Child's Nerv Syst (1995) 11:66-79 9 Springer-Verlag 1995
Pierre Lasjaunias Francis Hui Michel Zerah Ricardo Garcia-Monaco V6ronique Malherbe Georges Rodesch Akio Tanaka Hortensia Alvarez
Received: l0 August 1993
R Lasjaunias ([N~)9E Hui R. Garcia-Monaco. G. Rodesch A. Tanaka 9H. Alvarez Unit6 de Neuroradiologie Vasculaire Diagnostique et Thdrapeutique, Centre Hospitalier de BicStre, 78, Rue du Gdndral Leclerc, F-94275 Le Kremlin Bic~tre, France Fax: 33.1.45.21.34.33 M. Zerah Unit6 de Neurochirurgie Pddiatrique, Centre Hospitalier de BicStre, 78, Rue du Gdn6ral Leclerc, F-94275 Le Kremlin Bic~tre, France V. Malherbe Unit6 de Neurologie P6diatrique, Centre Hospitalier de BicStre, 78, Rue du G6ndral Leclerc, F-94275 Le Kremlin BicStre, France
Cerebral arteriovenous malformations in children Management of 179 consecutive cases and review of the literature
Abstract Over the past 10 years (1982-1992), we have been actively involved in the management of 179 cerebral arteriovenous malformations (CAVMs) in children and infants. Seventy-seven were true vein of Galen malformations (VGAMs) and 102 were pial AVMs (PAVMs), i.e., developed in the subpial space. Hemorrhage occurred as the first symptom in 50% of the children with pial AVMs, but was present in none of the VGAM cases. Only 31 children were found to be unsuitable for endovascular treatment, and in 124 cases embolization was indicated as the primary treatment (104 embolization performed). Only 21 children underwent a direct surgical approach (none in the VGAM group). In the embolized group in whom treatment has been completed (n=56), 8 children died, 39 have an anatomical cure, and 34 are clinically normal. In the group under treatment (n=48), 16 are not normal. The problems are timing and the aims (total or partial treatment) of the therapeutic procedures. In the nonembolized group (n=31), 8/13 of the pial lesions were
Introduction Following Raimondi's, Gerosa's and Hoffman's publications and written contributions over the past 10 years on cerebral arteriovenous malformations (CAVMs) in children, the trend has moved from anecdotal case reports to short series, offering an understanding of the disease, ther-
operated on (no mortality, 2 patients with moderate neurological deficits). In the VGAM group 13/18 died and 4 had spontaneous thrombosis (only 1 is neurologically normal). In the nonembolized group 13 lesions have been completely excluded, but only 5 patients are neurologically normal. This fact again stresses the need for prognostic evaluation before treatment and a clear definition of the treatment aims. Analysis of a large number of published series on the management of children with AVMs (1017 cases) reveals inconsistencies that hamper proper evaluation and comparison. In our experience, endovascular treatment always seems to be the best primary treatment in both VGAMs and PAVMs. However, management of children with these lesions requires a large multidisciplinary team, which is the only way of offering the most suitable and effective treatment, the sole guarantee of a good result.
Key words Arteriovenous malformations 9 Vein of Galen Embolization
apeutic strategies, and various specific managements [5, 22, 23, 35, 47, 48, 52]. It is in the pediatric population that the most recent advances have been made. Consequently, it is also in this population that the strongest misconceptions persist. Historical contributions from the neurosurgical point of view have shown their limitations in the management of these difficult lesions (see Tables 6-8), and thus the gaunt-
67
let has been thrown d o w n to the alternatives of interventional neuroradiology. However, purely technical solutions have failed to provide satisfactory m i d - t e r m results, despite spectacular images [10, 15, 27, 41]. U n n e c e s s a r y hasty i n t e r v e n t i o n s have also impaired the quality of the results, b e t r a y i n g the difficulty of establishing satisfactory patient selection from a purely technical point of view. The reduction of the therapeutic challenge to a strictly m o r p h o logical goal ignores f u n d a m e n t a l aspects of neonatal and infant a n a t o m y and fluid p h y s i o l o g y [3, 32, 59]. For the past 10 years, our group has been i n v o l v e d in the m a n a g e m e n t of 179 CAVMs in children. The purpose of this paper is to present the results in this series with a m i n i m u m of a 1 year clinical and morphological followup. Obviously, the classical division according to size and topography does not help in predicting the natural history in these patients, as it refers exclusively to the risks involved in direct surgery. Size and topography will not be discussed here, as they are not relevant to e n d o v a s c u l a r m a n a g e m e n t . Differentiation b e t w e e n vein of G a l e n malformations ( V G A M s ) and pial AVMs (PAVMs) relates to the anatomic space in which these lesions develop: they are respectively s u b a r a c h n o i d and subpial [5]. In addition, these m a l f o r m a t i o n s represent two different groups of dev e l o p m e n t a l m a l f o r m a t i o n V G A M in the e m b r y o n i c period, PAVMs in the fetal or possibly postnatal stage [5]. The capacity to distinguish these two entities has always been regarded by our group as essential to proper comparisons of the m a n a g e m e n t of AVMs in children.
