Neuro -radiology

Neuroradiology (1990) 31:471 475

@ Spiinger-Verlag 1990

Posterior fossa arteriovenous malformations Angioarchitecture in relation to their hemorrhagic episodes R. Garcia Monaco, H. Alvarez, A. Goulao, Ph. Pruvost, and P. Lasjaunias H6pital de Bic6tre, Unit6 de Neuroradiologie, Kremlin Bicetre C6dex, France

Summary. Classically, posterior fossa arteriovenous malformations (PF-AVMs) have been considered as lesions with high tendency to bleed. However, careful analysis of clinical and autopsy data from the literature demonstrate that in fact the incidence of hemorrhage is similar in infratentorial and supratentonal locations. The clinical perception that most of the diagnosed PF-AVMs had bled does not mean that they have a high risk of hemorrhage. It suggests that bleeding is the dominant way of expression, since other symptoms are uncommon. Furthermore, angioarchitectural analysis is similar in supra- an infratentorial locations. The causes of hemorrhage are probably related to venous anatomic and hemodynamic changes or associated aneurysms as in other intracranial AVMs. Key words: Arteriovenous malformation - Posterior fossa - Angioarchitecture - Intracranial hemorrhage

In recent years the diagnosis of posterior fossa arteriovenous malformation (PF-AVMs) has increased as a result of improved neuroradiologic techniques, raising the clinical incidence from 5% in the early 70's [1, 12] to 15% of all intracranial arteriovenous malformations (AVMs) [1, 5]. The therapeutic approach has also improved, mainly after the development of microsurgical techniques [16] and endovascular procedures [2, 10], diminishing the number of untreatable lesions considerably. In spite of all these advances PF-AVMs are still ill-understood and pose several questions without satisfactory answers, among which is, "Do they really bleed more than supratentorial lesions"? In order to find the answers to this and other questions regarding hemorrhage from PF-AVMs we have reviewed the clinical and angiographic records of our series of 32 patients and analyzed the current available clinical and autopsy data.

Material and methods Between 1981 and 1989, 319 patients with intracranial AVMs were referred to one of us (P. L.) for therapeutic management. Among them, 32patients presented true AVMs in the posterior fossa (10%). Patients with AVMs located at the level of the foramen magnum, or with dural fistulas, cavemomas and venous anomalies (the so-called "venous angiomas") are excluded. The average age of the patients was 30 years. Five belonged to the pediatric population (under 15). The topography and clinical presentation are summarized in tables I and II respectively. All patients were studied by complete selective angiography. CT was also performed in all and MRI in twelve. The following aspects of angioarchitecture were particularly searched for: venous ectasia, venous stenosis, venous thrombosis, associated aneurysms, transdural supply and transmesencephalic supply.

Table 1. Topography of AVMs (as inferred by angiographical, CT and MRI analysis) Topography

Number of patients

Percentage

Cerebellar Brain stem Cerebellar and brain stem Total

24 4 4 32

75% 12.5% 12.5% 100%

Signs/symptoms

Number of patients

Percentage

Hemorrhage Neurologic deficit Incidental finding Cardiac failure a Cranial bruit a

25 3 2 1 I

78% 9.5% 6.5% 3% 3%

Table 2. Presenting symptom

a Pediatric patients

472

Fig. 1. Right vertebral angiogram (lateral view) shows an inferior vermian AVM with an intra-lesional arterial aneurysm (arrow)

Discussion

Fig.2. Left vertebral angiogram (AP view) shows a cerebellar AVM, involving the vermis and right hemisphere. Proximal arterial aneurysm of the fight superior cerebellar artery (arrow)

PF-AVMs are still a hazardous group of vascular lesions, despite the technical improvements in both surgical and endovascular therapy. In clinical practice from 72% to 92% of patients present with intracranial hemorrhage [1, 6, 12, 16]. Though apparantly supporting the classical concept that PF-AVMs have a high tendency to bleed, we believe this is a misconception since it transforms retrospective clinical impressions into an epidemiologic remark. Further support to this observation can be derived by comparing the reported clinical and autopsy data of PFAVMs. Most important clinical series [1, 16], like ours, state that PF-AVMs represent approximatelly 10-15% of all intracranial AVMs. 80% of patients of

Results

Posterior fossa hemorrhage was present in 78% of our patients. There was no difference regarding sex in relationship to bleeding episodes. 24 of the 28 patients (86%) with cerebellar AVMs, presented with hemorrhage and only one with a neurologic deficit. From our 4cases of brain stem AVMs, one presented with hemorrhage, another with an objective cranial bruit, the remaining two having a neurologic deficit as presenting symptom. The angioarchitectural analysis is summarized in table III. 19 of 20 patients with venous stenosis had associated ectasias, as the former produces an obstacle to draining veins. 6 of the 7 patients with associated aneurysms had bled. A "nidus" type of AVM was encountered in 20 patients (62.5%). Micro-arteriovenous malformations (micro AVMs) were diagnosed in 5 patients (15.5%), 4 of which presented with hemorrhage and the latter as an incidental finding in a Rendu-Osler-Weber disease. 4patients (12.5%) had malformations associated with vein of Galen ectasia. Multiple AVMs were observed in 3 patients (10%).

