Читать книгу Microneurosurgery, Volume IIIA - Mahmut Gazi Yasargil - Страница 12

3 Pathological Considerations

Pathogenesis

The pathogenesis of angioma is generally attributed to maldevelopment of the cerebral vascular system occurring during the second to fifth stage of Streeter’s craniocerebral vascularization. However, the underlying anomaly ultimately responsible for the vascular malformation still remains a matter of controversy.

Old hypotheses assumed that embryologic forerunners of arteries and veins were separate. Based on meticulous injection techniques Evans (1911) was the first to show that a primary vascular plexus existed as a capillary network, preceding the more definable vascular system. In a process called metamorphosis, fusion of some of the channels of this primordial vascular plexus and dissolution of others takes place and ultimately leads to the differentiation of arteries and veins (Dandy 1928). This process of angiogenesis is controlled by hemodynamic and genetic factors. There is a steady development, not only of the arteries and veins, but also of the capillaries during successive embryological development phases.

All investigators concerned with the problem of pathogenesis of cerebral vascular malformations uniformly accept that an error of development occurs during this very early metamorphotic phase.

Dandy (1928) postulated a retention of primordial vascular connections between arteries and veins. Olivecrona and Ladenheim (1957) assumed an embryonic agenesis of the capillary system, ultimately resulting in discharge of arterial blood directly into the venous system through a tangle of abnormal vessels of varying caliber.

A concept basically similar to that of Dandy was introduced by Kaplan and Meier (1958). Based on observations made in specimens obtained at autopsy, they concluded that arteriovenous malformations within the cerebral hemisphere represent a perpetuation of a primitive arteriovenous communication, which otherwise would be replaced by an intervening capillary network during the normal embryological development of the cerebral vascular system.

Hamby (1958) approached the problem from a hemodynamic standpoint, concluding that the basic characteristic of arteriovenous malformations is a lack of vascular resistance in the area involved by the lesion. Since the normal cerebral vasculature resistance is provided by the capillary bed, Hamby’s concept is similar to that of Olivecrona and Ladenheim and based on the agenesis of capillaries.

Gold et al. (1964) and Lagos (1977) recognized two types of vascular malformation: 1) a direct end-to-end anastomosis between the arteries and veins of normal structure, representing arteriovenous fistulae and 2) a network of poorly differentiated and immature vessels interposed between the arterial and venous system, representing typical arteriovenous malformations.

Stein and Wolpert (1980) and Warkany et al. (1984) assumed an arrest of normal development of primitive arteries, capillaries and veins, resulted in the formation of direct arteriovenous communications through immature, poorly differentiated vessels, without an intervening capillary bed.

Parkinson and Bachers (1980) maintained that the essential feature of arteriovenous malformations is a shunt responsible for the short-circuiting of the arteriocapillary bed and proposed the descriptive definition of a “congenital arteriovenous fistulous malformation” occurring as a consequence of a local angioblastic error.

Based on Sabin’s (1917) original concept of the development of the primitive vascular plexus, Garretson (1985) recently proposed that AVMs arise from persistent direct connections between the future arterial and venous sides of the primitive vascular plexus, with failure to develop an interposed network.

In summary, most of the theories developed to explain the origin of cerebral vascular malformations have in common the hypothetical concept of total agenesis, or poor development of the capillary network. It is known, however, that normal angiogenesis takes place in a capillarofugal direction and that the predisposing factor for the formation of arteries and veins lies within this primordial capillary network. If there is a primary agenesis of the capillary network and therefore of the driving force for the development of arteries and veins, then this territory must be ultimately avascular.

If, however, the theory of primary capillary agenesis is not correct, one must assume a secondary destruction or disappearance of capillaries, in order to explain the absence of a capillary network as the pathogenetic mechanism for vascular malformations. Such a secondary destruction would have to occur through the action of a factor having the capacity to destroy capillary vessels after arteries and veins have been formed from them. In such a situation the arteries and veins would then form direct communications.

A capillary destroying factor has not yet been found. Also, if this theory of secondary destruction of capillaries is correct, one would expect to see only cases with direct arteriovenous fistulae, rather than all the commonly known varieties of AVMs in which coiling convoluted vessels are interposed between arteries and veins.