Patients and methods Of the 179 consecutive cases referred to us, 77 were classified as VGAMs and 102 as PAVMs. Our series is presented in two different ways: a series profile and summary of clinical presentation (Tables 1, 2) and a report of therapeutic results, depending on the mode of treatment chosen and whether or not that treatment has been completed (Tables 3-5). The patients treated are discussed as three separate groups: those who have completed embolization treatment, those still undergoing embolization treatment, and those not treated at all by embolization. Twenty-four cases were managed contrary to our recommendation or were treated at other centers and are therefore excluded from our discussion. Although it cannot be directly compared to neurosurgical series, our experience demonstrates that there is a different way of manag-
Table 2 Clinical presentation of 179 consecutive cases of cerebral AVM referred for consultation (VGAM vein of Galen aneurysmal malformation)
VGAM Age at diagnosis Antenatal Neonate Infant Child
18 28 24 7
Hemorrhagic onset Total
77
PialAVM
Total
1a 13 15 73
19 41 39 80
47 102
2b
49 179
a Pial and dural arteriovenous shunts; major neonatal cardiac failure b All late onset with venous outlet changes Table 3 Group of patients in whom embolization has been completed
No neurological abnormality Neurological deficitb or mental retardation Death ~ Neurological complications Transient Permanent
VGAM (n=28)
Pial AVM Total (n=28) (n=56)
19 6
15a 8
34 14
3d
5e
8
3 2
3 2
6 4
a Two high-flow fistulas excluded, one microlesion remaining b See text for details c Fatal clinical outcome, different from operative mortality (see Tables 6, 7) a One venous approach; in one case poor timing after ventricular shunting; one patient would be rejected for treatment today e Includes one patient with delayed recurrent hemorrhage after partial embolization Table 4 Group of patients still undergoing embolization treatment
No neurological abnormality Neurological deficita or mental retardation Complications Transient Permanent
VGAM (n=24)
Pial AVM n=24)
Total (n=48)
19 5
13 11
32 16
2 1
2 3
2
a See text for details Table 1 Treatment profile of 179 cases of cerebral arteriovenous malformations (AVM) in children (< 15 years old)
Cases embolized Treatment completed Under treatment Cases rejected for embolization Cases expected for embolization Proposed for embolization Confirmed to come Cases managed differently from our proposal
104 56 48 31 20 13 7 24
ing patients with these lesions and therefore illustrates a different approach to decision making during the pretreatment and follow-up phases. Based on our experience with the use of bucrylates in more than 400 CAVMs cases, we consider the results to be stabilized at 6 months to 1 year following the final embolization (Figs. 1, 2). We have not observed revascularization at later follow-ups when the angiographic evaluation at 6 months to 1 year was strictly normal. When slight hyperemia is demonstrated at 6 months, even without evidence of arteriovenous shunt, another control angiogram is performed 1 or 2 years later. At present, we would not yet suggest relying on magnetic resonance angiography (MRA) to assess the stability of acom-
68
Fig. 1 A Coronal and B sagittal magnetic resonance images (MRI) of a vein of Galen aneurysmal malformation (VGAM) in an infant presenting with macrocrania and ventricular enlargement. Note the patency of the mesencephalic aqueduct. C coronal and D sagittal follow-up MRI 1 year after successful transarterial embolization. Note the expected classical shrinkage of the venous ectasia, the reduction of the mass effect on the vermis and the tectum, and the normal size of the ventricles (without ventricular drainage). The corpus callosum has lost the laminar aspect it had at the pretherapeutic stage (B) and has grown normally (D)
Table 5 Group of patients not treated by embolization in whom treatment is complete VGAM (n=18) CAVM exclusion (children alive) Spontaneous Post surgical No neurological abnormality Neurological deficit or mental retardation a Death
Pial AVMs (n=13)
Total (n=31)
4 0
1 8
13
1 4b
4 7
5 11
2
15
13
a Includes post-therapeutic (nonendovascular) neurological complications (2 cases); see text for details b Includes one disabled child with a partially thrombosed VGAM
plete occlusion. At this time, all children are clinically evaluated. The clinical assessment is based on neurocognitive examination using the Denver and Brunet Leizine tests. All children are followed up with a clinical examination every year and MRI every 2 years even after complete anatomic exclusion. This ongoing clinical follow-up is mandatory in the pediatric population since therapeutic success can only be truly evaluated when brain maturation is complete. Treatment strategy and rationale by age group have been published in the past few years by our group [19, 30, 50, 59] and will not be discussed in the present paper.
Results A l t h o u g h o u r series is a l a r g e o n e , it d o e s n o t r e f l e c t a true population profile with strong epidemiological signifi-
69
Fig. 2 A - D Tectal hematoma in a 14-year-old boy. A Nonenhanced axial CT images. B T2-weighted MRI confirms the location of the lesion and suggests a cavernoma. C Left vertebral angiography discloses a tectal arteriovenous malformation (AVM) fed by a long circumferential artery and draining into a posterior mesencephalic vein towards the vein of Galen. Superselective embolization was performed with N-butyl cyanoacrylate. D Control angiogram 6 months after the endovascular procedure, confirms the total exclusion of the AVM
cance, but rather a w i d e referral pattern spread across Eur o p e and the M i d d l e East. It is o f interest to recall that for the y e a r 1989 a survey on the E u r o p e a n continent (25 countries) [48] s h o w e d that in 530 m i l l i o n p e o p l e 178 interventions had been p e r f o r m e d for v a s c u l a r disease in children, i.e., about one intervention per y e a r p e r 3 m i l l i o n i n h a b i tants. M o s t active n e u r o s u r g i c a l centers in E u r o p e p e r f o r m b e t w e e n 0 and a m a x i m u m o f 15 ( D e n m a r k ) interventions for such d i s e a s e in children, w h i c h in m a n y cases consists o f c a v e r n o m a s . A l l our patients were referred to us b y neu-
70
rosurgeons or following neurosurgical advice. As would be expected, only the more difficult cases are referred to us; the easier ones are managed in their own centers. We realize that a breaking down the analyis according to disease and age would appear of a more immediate practical value, but we think that this would hide the need for wide exposure to the problem in order disclose the differences in disease management and that would focus only on individual technical challenges. However, our results clearly distinguish VGAMs from PAVMs. The great majority of patients (over 87%) were referred to us without delay, i.e., at the age of onset. One hundred four patients underwent embolization (Table 1), and the same primary onset treatment was indicated in a further 20. Thirty-one were not felt to be candidates for the endovascular approach, either because the lesion was thrombosed, or the clinical status was irreversibly poor, or direct surgery without embolization was felt to be the best treatment strategy to offer.