Table 3. Angioarchitecture of PF-AVM a Angioarchitecture

Patients (n=32)

Percent- Patients age who bled (n =25)

Percentage

Venous ectasia Venous stenosis or thrombosis Aneurysms Transdural supply Transmesencephalic supply

24

75%

18

72%

22 7 6

68% 22% 19%

18 6 6

72% 24% 24%

6

19%

4

16%

a The difference between the overall population and the group that have bled is minimal since 78% of them are the same patients

473

Fig.3. a Large cerebellar AVM supplied by superior and inferior cerebellar arteries. Venous drainage through ectatic precentral and Galen vein. b Venous collateral circulation through cortical, occipital, parietal and fronto-orbital veins due to straight sinus thrombosis

this group have had an hemorrhagic episode. Supratentorial lesions, in similar clinical series [1, 16], represent 85-90% of intracranial AVMs, only 40% of which had presented previously an hemorrhagic episode. In contradistinction, most important autopsy series [3, 9] state that approximatly 25% of AVMs are located in the posterior fossa. Applying the incidental remark on hemorrhagic risk cited above to the autopsy incidence of AVMs, one would expect that from every 5 patients with hemorrhagic AVMs, 3 would belong to the supratentorial location and 2 to the posterior fossa. This is in contradiction with the clinical experience in which of 5 patients, 4 were supratentorial and 1 infratentorial. The fact that 80% of the patients with diagnosed PF-AVMs had bled does not mean that this localization has a higher risk of hemorrhage. It suggests that it is the dominant expression of cerebellar AVMs, since headaches and neurologic deficit are uncommon, as in micro-AVMs [14]. Moreover, there is an important group of "silent", asymptomatic PFAVMs, more frequent than supratentorial lesions, as shown by the difference between clinical and autopsy series (10%). These autopsy series [3, 9] were not

correlated with the involvement of the AVM in the cause of the death. The particular cerebellar functional anatomy and the possibility of transfer of functions to neighbouring areas allow normal cerebellar activity even with only 2/3 of the parenchyma. This, although speculative, may be a possible explanation for the clinical tolerance of PF-AVMs. On a prospective aspect, Crawford [5] concluded that a previous hemorrhage, patients older than 60 years and temporal lobe localization were worsening factors to the overall hemorrhagic risk. Infra tentorial location per se, did not represent a higher risk of hemorrhage during a 20 year follow-up in his group of patients. In prospective analysis and long term studies of other authors [4, 7, 8], risk factors did not include posterior fossa located AVMs. The angioarchitectural analysis of our series of PF-AVMs did not show major differences in comparison with supratentorial lesions, already described in a previous report [15]. The presence of associated aneurysms or venous stenosis and thrombosis are strongly related to hemorrhage [3, 11, 15]. Interestingly, these risk factors have equal incidence in both supra- and infratentorial localizations. The possible cause of hemorrhage could be rupture of an associated aneurysm encountered in 24% of our patients. Although the concomitance of AVMs and aneurysm is well known (Figs. 1, 2), some previous reports indicate that it is rare in infratentorial localization [11, 16]. Not so in our series; the in-

474

Fig.4. Superior vermian (Culmen) micro AVM. Superior vermian artery (solid arrow). Superior vermian vein (open arrow)

cidence is similar in the posterior fossa and in the supratentorial region. This different appreciation may result from the small number of AVMs previously reported in that location or the different quality of modern angiographic techniques. It is remarkable that 6 of our 7 patients with associated aneurysms had bled, reinforcing the concept that they deserve rapid treatment. Other causes of hemorrhage could be related to venous hemodynamic changes secondary to obstacles in draining veins [15]. Venous stenosis may be due to kinking of a large ectatic vein or anatomical conditions. Thrombosis is probably related to coagulation disturbances, often associated with AVMs [13]. Whatever the etiology of the obstacle, it increases the venous pressure proximal to the stenosis leading to further venous ectasias and rerouting of blood (Fig.3). The retrograde venous hypertension could favour the rupture of the vessel and the appearence of communicating hydrocephalus by interfeting with CSF reabsorption [10, 15]. The existence of micro-AVMs is another common factor in both brain and PF-AVMs that can be responsible for the hemorrhage. This special type of