For this reason it seems appropriate to discuss another hypothesis: There is neither a primary agenesis nor a secondary destruction of capillaries, but a local or regional disease of capillaries. In a given primitive vascular territory, the normal development of capillaries is disturbed, however, these capillaries do not disappear entirely, but proliferate and thereby develop metamorphotic, dysplastic vessels (Luschka 1854, Dandy 1928). This disease may be defined as a ‘proliferative capillaropathy’ of unknown origin (Fig 3.1). It is characterized by maldevelopment of an area of the primordial capillary plexus into metamorphotic vessels. These vessels do not fulfil the histologic criteria of arteries, veins or capillaries. It is, in fact, well known, that it is difficult if not impossible to typify histologically the vessels comprising the core of a vascular malformation. These vessels have been called “unidentifiable type of vessels” by Hamby (1958) (Fig 3.2) and “structural hybrids” by Burger and Vogel (1976).

Fig 3.1A–G Artist’s drawing of the different types of cerebral vascular malformations.

A Arterial malformation.

B Arteriovenous fistulous malformation.

C Arteriovenous plexiform malformation.

D Arteriovenous plexiform micro-malformation.

E Cavernous malformation.

F Capillary malformation (telangiectasia).

G Venous malformation.

Fig 3.2 The unique drawing of Hamby (1958) showing the complex vascular composition of AVM-nidus.

In a histologic study of three cases of arteriovenous malformations, Sorgo (1938) classified the vessels constituting the malformation into three main types, according to the composition of their wall. He also found similarities between the wall of vessels composing arteriovenous malformations and the wall of normal embryonic vessels. Based on his observations he postulated that at least one of the described types of vessels may well arise from capillaries.

The results of recent electron microscopic studies are in accordance with our proposed concept. Meyermann and Yaşargil (1981) found that the ultrastructural composition of small vessels of 41 surgically obtained arteriovenous malformations could be divided into two distinct types; vessels with a closed and vessels with a fenestrated endothelial cell layer. This second type of vessel, characterized by a fenestrated endothelial coat is clearly abnormal, since fenestrations of endothelial cells do not occur in the normal brain vasculature with exception of the area postrema, choroid plexus, pineal and pituitary glands, intercolumnar tubercle, and certain nuclei within the hypothalamus (Lee 1971). Another observation of this study was the sprouting of new capillaries in the fibrotic arachnoid surrounding superficial pathological vessels. This finding supports the concept of a proliferative capillaropathy (Figs 3.3, 3.4).

Depending on the extension and distribution of the capillary disease involving the primitive vascular plexus, vascular malformations may therefore be defined as localized, multiple or diffuse collections of metamorphotic vessels, abnormal in number, in structure and in function.

The result of this primary disease of capillaries is a mal-production and therefore a mal-formation of both arteries (or arterioles) and veins (or venules), i.e. a metamorphotic angiodysplasia or capillaropathy.

Fig 3.3A–B A Sinusoid-type vessels of AVMs are coated by fenestrated endothelial cells. The fenestrae are indicated by arrows. In the normal cerebral vasculature this type of endothelial coat is only present in certain distinct areas of the CNS. The cytoplasm of the endothelial cells is filled with cross-sectioned filaments and some vacuoles. The arrowhead indicates a so called Weibel-Palade body. This organelle can only be found in endothelia, and is a rare feature in a normal cerebral vessel wall. Bar = 1 μm B Although some gaps in the endothelial cell layer of AVM are demonstrated as in A, some cell contacts of adjacent endothelia are tight as seen in normal cerebral vessels. Bar = 1 μm. By courtesy of Dr. R. Meyermann.

Fig 3.4 Arteriovenous malformation surgically resected from the left occipital lobe of a 24 year old female patient (see Fig 3.78). Note the considerable variation of vessel size with dilated, partially arterialized veins (V) and occasional small arteries (arrow). In the lower half malformed compact vessels with little or no intervening parenchyma prevail, thus resembling a cavernous angioma (Elastica van Gieson, x10). By courtesy of Prof. P. Kleihues, Zurich.

From such a dysplastic vascular plexus may arise all known and angiographically observable types of “malconnection”. Persistence of the embryonal plexus will lead to a pure plexiform type containing vessels without direct arteriovenous fistulae. A gradual but incomplete destruction of the embryonal plexus will result in a mixed type of malformation, composed of both plexiform convolutions and direct arteriovenous fistulae. The preponderance of plexiform or fistulous vessels depends on the degree of destruction of the plexus. Gradual complete destruction of the plexiform parts will ultimately result in pure, direct arteriovenous fistulae (Table 3.1).