Completed embolization group (n=56; Table 3) None of the patients with VGAMs (n=28) underwent surgery, whereas in the PAVM group (n=28) 13 underwent a direct surgical approach following partial embolization. Fig. 3 Left carotid angiogram oblique view in a 13-year-old female. This brings to 21/155 the number of children who bene- The micro-AVM (arrow) of the rolandic region responsible for an fited from a direct surgical approach with or without em- intracerebral hernatoma was not considered for endovascular therapy and the patient was referred for surgery bolization (Figs. 3, 4). Ventricular shunting is not regarded here as surgical treatment of the AVM. Three out of the 28 PAVM cases received radiosurgical complementary treatment following partial embolization with bucrylate (Fig. 5). Sixty sessions were needed in the V G A M cases (aver- abnormality were related either to embolization or ventricage 2 per patient) and 67 sessions in the PAVMs (2.4 per ular shunting. With the exception of the four complications patient). Eight children died in this group despite treat- mentioned (Table 3) and six additional transient ones (afment, seven of them following embolization of an other- ter over 300 arteries embolized with bucrylate), no patient wise nonsurgically treated AVM. In none of these cases declined in this respect who was not severely impaired or was there a surgically related death. However, in four of disabled when referred. these seven patients the decision would have been different today: three would have been rejected for embolization (all were neonates with AVMs and severe multiorgan fail- Ongoing embolization group (n=48; Table 4) ure), and one would have undergone embolization either without previous ventricular shunt or a few days after In the group of patients who have not yet completed their shunting, or at least following control computed tomogra- embolization treatment we are confronted with a new popphy. Two patients died after a hemorrhagic accident (see ulation of patients who were not present as such in the past. below) and the last one following a transtorcular venous Sixteen are not normal, and one might ask, if the treatment is not completed, why not continue it? The decision to stage approach. A significant number of children in this group are ana- a procedure carries some significant advantages, particutomically (n=39) or clinically (n=34) normal. Not all chil- larly in children where the length of the procedure is limdren cured are normal, and some who are clinically nor- ited by puncture time or fluid volumes [5]. In this group mal cannot be cured. Of the surviving children, the num- the problem is the time between two consecutive sessions. bers are respectively 39/48 and 34/48. In the 14 clinically There is no optimal definitive frame; however, with expeabnormal children, only 2/14 had severe mental retarda- r i e n c e each interventional neuroradiologist can determine tion; (>20% of chronological age) 12/14 had moderate whether or not a better technical result can be achieved afmental retardation or mild neurological symptoms (hemi- ter waiting for a few weeks or months. This will have to anospia, hemiparesis, or seizures): 4/14 cases of clinical be balanced with the hemodynamic changes induced by the
71
Fig. 4 A Left vertebral angiogram in a 13-year-old girl with a left occipital AVM. Previous proximal embolization led to angiogenesis that could not be controlled by further endovascular therapy. B Postsurgical left vertebral angiogram. Surgery excluded the lesion secondarily, with technical difficulties due to the postembolization induced angiogenesis Fig. 5 A Seven-year-old boy with a corpus collosum AVM fed by en passage transcallosal arteries of the left pericallosal artery and draining into the internal cerebral vein. The lesion was embolized superselectively and the remainder (B) seen 6 months later treated by radiosurgery
72
actual embolization and the hemodynamic level needs to ensure a proper or preserved maturation process. The decision will depend on the projected clinic~il status ,of the patient over the ensuing few weeks and accurate analysis of the venous angioarchitecture [5]. In the abnormal group, 14/16 presented with either mild mental retardation (< 20% of chronological age), which in infants can be rapidly compensated, or a mild deficit related to ventricular shunting or a previous hemorrhagic episode. The remaining cases will never be cured with the present treatment modalities, either because the lesion is unreachable (even by combined approaches) or because the disease is multifocal [20, 25]. In this subgroup the treatment will be partial and repeated over time with large intervals between sessions. The aim of such partial treatment is to erase existing dangerous portions of the AVM progressively (flow-related aneurysms are exceptional at that age; false aneurysms are seen following a hemorrhagic episode; venous ectasias are frequent in that population) or its effects on the adjacent brain (venous congestion, venous thrombosis) [5, 31, 55]. In our experience partial embolization is an acceptable therapeutic objective only if a complete cure cannot be obtained at a satisfactory level of risk, and if the procedure is performed with a permanent agent like bucrylate. The high rate of permanent neurological complications (3/48) observed in this group reflects the aggressive management in two difficult cases of VGAM cases, and one noncurable pial AVM (3 transarterial approaches).
Nonembolization group (n=31; Table 5) No VGAM has been operated on. Thirteen out of 18 patients with nontreated VGAMs died. Of the remaining 5, 4 experienced spontaneous thrombosis (only one is neurologically normal). The concept of a favorable spontaneous evolution of VGAMs [4, 58] is inaccurate and dangerous, since all the spontaneously thrombosing VGAMs that we have seen were easy forms that could have been treated with the expectation of a high rate of normal neurological outcome. In our series, PAVMs (nongalenic AVMs; n=l 3) had a poorer prognosis in neonates and infants than VGAMs (n=18). In the pial group eight surgical attempts were performed as the first (and only) treatment modality, seven of them in patients older than 2 years of age. All patients had their AVMs removed and all are alive; two of the eight have moderate permanent neurological deficits. It is also important to note that only 5/31 children alive in this group are normal, and 13/31 of the lesions are completely excluded now. This underlines again the need for early management, which does not necessarily mean an emergent approach. In this group the 50% mortality (15/31) is hardly reducible and should include the six additional cases that would not be emboIized (21/155) today.
Discussion
Our experience in the management of CAVMs in generfil started in 1982 and encompasses 651 cases studied and 372 embolized with bucrylate using a consistent technique with few catheter changes. This pediatric series therefore reflects roughly one-third of grossly all CAVMs seen or embolized by our group. Literature review Few reports have appeared in the world literature dealing with series of AVMs managed in childhood, either PAVMs (Tables 6, 7) or VGAMs (Table 8). Most are technically oriented towards surgery, which seldom appears to target the lesion itself directly. Endovascular approaches are either ignored or insufficiently documented with a limited literature search. In addition, few publications on radiation therapy series are available for this population [1, 11, 17, 33] (Table 7). Finally, confusing anatomic classifications greatly hamper proper comparison between series. In sum, the different contributions show no real common approach to CAVMs in children. Generally speaking, a review of the literature for children is frustrating, in contrast to the situation for adults.