AVM has been recently reviewed [13]. They are cortical, with a single "normal"-sized feeding artery and a "normal"-sized draining vein (Fig.4). Their characteristics are similar in both supra- and infratentorial localizations. The classical tendency for small AVMs and micro-AVMs to bleed more than others also affects the retrospective analysis and the fact that these small or micro-lesions can hardly reveal themselves in any other way. Prospective studies do not confirm this increased hemorrhagic tendency [5]. From analyzing the topography in relationship to hemorrhage in the posterior fossa we have noted that 87% of patients with AVMs that involved the cerebellum had bled, while only 25% of pure brain stem lesions presented with hemorrhage. We believe that the latter are more likely to present earlier with neurological deficit, due to the eloquence of the brain stem with its numerous important functions in proportion to its small size, so any pathologic condition will rapidly produce symptoms. This can also be said of the spinal cord. However, the low incidence of brain stem AVMs in our series does not allow us to generalise about their behaviour and further experience is needed.

475

Conclusion

As previously suggested the otherwise classical idea that PF-AVMs have a higher tendency to bleed should no longer be sustained. A careful analysis of clinical and autopsy series clearly demonstrates that they do not bleed more than supratentorial AVMs. All the main characteristics of AVMs are equally seen in both the supra- and infratentorial locations, and angioarchitectural analysis shows no significant differences. PF-AVMs usually present clinically with hemorrhage, since it is almost the dominant way of expression of cerebellar AVMs, Possible causes of hemorrhage could be linked to venous hemodynamic changes or associated aneurysms as in other intracranial AVMs.

References 1. Batjer H, Samson D (1986) Arteriovenous malformations of the brain: natural history study. J Neurosurg 64:849 856 2. Berenstein A, Kricheff II (1979) Catheter and material selection for transarterial embolization: technical considerations. I. Catheters. Radiology 132:619 632 3. Berenstein A, Lasjaunias P (1990) Endovascular treatment of brain, spine and spinal cord lesions. Surgical angiography, vol IV. Springer, Berlin Heidelberg New York Tokyo 4. Cophignon J, Thurel C1, Djindjian tL Rey A, Visot A, Le Besnerais Y, Houdart R (1978) Cerebral arteriovenous malformations. Modern aspects of investigations and treatment. Prog Neurol Surg 9:195-230 5. Crawford P, West C, Chadwick D, Shaw M (1986) Arteriovenous malformations of the brain: natural history in unoperated patients. J Neurol Neurosurg Psychiatry 49:1-10 6. Drake CG, Friedman AH, Peerless SJ (1986) Posterior fossa malformations. J Neurosurg 64:1-10

7. Forster D, Steiner L, Hakanson S (1972) Arteriovenous malformations of the brain. A long-term clinical study. J Neurosurg 37:562-570 8. Graf C J, Perret GE, Toruer JC (1983) Bleeding from cerebellar arteriovenous malformations as part of their natural history. J Neurosurg 58:331-337 9. Jellinger K (1986) Vascular malformations of the central nervous system: a morphological overview. Neurosurg Rev 9: 177-216 10. Lasjaunias P, Berenstein A (1987) Endovascular treatment of craniofacial lesions. Surgical angiography, vol II. Springer, Berlin Heidelberg New York Tokyo 11. Lasjaunias P, Piske R, Terbrugge K, Willinsky R (1988) Cerebral arteriovenous malformations (C. AVM) and associated arterial aneurysms (AA). Acta Neurochir 91 : 29-36 12. Samson D, Batjer H (1985) Arteriovenous malformations of the cerebellar vermis. Neurosurgery 16:3 13. Sutherland GR, King MR, Drake CG, Peerless SJ, Vezina WC (1988) Platelet aggregation within cerebral arteriovenous malformations. J Neurosurg 68:198 204 14. Willinsky R, Lasjaunias P, Comoy J, Pruvost P (1988) Cerebral micro arteriovenous malformations (mAVMs). Review of 13 cases. Acta Neurochir 91:37-41 15. Willinsky R, Lasjaunias P, Terbrugge K, Pruvost Ph (1988) Brain arterio-venous malformations. Analysis of the angioarchitecture in relationship to hemorrhage. J Neuroradiol 15: 225-237 16. Yasargil MG (1988) AVM of the brain, clinical considerations, general and specific operative techniques, surgical results, nonoperated cases, cavernous and venous angiomas, neuroanesthesia. Microneurosurgery, vol IIIb. Thieme, Stuttgart, pp 317-396 Received: 25 July 1989 Dr. P. Lasjaunias H6pital de Bic~tre Unit6 de Neuroradiologie 78 G6n6ral Leclerc F-94275 Kremlin Bic6tre Cedex France