It is also evident that vascular resistance will be highest in the pure plexiform and lowest in the pure fistulous types, explaining the angiographic observation that the flow through plexiform lesions is slower than through fistulous lesions.

The different types of arteriovenous malformations may be demonstrated angiographically. Based on their angiographic appearance, arteriovenous malformations may therefore be divided into three main types (Table 3.2).

Traditionally, descriptions of cerebral vascular malformations used in classifications include 1. the composition of the vascular wall, 2. the presence or absence of an intervening brain parenchyma between the vascular spaces of the malformation, and 3. the state (normal or gliotic) of the intervening neural tissue. Based on these morphological parameters vascular malformations are divided into four main types: 1. arteriovenous malformations 2. venous malformations 3. cavernous malformations and 4. capillary malformations (or telangiectasias) (McCormick 1966).

Despite this attempt to separate various different forms, certain observations support the hypothesis of a single underlying primary lesion.

Transitional forms exhibiting the histologic characteristics of more than one of the above mentioned types are sometimes encountered within the same malformation. It is, in fact, difficult to distinguish histologically between telangiectasia and venous angioma. Also telangiectasias have been reported to be a component of venous angiomas (McCormick 1966, Manuelidis 1950). Combinations of cavernous and telangiectasias (Roberson et al. 1974), as well as venous angiomas and arteriovenous malformations (Huang et al. 1984), have been reported to occur within the same malformation. Also multiple lesions of different histologic types can occur in the same individual (McCormick 1966).

Although absence of capillaries has usually been described as the hallmark of arteriovenous malformations, abnormal proliferation of capillaries may be observed within the malformation or even in adjacent tissue. Hamby (1958) in a unique histologic study of a specimen of an arteriovenous malformation of the brain, demonstrated not an agenesis or absence of capillaries, but a multitude of different types of capillary-like vessels, clearly distinguishable from the entering arteries and the draining thin-walled tortuous veins. These capillary-type vessels found in the central core of the malformation form a complex of coiling and intercommunicating vessels (see Fig 3.2).

Dilated capillaries or capillary-like spaces are found in telangiectasias, which are therefore also called capillary malformations, as well as in cavernomas (Huang et al. 1984). By the same reasoning certain vascular malformations of the subcutaneous tissue are also called capillarovenous malformations (Merland et al. 1983). In histologic studies of Cabanes et al. (1979) cases of venous angioma with a clear participation of capillaries are demonstrated.

We should also, perhaps, remember Virchow’s statement of 1851 – that “one type of angioma can transform into another by changes in flow and pressure or by cellular proliferation”.

Histologically, the presence or absence of intervening neural parenchyma, as well as its state (normal or gliotic) are used as parameters for classifying vascular malformations. Usually, arteriovenous malformations surround gliotic tissue, venous malformations and telangiectasias have normal intervening tissue, and cavernous malformations contain no intervening parenchyma. Both histologic studies and intraoperative observations show, however, that an intervening neural parenchyma and even gliosis within it may occur with all types of cerebral vascular malformations.

Cavernous malformations are classically described as being compact, with the vascular spaces being contiguous with one another and lacking intervening tissue. During operation on such lesions, however, one may observe through the operating microscope, small cavernous spaces located at the periphery of the mass and being clearly separated from it by brain parenchyma.

In a histologic study, Manuelidis (1950) clearly demonstrated neural tissue between the vascular spaces of an otherwise typical case of cavernous angioma.

A finding common to all types of cerebral vascular malformation is spontaneous thrombosis, occurring most frequently in the venous space of the lesion. Although such spontaneous thromboses have been most often reported in cases of true AVM, they clearly also occur with the other types, especially venous and cavernous malformations.

The histological character of the resected lesions and the relative frequency are given in Table 3.3.

Table 3.3 Histological findings in 398 AVMs

Mixed type	374 cases	(94.0%)
More arterial	12 cases	( 3.0%)
More venous	12 cases	(3.0%)
	398 cases

Not investigated 16 galenic and 2 fistulous lesions.

Hemorrhage, which most frequently occurs in arteriovenous malformations, may also be observed with other types of vascular anomalies. Microscopic hemorrhages with foci of hemosiderin laden macrophages are frequently found in arteriovenous malformations, but may also be seen in venous, cavernous and even capillary-type malformations.