Age groups The concept of pediatric age varies between 15 and 20 years. There is little experience on the management of nongalenic AVMs in neonates and infants and no reports on antenatal series.
Topography Differentiation between cortical, deep, and infratentorial AVMs is interesting technically, but basically the topography is known to be the least important factor in the anticipated natural history [5, 6, 12, 44]. Choroidal AVMs, which are seldom analyzed separately, are probably a different entity within the nongalenic category of AVMs.
Time span covered The series extend from i941 to the present and therefore mix different degrees of knowledge, techniques, and management.
Series size The series are small and the cases are often difficult to analyze since, for example, a distinction between VGAM and
73
Table 6 Literature review: surgical treatment of pial AVMs (TE total exclusion, PE partial exclusion, VP ventriculoperitoneal) Authors
Upper age limit (years)
Study
Total no. of cases
Operated CAVMs/VGAMs
Gerosa et al. [22]
16
19541979
56
Eir~s et al. [16]
17
19751985
Fong and Chang [18]
16
Amacher et al. [2]
Patients with total exclusion
Operative mortality
Remarks
Neurologically normal
Neurologically abnormal
38
17 (45%)
6 (16%)
3 (8%)
Partial treatment included
17
14
9 (64%)
2 (14%)
0
Partial treatment not included
1971 1987
39
27
23 (85%)
3 (11%)
1 (4%)
No distinction made between TE and PE. All cases taken to be TE
18
?
20
20
15 (75%)
1 (5%)
0
Celli et al. [9]
15
19511980
19
11
6 (55%)
5 (45%)
0
Yasargil [58]
15
v
60
60
54 (90%)
6 (10%)
0
Martin and Edwards [37]
?
?
35
35
30 (86%)
1 (3%)
1 (3%)
Ages not stated
Mori et al. [42]
15
?
28
18
5 (28%)
3 (17%)
Some discrepancy between tables and text
Mazza et al. [38]
16
19711982
24
18
15 (83%)
1 (6%)
2 (11%)
Lapras et al. [29]
15
?
65
62
26 (42%)
25 (40%)
6 (10%)
Total actually 65, but 2 had VP shunts only and 1 had embolization (excluded)
Humphreys [24]
17
19541984
100
72
43 (60%)
23 (32%)
6 (8%)
Figures do not tally with text: 74 patients, but only 72 accounted for; AVM removed 54 times; no distinction made between TE and PE; all cases taken to be TE
Partington et al. [45]
16
19851988
12
12
11 (92%)
1 (8%)
0
Stereotactic surgery; includes cavernoma and venous angioma
Garza-Mercado et al. [21]
18
1968 1985
19
12
3 (25%)
4 (33%)
4 (33%)
Malik et al. [36]
18
1941 1989
46
27
23 (85%)
2 (7%)
2 (7%)
Laine et al, [28]
20
20
20
12 (60%)
4 (20%)
4 (20%)
Tamaki and Ehara [54]
19
19501980 1970 1990
25
25
23 (92%)
2 (8%)
0
No distinction made between TE and PE; all cases taken to be TE
Suarez and Viano [53]
15
?
15
10
7 (70%)
0
3 (30%)
Computation of percentage not accurate
Nelson et al. [43]
19
19781990
13
11
7 (64%)
1 (9%)
Surgery alone, embolization alone or combined
Total
15 to 20
19411990
613
492
278-330 (44-67%)
86 (17%)
36 (7%)
Over 19 patients below 16 years; only 4 were cured and neurologically normal out of 7 surgically treated
112 counted as TE and neurologically normal
74
Table 7 Literature review: radiotherapy of pial AVMs Authors
Upper age limit (years)
Study
Total no. Operated Total exclusion OP Remarks of cases CAVMs/ mortality VGAMs Neuro- Neurologically logically normal abnormal
Altschuler et al. [1] Loeffler et al. [3]
18 15
1987-1988 1986-1988
18 5
18 5
3 (17%) 0% 3 (60%) 0%
Colombo [11] Total
15 15 to
1984-1989 24 1984-1989 47
24 47
11 (46%) 0% (17% to 0%
18
PAVMs is not always clear. The most recent reports no longer mix the VGAMs and the PAVMs; however, within VGAMs, vein of Galen aneurysmal dilations (VGADs) [32] and true VGAMs often remain undifferentiated, particularly by those who still use the four types of Yasargil's classification [57].
60%)
Other 11 cases: no angiography done 1 case not scheduled for angiography yet 6 cases - no angiography 18 cases with no follow-up angiography
thereafter. Persistence of anatomic confusion between VGAMs (Fig. 6) with transdiencephalic and subependymal supply and tectal PAVMs with transmesencephalic supply still hinders comparisons.
Clinical aspects
Operated cases The problem is that most lesions are not directly attacked. In large groups of nonoperated patients, outcomes are not detailed. Partial surgical treatment is not promoted as such, but is often performed with feeder ligations; this intervention differs from partial embolization with bucrylate and introduces confusion in comparisons of the two concepts and associated results. Many of the children included in the surgical series are operated on as adults. Control of anatomic exclusion and clinical status are often difficult to assess, with few follow-up angiograms performed and neurocognitive testing carried out only rarely.
