations Neurosurgery 16:421–430, 1985.

NEUROSURGERY VOLUME 60 | NUMBER 2 | FEBRUARY 2007 | 275

REYNS ET AL.

49. Wisoff JH, Berenstein A: Interventional neuroradiology, in Edwards MS, rhage or other symptoms, as well as the disability AVMs can create, we
Hoffman JH (eds): Cerebral Vascular Disease in Children and Adolescents. continue to advocate an aggressive stance in their management.
Baltimore, Williams and Wilkins, 1989, pp 139–157.
50. Yamamoto Y, Coffey RJ, Nichols DA, Shaw EG: Interim report on the radio- Douglas Kondziolka
surgical treatment of cerebral arteriovenous malformations: The influence of Pittsburgh, Pennsylvania
size, dose, time, and technical factors on obliteration rate. J Neurosurg
83:832–837, 1995.
51. Yamamoto M, Jimbo M, Ide M, Lindquist C, Steiner L: Gamma knife radio-
surgery in cerebral arteriovenous malformations: Postobliteration nidus
T he authors present their considerable experience (100 patients) with
the radiosurgical treatment (dedicated linear accelerator) of pedi-
atric intracranial AVMs. The indications are clearly defined and include
changes observed on neurodiagnostic imaging. Stereotact Funct Neurosurg
a clinically stable course and no impending intracranial hypertension.
64 [Suppl 1]:126–133, 1995.
Their results are encouraging with 70% of the patients having angio-
graphically recorded complete obliteration and no substantial differ-
ence between the pediatric and adult populations in terms of “thera-
COMMENTS peutic timing.” As expected, the two major predictive factors are
represented by a limited target volume and no previous endovascular

T he authors have compiled their extensive experience treating pedi-

atric arteriovenous malformations (AVMs) using linear accelerator
radiosurgery between 1988 and 2002, a period during which substantial
procedures. The latter seems to further stress the unsolved question of
appropriate therapeutic algorithms in AVMs. After documented oblit-
eration, there was only 2% posttreatment bleeding and no hemorrhage.
technological changes were occurring in radiosurgical practice. Among Finally, the morbidity rate was low (5%) and directly related to the
these developments were improved imaging and volumetric treatment nidus volume, whereas none of these patients ever complained of car-
planning and the development of radiation delivery systems capable of cinogenic sequelae.
highly conformal treatment. Nonetheless, their results are very good This interesting report deserves several specific comments. Age
with obliteration rates and treatment-related morbidity rates consistent definitively seems to be a predictive factor for AVM obliteration (3, 4);
with contemporary expectations. This is the largest pediatric experi- this is likely owing to either the peculiar radiosensitivity of immature
ence published thus far. The authors were able to obtain follow-up data vessels in the pediatric population or to focally (nidus wall, endothelial
for at least 3 years for every patient treated and have confirmed AVM cells) higher levels of specific growth factors (e.g., epidermal growth
obliteration angiographically in each instance. These aspects of their factor and fibroblast growth factor) in infants and children (1, 2).
report eliminate potential biases that appear in some other studies and Despite the reported rewarding results, the adopted “Talairach-based”
make it of considerable value in assessing the role of radiosurgery in procedure seems cumbersome, particularly when compared with cur-
treating pediatric AVMs. With regard to interval follow-up of treated rent radiosurgical techniques such as double staged localization under
patients, we reserve angiography for those patients in whom there is general anesthesia, single shot treatment, annual angiographic control
apparent obliteration on magnetic resonance imaging scans. This is of rather than magnetic resonance imaging plus magnetic resonance
particular relevance for children who require general anesthesia for angiography for at least 24 to 36 months. However, considering the
angiography. We have also found computed tomographic angiography substantial case material, the appropriate clinical and radiological
to be superior to magnetic resonance imaging and magnetic resonance remarks, and the long-term follow-up period, this retrospective study
angiography not only in assessing the residual nidus, but also in delin- represents a relevant contribution to the literature.
eating the AVM for treatment planning purposes.
Massimo Gerosa
Paul H. Chapman Verona, Italy
Boston, Massachusetts
1. Nicolato A, Lupidi F, Sandri MF, Foroni R, Zampieri P, Mazza C, Maluta S,

A
Beltramello A, Gerosa M: Gamma knife radiosurgery for cerebral arteriove-
VMs are the most common cause of pediatric stroke. As shown in
nous malformations in children/adolescents and adults. Part I: Differences in
this series, a wide variety of clinical problems can be associated
epidemiologic, morphologic, and clinical characteristics, permanent complica-
with a brain AVM and multimodality management is often necessary. tions, and bleeding in the latency period. Int J Radiat Oncol Biol Phys
This report makes a number of observations. First, the 70% overall AVM 64:904–913, 2006.
obliteration rate was reasonable. However, no AVM larger than 10 ml3 in 2. Nicolato A, Lupidi F, Sandri MF, Foroni R, Zampieri P, Mazza C, Pasqualin A,
volume was obliterated (n = 4). Thus, the care of patients with larger Beltramello A, Gerosa M: Gamma Knife radiosurgery for cerebral arteriove-
AVMs must be questioned. Perhaps embolization, resection, or staged nous malformations in children/adolescents and adults. Part II: Differences in
radiosurgery may need to be considered for these larger lesions. obliteration rates, treatment-obliteration intervals, and prognostic factors. Int
Secondly, the authors used a fairly uniform dose selection method, which J Radiat Oncol Biol Phys 64:914–921, 2006.
3. Pollack BE, Gorman DA, Coffey RJ: Patient outcomes after arteriovenous mal-
could perhaps have been tailored individually to specific AVMs. One
formation radiosurgical management: Results based on a 5- to 14-year fol-
patient developed monocular blindness after radiosurgery for an AVM
low-up study. Neurosurgery 52:1291–1297, 2003.
near the optic nerve. Was this related to the dose received or to dose plan 4. Shin M, Kawamoto S, Kurita H, Tago M, Sasaki T, Morita A, Ueki K, Kirino T:
conformality? Other centers have promoted the concept that children Retrospective analysis of a 10-year experience of stereotactic radio surgery
may obliterate their AVMs faster than adults with similar lesions. This for arteriovenous malformations in children and adolescents. J Neurosurg
remains a hypothetical concept. Because of the lifelong risk of hemor- 97:779–784, 2002.

276 | VOLUME 60 | NUMBER 2 | FEBRUARY 2007 www.neurosurgery-online.com