From these pathologic-anatomic observations it becomes evident that cerebral vascular malformations have characteristics in common with respect to their histologic nature, their vascular composition, and regressive changes, irrespective of their type.

Using cerebral angiography, the different morphologic types of vascular malformation described above can usually be distinguished (Tables 3.2, 3.4). Arteriovenous malformations typically appear during the arterial phase of the angiogram and are characterized by large feeding arteries, a more or less compact conglomeration of coiled vessels and prominent draining veins. Venous angiomas most frequently appear during the venous phase and are characterized by numerous dilated, linearly arranged medullary veins, producing an umbrella-shaped configuration and converging towards a markedly dilated central parenchymal vein. Cavernous angiomas may cause an avascular mass effect, but remain invisible with usual angiographic techniques, owing to their slow circulation and the lack of prominent feeding arteries. A blush, representing pooling of contrast material within the vascular spaces of the lesion, may however appear, if either prolonged injection angiography (Numaguchi and Nishikawa 1979) or a repeated injection series (Huang et al. 1984) is performed. In telangiectasias, angiography is usually negative, owing to their small size and their slow circulation time. Occasionally, however, telangiectases may show a small stain or blush during the venous phase of the angiogram (Huang et al. 1984).

Although the different morphologic types of cerebral vascular malformation are distinguishable on angiography, certain observations support the concept of a single underlying cause, common to all types of vascular malformation. One may occasionally demonstrate both transitional forms as well as the coexistence of two or more different types of lesion within the same vascular malformation. A pure venous type of malformation was demonstrated angiographically in 1958 by Krayenbühl and Yaşargil. The histological examination showed no arterial component in the lesion (Fig 3.5). According to Huang et al. (1984) 14% of cases of venous angioma contain fine arteries which form a reticular blush in the arterial phase of the angiogram. This indicates the presence of an arterial or low-flow arteriovenous component in certain venous angiomas. It clearly contradicts the classical definition, according to which venous angiomas lack arteriovenous shunts or an arterial component and become visible only in the late venous phase of the angiogram. Similar observations on venous angiomas with an arterial component were reported by Moritake et al. (1980). The close embryological relationship between apparently different pathologic entities such as venous angioma and arteriovenous malformation is demonstrated in a case reported by Huang et al. (1984), in which a typical venous angioma contained within it, two small arteriovenous malformations. Furthermore, all three coexistent vascular malformations had a common venous drainage! In cases of telangiectasia, Rosenbaum (1974) has observed an early appearing blush and early draining veins, suggesting the presence angiographically of small or cryptic arteriovenous malformations.

Fig 3.5A–C This may be the first angiography demonstration of a venous angioma. (54 year old male presenting with subarachnoid hemorrhage. From Krayenbühl, H., M. G. Yaşargil: Series Chirurgia Geigy 4: 76 1958.)

A Normal arterial phase of carotid angiography.

B Venous phase of carotid angiogram after a SAH shows the lesion in the right temporal lobe. It drains into the dilated basal vein of Rosenthal. In 1958 this malformation was called “Arteriovenous malformation visible only in the venous phase”.

C Histological examination shows venous malformation with arterial components.

It is interesting to note, that during operation for cavernous angiomas, one regularly observes with the help of the operating microscope, slightly dilated arteries entering the cavernous space of the lesion and thus indicating an arterial participation in their supply.

In cases of venous angioma it is a frequent angiographic finding that the medullary or subependymal veins adjacent to the angioma are hypoplastic or even absent and that the adjacent superficial or cortical veins may be poorly developed. Also hypoplasia of the internal cerebral veins, poor development or even absence of certain major subependymal veins and a paucity of superficial cortical veins have occasionally been observed (Huang et al. 1984). Veins pursuing an unusual course, most probably representing persistent fetal or intrauterine venous structures, are frequently observed angiographically in such cases (Huang et al. 1984).

Similar anomalies of the venous system may also be observed in cases of arteriovenous malformation. Unfortunately, angiographic study of the venous drainage patterns of cerebral vascular malformations has been generally neglected (see below). Review of our own angiographic material disclosed an unsuspected 30% incidence of associated anomalies in the venous drainage system of AVMs similar to those reported to occur with venous angiomas (see Vol. III B, Table 9.2).