Mortality in the operated group Patient selection being imprecise, the interpretation of "mortality" varies and the figures range from 0% to 35%, depending upon the aggressiveness of the team in desperate situations. Technical morbidity/mortality is not distinguished from morbidity-mortality "despite attempted treatment." This absence of precision tends to promote unecessarily heroic procedures in nonfatal situations. The same remarks apply to reports of series with endovascular treatment of V G A M that have appeared in the literature where, for example, mental retardation in young children (often present) is seldom mentioned or tested. Most cases are published with insufficient clinical or radiological follow-up, and results are based on the immediate pictural aspect with no further evaluation. Some cases of anatomic exclusion of lesions are even counted as technical successes despite the fact that the child died shortly
In our series the number of neonates and infants is very high due to our specific involvement in VGAMs (Table 2), as most neonatal and almost all antenatal AVMs diagnosed are VGAMs. The symptom in neonates is almost consistently congestive heart failure. The importance of c o n g e s tive heart failure varies from one case to the next [19]; proper medical management is particularly crucial for timing treatment. Seventy-seven out of 99 neonates and infants actually presented with either cardiac :failure or cerebral hydrodynamic disorders [19, 59]. Sixteen out of 102 cases were referred with the erroneous diagnosis of VGAM; all were PAVMs with a dilation of the vein of Galen. Seven of 16 presented as neonates or in infancy; 4 of 16 had a hemorrhagic onset, compared to 2/77 with VGAMs [30, 49] and 43/86 with PAVMs without vein of Galen dilation. The increase in the number of patients with PAVMs (and the parallel decrease in VGAM) with age agrees with the findings in most series, due to better clinical assessment of neonates at birth. It is surprising, however, that only one "pial" lesion was diagnosed in utero; in fact, this malformation was diagnosed as a V G A M and turned out to be a giant dural lake associated with meningeal and pial arteriovenous shunts.
Bleeding as the first symptom It is important to stress that contrary to common opinion, VGAMs do not bleed unless secondary venous changes reroute the drainage into the pial veins. Conversely, almost half of PAVMs had a hemorrhagic onset, which conforms to large series in the literature. While waiting for embolization, 4/111 actually bled (3/4) or rebled (1/4). One of
1950-1985
? National review
20 years
15 years 28 months
7 years
1 month
3 years
12 years
18 years
20 years
Johnston et al. (includes Hoffman's series) Yasargil [57] Maheut et al. [35]
Merland et al. [39]
Ciricillo et al. [10]
Casasco et al. [7] b
Mickle [40] (transtorcular embolization)
Wisoff et al. [56] (Berenstein)
Total
357
33
26
7
14
6
14 53
191
13
218
33
24
7
13
6
14 26
89
6
No. of No. of V G A M s cases treated
84 (39%)
17 (52%)
5 (21%)
5 (71%)
1 (8%)
9 (64%) 3 (12%)
27 (30%)
3 (50%)
Neurologically normal
43 (20%)
5 (15%)
13 (54%)
0
0
2 (33%)
5 (19%)
19 (21%)
2 (33%)
Neurologically abnormal
Patients treated by direct approach a
a Surgical, transarterial, transvenous b Includes only cases treated; no patient selection introduced; interventional neuroradiological series
1950-1990
1977-1988
?
1988-1990
1978-1989
1979-1986
1975-1985
10 months
Johnston et al. [26] (personal series)
Study
Upper age limit
Authors
Table 8 Literature review: treatment of V G A M s
84 (38%)
6 (18%)
6 (25%)
18 treated patients suffered late death or were not accounted for; impossible to assess the amount of stable complete exclusion
The normal patient with TE has had embolization with particles; no followup angiography Surgery alone, embolization alone or combined; no follow-up angiography Two infants had VP shunts inserted; one survived with mild deficit; the other succumbed to sudden death Includes surgery and embolization; 5 late deaths not included 7 (54%)
0
Another patient died 2 years after embolization with balloons and particles
Result in 5 unknown
One patient died 3 months after total obliteration; not considered an operative death;2 patients had embolization intraoperatively
Remarks
3 (50%)
18 (69%)
5 (36%)
38 (43%)
1 (17%)
Operative mortality
tJI
76
Fig. 6A, B Subependymal arterial supply (arrow) to a VGAM of the choroidal type in a 7-year-old boy
Table 9 Overall results in 87 children with cerebral AVMs (Bic~tre 1982-1992) VGAM (n=46) Patients treated and treatment 28 completed (embolization, operation, combined treatment) Mortality in the treated group 3 (10.7%) No. of patients alive 30 (65.2%) No. of patients anatomically cured 27 (90%) No. of patients neurologically normal 20 (66.7%)
Pial AVMs (n=41) 36 2 (5.5%) 34 (82.9%) 25 (73.5%) 19 (55.9%)
the three suffered delayed hemorrhage 1 month before his scheduled first session of embolization. Two of the three had diencephalic and mesencephalic AVMs, respectively; both AVMs bled between the third and fourth sessions. The first of these patients, who was initially nonoperable, was subsequently operated on with success and had moderate residual sensory motor deficit; the second had a nonoperable AVM and died from that hemorrhage. The fourth patient died from a recurrent hemorrhage 5 years after the first bleeding episode and 1 year following the embolization session in which incomplete occlusion had been carried out; our group had felt that the therapeutic risks involved in completing the treatment by any method were too high. None of the patients belonged to the VGAM group.
Hydrodynamic disorders Management of hemo- and hydrodynamic disorders has already been addressed [19, 59]. However, it should be recalled that ventricular shunting in cerebral AVMs in in-
rants, and particularly in VGAMs, carries a high morbidity [51, 59], which alters the possible clinical outcome following the shunt treatment itself. Hydrodynamic disorders should be managed early by reduction or occlusion of the arteriovenous shunt. As a matter of fact, most infants described as clinically normal before consultation had some degree of mental retardation at the first examination and evaluation. Most of them recovered after partial treatment, and the remainder are still improving as treatment is completed. The picture is different for the infant with significant mental retardation (chronological age >20% above tested age), in whom improvement is usually incomplete. The prominent anatomic features relate to the differences that exist between the neonate and infant age groups versus older children [3, 8, 51, 59]. Even if pial lesions at that age manifest with features that reproduce the VGAM pattern, symptoms will be significantly different, with rare hydrodynamic disorders and frequent focal neurological symptoms or hemorrhage. PAVM will rapidly produce localized (or asymmetrical) brain "atrophy" (melting brain syndrome) and not diffuse macrocrania with bilateral brain damage, in contrast to VGAMs. This atrophy may represent the local effect of disordered hydrodynamics induced by local pial venous congestion. In dural sinus hyperpressure, the increased fluid pressure associated with head enlargement may constitute a way of restoring an efficient pressure differential between intradural veins and dural sinus circulations [46, 59]. Both can induce reversible tonsillar prolapse, which expresses the posterior fossa hydrovenous [3, 59] disorders and responds to appropriate treatment. These observations stress the need for early consultation with a competent team to avoid unnecessary fatalism or delay. This also shows that the prognosis of symptomatic neonates and infants with CAVMs is not primarily dependent on our capacity to eradicate the arteriovenous shunt, but on the degree and type of cerebral damage already present.
77
Therapeutic objectives Cerebral arteriovenous shunts are different in children than in adults: multifocal lesions, induced remote arteriovenous shunts [20, 25], venous thromboses, systemic phenomena [13, 14, 55], large venous ectasias, high-flow lesions, rapid atrophy, and others. Conversely, high-flow angiopathic changes are not seen with the same frequency; flow-related arterial aneurysms are Seen only exceptionally [31], whereas proximal occlusive arteriopathy is more frequent. For this reason, management protocols derived from experience in adults cannot be applied to the child population. In particular, adult-based grading only indicates the expected surgical outcome if the operator has the same skills as the one who made the grading. This biais is particularly inappropriate for in children, in whom: (I) cerebral eloquence is difficult to assess, particularly in the first years of life; (2) most lesions are fistulas or multifocal and (3) because of their flow usually involve the entire venous system; (4) the possibilities of recovery in children are different from adults. A perverse effect is the belief that this grading applies to the spontaneous evolution of the lesion. It is clear that an AVM that is difficult (high-grade) to operate on is not necessarily one that is dangerous for the patient if not operated upon, or at least more dangerous than a low-grade AVM. Nevertheless, it is hardly in the scope of our paper to decide whether or not more cases could have been operated upon. The decision process in children includes, in addition to the conventional objectives, various peculiar factors such as anatomic analysis of veins and the myelinization process. Thereafter, progressive deficits related to congested cerebral veins, poorly controlled seizures, hemorrhages with or without specific arterial or venous changes upstream and downstream of the AVMs, or headaches in children without hydrocephalus or macrocrania can become immediate objectives for staged or partial treatment. In our experience, no child considered completely cured at follow-up showed any recurrent arteriovenous shunting or any new symptom at later clinical or angiographic follow-up. Normal maturation is an important follow-up criterion in neonates and infants; even without deficit, hemorrhage, or seizure; it helps in assessing treatment quality and success. Failure to obtain a normal maturation process can even constitute a therapeutic failure if the optimum moment for intervention has been missed. This points to the
difference between early irreversible damage and damage by persistent disorder and secondary destruction. Brain calcification, which can be observed in both instances, indicates failure to restore normal hemo- and hydrodynamic conditions in the maturing brain. Concerning VGAM management, the transarterial approach with proper patient selection offers good results and represents a significant change in current notions concerning that disease. The indication for treatment will lead to technical choices based on the optimum moment to intervene.
Conclusion Our group now has 10 years of experience in the management of CAVMs in children [9]. The adult-based grading has not demonstrated its validity in the management of CAVMs in the pediatric population, in particular in the first few years of life when the anatomy and physiology are different, and because of the specificity of the angioarchitecture. An understanding of the interrelationship between the lesion to be treated and the other structures has shown its value and permits better patient selection. In our experience, endovascular management of CAVMs in children offers the largest range of possibilities and results. The traditional neurosurgical approach has limited indications and no place in partial treatment unless a proper pediatric interventional neuroradiology team is not available. Prediction of the risks of CAVMs in children are different, depending on the anatomophysiologic status of the child: before 2 years of age, the lesions are mainly VGAMs causing systemic and hydrodynamic disorders for which the best therapeutic option is the endovascular approach. Nongalenic lesions at the same age have a very poor prognosis. After 5-7 years of age, the physiology, type, and architecture of these lesions are closer to those in adults and prompt a similar type of management, which in our group is primarily endovascular, followed by surgery each time anatomic exclusion cannot be obtained by embolization. Between these ages management will still be mainly endovascular but, because of late referral, it will often be applied to already retarded children with unpredictable outcomes.
References 1. Altschuler EM, Lunsford LD, Coffey RJ, Bissonette DJ, Flickinger JC (1989) Gamma knife radiosurgery for intracranial arteriovenous malformations in childhood and adolescence. Pediatr Neurosci 15:53-61
2. Amacher AL, Drake CG, Hovind L (1979) The results of operating upon cerebral aneurysms and angiomas in children and adolescents. II. Cerebral angiomas. Child's Brain 5:166-173
3. Andeweg J (1989) Intracranial venous pressures, hydrocephalus and effects of cerebrospinal fluid shunt. Child's Nerv Syst 5:318-323
78
4. Beltramello A, Perini S, Mazza C (1991) Spontaneously healed vein of Galen aneurysm. Child's Nerv Syst 7:129-134 5. Berenstein A, Lasjaunias P (1993) Surgical neuroangiography, vol 4: Eudovascular treatment of cerebral lesions. Springer, Berlin Heidelberg New York, pp 267-316 6. Brown RD, Wiebers DO, Forbes G, O'Fallon WM, Piepgras DG, Marsh WR, Maciunas RJ (1988) The natural history of unruptured intracranial arteriovenous malformations. J Neurosurg 68:352-357 7. Casasco A, Lylyk P, Hodes JE, Kohan G, Aymard A, Merland JJ (1991) Percutaneous transvenous catheterization and embolization of vein of Galen aneurysms. Neurosurgery 28:260-266 8. Castro ME, Portnoy HD, Maesaka J (1991) Elevated cortical venous pressure in hydrocephalus. Neurosurgery 29:232-238 9. Celli P, Ferrante L, Palma L, Cavedon G (1984) Cerebral arteriovenous malformations in children. Clinical features and outcome of treatment in children and in adults. Surg Neurol 22:43-49 10. Ciricillo SF, Edwards MSB, Schmidt KG, Hieshima GB, Silverman NH, Higashida RT, Halbach VV (1990) Interventional neuroradiological management of vein of Galen malformations in the neonate. Neurosurgery 27:22-27 11. Colombo F (1991) Linear accelerator radiosurgery of cerebral arteriovenous malformations in the pediatric age group. In: Raimondi AJ, Choux M, Di Rocco C (eds) Cerebrovascular diseases in children. Springer, Berlin Heidelberg New York, pp 159-167 12. Crawford PM, West CR, Chadwick DW, Shaw MDM (1986) Arteriovenous malformations of the brain: natural history in unoperated patients. J Neurol Neurosurg Psychiatry 49:1-10 13. Cronqvist S, Granholm I, Lundstrom NR (1972) Hydrocephalus and congestive heart failure caused by intracranial arteriovenous malformations in infants. J Neurosurg 36:249-254 14. Cumming GR (1980) Circulation in neonates with intracranial arteriovenous fistula and cardiac failure. Am J Cardiol 45:1019-1024 15. Dowd CF, Halbach VV, Barnewell SL, Higashida RT, Edwards MSB, Hieshima GB (1990) Transfemoral venous embolization of vein of galen malformations. AJNR 11:643-648 16. Eiras J, Gomez-Perun J, Carcavilla LI, Alberdi J (1987) Surgical experience with arteriovenous malformations in children. Child's Nerv Syst 3:156-160
17. Fabrikant JI, Frankel KA, Phillips MH, Levy RP (1989) Stereotactic heavy charged-particle Bragg peak radiosurgery for intracranial arteriovenous malformations. In: Edwards MSE, Hoffman HJ (eds) Disease in childhood and adolescence. Williams & Wilkins, Baltimore, pp 389-409 18. Fong D, Chan ST (1988) Arteriovenous malformation in children. Child's Nerv Syst 4:199-203 19. Garcia-Monaco R, De Victor D, Mann C, Hannedouche A, Terbrugge K, Lasjaunias P (1991) Congestive cardiac manifestations from cerebrocranial arteriovenous shunts. Endovascular management in 30 children. Child's Nerv Syst 7:48-52 20. Garcia-Monaco R, Rodesch G, Terbrugge K, Burrows E Lasjaunias P (1991) Multifocal dural arteriovenous shunts in children. Child's Nerv Syst 7:425-431 21. Garza-Mercado R, Cavazos E, TamezMontes D (1987) Cerebral arteriovenous malformations in children and adolescents. Surg Neurol 27:131-140 22. Gerosa MA, Capellotto P, Licata C, Iraci G, Pardatscher K, Fiore DL (1981) Cerebral arteriovenous malformations in children (56 cases). Child's Brain 8:356-371 23. Hoffman HJ, Chuang S, Hendrick ED, Humphreys RP (1982) Aneurysms of the vein of Galen. J Neurosurg 57:316-322 24. Hurnphreys RP (1986) Hemorrhagic stroke in childhood. J Pediatr Neurosci 2:1-10 25. Iizuka Y, Rodesch G, Garcia-Monaco R, Alvarez H, Burrows R Hui F, Lasjaunias P (1992) Multiple cerebral arteriovenous shunts in children report of 13 cases. Child's Nerv Syst 8:437-444 26. Johnston IH, Whittle IR, Besser M, Morgan MK (1987) Vein of Galen malformation: diagnosis and management. Neurosurgery 20:747-757 27. King WA, Wackym PA, Vinuela F, Peacock WJ (1989) Management of vein of Galen aneurysms. Combined surgical and endovascular approach. Child's Nerv Syst 5:208-211 28. Laine E, Jomin M, Clarisse J, Cornbelles G (1981) Les malformations art6rio-veineuses c6r6brales profondes. Classification topographique, Possibilit6s et r6sultats th6rapeutiques propos de 46 observations. Neurochirurgie 27:147-160 29. Lapras C, Gharbi S, Mottolese C (1990) Traitement chirurgical des angiomes de l'enfant. Arguments pour une attitude chirurgical active. Pediatric 45S:231s-237s
30. Lasjaunias P, Terbrugge K, Piske R, Lopez-Ibor L, Manelfe C (1987) Dilatation de la veine de Galien. Formes anatomo-cliniques et traitement endovasculaire ~tpropos de 14 cas explor6s et/ou trait6s entre 1983 et 1986. Neurochirurgie 33:315-333 31. Lasjaunias P, Piske R, Terbrugge K, Willinsky R (1988) Cerebral arteriovenous malformations (CAVM) and associated arterial aneurysms (AA). Acta Neurochir 91:29-36 32. Lasjaunias P, Garcia-Monaco R, Rodesch G, Terbrugge K, Zerah M, Tardieu M, De Victor D (1991) Vein of Galen malformation endovascular management of 43 cases. Child's Nerv Syst 7:360-367 33. Loeffier JS, Rossitch E, Siddon R, Moore MR, Rockoff MA, Alexander E III (1990) Role of stereotactic radiosurgery with a linear accelerator in treatment of intracranial arteriovenous malformations and tumors in children. Pediatrics 85:774-782 34. Loirat C, Guillemont C, Pillion G, Macher MA, Maisin A (1990) D6veloppement intellectuel des enfants atteints d'insuffisance r6nale grave pr6coce. Sere H6p Paris 66:1752-1756 35. Maheut J, Santini JJ, Barthez MA, Billard C (1987) Symptomatologie clinique de l'an6vrysme de l'ampoule de Galien. R6sultats d'une enqu~te nationale. Neurochirurgie 33:285-290 36. Malik GM, Sadasivan B, Knighton RS, Ausman HI (1991) The management of arteriovenous malformations in children. Child's Nerv Syst 7:43-47 37. Martin NA, Edwards MSB (1989) Supratentorial arteriovenous malformations. In: Edwards MSB, Hoffman HJ (eds) Cerebrovascular disease in childhood and adolescence. Williams & Wilkins, Baltimore, pp 283-308 38. Mazza C, Pasqualin A, Scienza R, Bazzan A, Da Pian R (1983) Intracranial arteriovenous malformations in the pediatric age: experience with 24 cases. Child's Brain 10:369-380 39. Merland JJ, Laurent A, Rufenacht D, Reizine D (1987) MAV de la r6gion de l'ampoule de Galien. Aspects anatomiques, cliniques et 6volution du traitemerit endovasculaire (1979-1986) ?~ propos de 10 cas. Neurochirurgie 33:349-352 40. Mickle JP (1991) The transtorcular embolization of vein of Galen aneurysms and update on the use of this technique in twenty-four patients. Concepts Pediatr Neurosurg 11:69-78 41. Mickle JP, Quisling RG (1986) The transtorcular embolization of vein of Galen aneurysms. J Neurosurg 64:731-735
79
42. Mori K, Murata T, Hashimoto N, Handa H (1980) Clinical analysis of arteriovenous malformations in children. Child's Brain : 13-25 43. Nelson PK, Niimi Y, Lasjaunias P, Berenstein A (1992) Endovascular embolization of congenital arteriovenous fistulas. In: Interventional neuroradiology. Neuroimaging clinics of North America. Saunders, Philadelphia, pp 309-331 44. Ondra SL, Troupp H, George ED, Schwab K (1990) The natural history of symptomatic arteriovenous malformations of the brain: a 24-year follow-up assessment. J Neurosurg 73:387-391 45. Partington MD, Davis DH, Kelly PJ (1989) Stereotactic resection of pediatric vascular malformations. Pediatr Neurosci 15:217-222 46. Quisling RG, Parker Mickle J (1989) Venous pressure measurements in vein of Galen aneurysms. AJNR 10:411-417 47. Raimondi A (1987) Pediatric neurosurgery. Springer, Berlin Heidelberg New York, pp 321-329 48. Raimondi A, Majno M (eds) (1990) First European Symposium: Pediatric Nenrosurgery in Europe. State of the art, 1989. Child's Nerv Syst 6 [Suppl] S 1-$66
49. Raybaud CA, Strother CM, Hald JK (1989) Aneurysms of the vein of Galen: embryonic considerations and anatomical features relating to the pathogenesis of the malformation. Neuroradiology 31:109-128 50. Rodesch G, Lasjaunias P, Terbrugge K, Burrows P (1988) L6sions vasculaires art6rioveineuses intra crfiniennes de l'enfant. Place des techniques endovasculaires fi propos de 44 cas. Neurochirurgie 34:293-303 51. Schneider SJ, Wisoff JS, Epstein FJ (1992) Complications of ventriculoperitoneal shunt procedures or hydrocephalus associated with vein of Galen malformation in childhood. Neurosurgery 30:706-708 52. Seidenwurm D, Berenstein A, Hyman A, Kowalski H (1991) Vein of Galen malformation: correlation of clinical presentation, arteriography, and MR imaging. AJNR 12:347-354 53. Suarez JC, Viano JC (1989) Intracranial arteriovenous malformations in infancy and adolescence. Child's Nerv Syst 5:15-18
54. Tamaki N, Ehara K (1991) Arteriovenous malformations. Indications and strategy for surgery. In: Raimondi AJ, Choux M, Di Rocco C (eds) Cerebrovascular diseases in children. Springer, Berlin Heidelberg New York, pp 59-74 55. Willinsky R, Lasjaunias P, Terbrugge K, Burrows P (1990) Multiple cerebral arteriovenous malformations (AVMs). Review of our experience from 203 patients with cerebral vascular lesions. Neuroradiology 32:207-210 56. Wisoff JH, Berenstein A, Choi IS, Friedman D, Madrid M, Epstein F (1990) Management of vein of Galen malformations. Concepts Pediatr Neurosurg 10:137-155 57. Yasargil MG (1988) Microneurosurgery. Thieme, Stuttgart New York pp 393-396 58. Zampella EJ, Aronin PA, Odrezin GT, Duvall ER (1988) Conservative management of thrombosed vein of Galen malformations. Pediatr Neurosci 14:264-271 59. Zerah M, Garcia-Monaco R, Rodesch G, Terbrugge K, Tardieu M, De Victor D, Lasjaunias P (1992) Hydrodynamics in vein of Galen malformation in 43 cases. Child's Nerv Syst 8:111-117
O v e r the past 10 y e a r s Dr L a s j a u n i a s and his group have b e e n a c t i v e l y inv o l v e d in the n e u r o r a d i o l o g i c a l m a n a g e m e n t o f a h u g e n u m b e r o f children and infants with a r t e r i o v e n o u s m a l f o r m a t i o n s (AVMs). The results p r e s e n t e d c l e a r l y d e m o n s t r a t e that e n d o v a s c u l a r t r e a t m e n t is one o f the best p r i m a r y forms o f treatment o f A V M s in children. H o w e v e r , D r L a s j a u n i a s also p o i n t s out that the m a n a g e m e n t o f such c h i l d r e n requires a large m u l t i d i s c i p l i n a r y t e a m to a l l o w the most suitable and effective treatm e n t in any situation. The r e v i e w o f
the literature clearly shows that there are few reports w o r l d w i d e on d e a l i n g with A V M m a n a g e m e n t in childhood. Thus this p a p e r discusses p o i n t b y point the role o f surgical t h e r a p y versus e n d o v a s c u l a r a p p r o a c h e s . This r e p o r t brings an additional focus on the different p e d i a t r i c age groups with t o p o g r a p h ical details o f A V M s and also the p r o b l e m o f surgical a p p r o a c h e s . A n o t h e r part o f the p a p e r presents the different i n t e r v e n t i o n a l p r o c e dures and their a s s o c i a t e d mortality.
In summary, this p a p e r is a further m i l e s t o n e in the current literature o f A V M in children. B a s e d on the enormous e x p e r i e n c e o f Dr L a s j a u n i a s and his team, and the huge n u m b e r o f patients studied and treated, prim a r y e n d o v a s c u l a r treatment is suggested, f o l l o w e d b y surgery. The literature r e v i e w strengthens the importance o f this paper. T h o m a s J. Vogl, Berlin