Our own operative findings have also demon strated clear overlaps of histological types of malformation within single lesions. There have been AVMs with a predominance of arterial or venous components, cavernous malformations with definite feeders bearing aneurysms, capillary cavernomas with no visible arterial or venous connections and virtually isolated from surrounding tissue by firm encapsulation, and venous malformations with arterial components found at operation and confirmed histologically but which could not be demonstrated angiographically.

Upon comparing the clinical features of the different types of cerebral vascular malformation, it becomes evident, that with the exception of a bruit and some symptoms associated with steal phenomena, which exclusively occur with certain high-flow arteriovenous malformations, all other symptoms such as epileptic seizures, hemorrhage, progressive neurological deficit, and headache, may occur with any type of vascular malformation albeit with some variations in incidence (Table 3.5).

There is therefore pathological, anatomical, angiographic, surgical and clinical evidence for a common underlying pathogenesis of all forms of cerebral vascular malformation, based upon a disease of capillaries. The seemingly distinct forms of cerebral vascular malformation described by pathologists, diagnosed angiographically by neuroradiologists and operated upon by neurosurgeons represent nothing more than different manifestations of the same disease.

This concept supports the theory of van Bogaert (1935), who doubted that the different types of cerebral vascular malformations represent different disease entities and expressed the opinion that there is only one Angioma-Disease (maladie angiomateuse) with a variety of subgroups. He was able to explain the pathogenesis of this disease by assuming a disturbance in the development of small vessels as the underlying mechanism.

Classification of Vascular Malformation

“The classification of the vascular malformations of the brain has been the subject of considerable discussion and the extensive literature on this topic reflects a varying and, at times, confusing nomenclature.” (Bebin and Smith 1982, p. 13). The confusion continues and applies not only to vascular malformations of the brain but also to those of all other organs. We agree with Mulliken (1983) “the words to describe the common vascular birthmarks reflect our ignorance of their pathogenesis”. There are majors problems with both nomenclature and classification.

Nomenclature

A. Both Greek and Latin roots are used: Vascular malformation (Latin roots), Angiodysplasia (Greek roots).

B. The suffix oma (= neoplasm) is commonly used not only for true vascular tumors such as hemangioblastoma, but also for vascular malformations. The use of the suffix osis (e. g. “angiomatosis”) has sometimes been inappropriate. The term should be reserved for diffuse or multiple lesions only.

At the present time the English version of “malformation” has found general acceptance and there is little point in entering further into sophisticated linguistic struggles.

Classification

As ever more sophisticated means of studying vascular malformations have developed, systems of classification have diversified from the early descriptive terms based purely on gross morphological observation. In some instances, old terminology has been retained, in others changed and in yet further (often simultaneous) publications regrouped under different headings.

Noran presented and discussed all the proposed classifications in the literature up to 1945 and concluded: “a comprehensive evaluation of the literature is warranted in order that one may arrive at some correlation between these various nomenclature and classification.”

Within the last 40 years further new concepts have been proposed. Table 3.6a contains some of the more notable historic and modern classifications, and shows the development of thinking regarding the malformations.

Virchow (1863) conducted his own thorough studies and described 4 main types of malformation and stated, as early as 1851: “one type of angioma can transform into another by changes in flow and pressure or by cellular proliferation.”

The venous anomalies, and plexiform angioma of Dandy’s classification (1928) would nowadays be called AVM, and the cyst with angioma in the wall a hemangioblastoma. We assume that he did not describe any “venous angiomas” as now recognized by Huang et al. (1984) and McCormick (1985).

Table 3.6a

Virchow (1863) 1. Angioma simplexTelangiectasia (can change to cavernoma) 2. Cavernous angioma 3. Racemous angioma a. Arterial (aneurysma anastomoseon) b. Venous angioma c. Arteriovenous aneurysm 4. Lymphangioma Dandy (1928) 1. Angioma a. Cyst with angioma in the wall (actually angioblastoma) b. Cavernous angioma c. Plexiform angioma (nowadays a form of AVM) 2. Arteriovenous aneurysm (nowadays a form of AVM) 3. Venous abnormalities (nowadays also AVM) Cushing – Bailey (1928) 1. Hemangioblastoma (true neoplasm) a. Cystic b. Solid α capillary β cellular γ cavernous (nowadays = cavernous angioma) 2. Angiomatous malformation a. Telangiectasias b. Venous angiomas c. Arterial or arteriovenous angiomas (AVM) Bergstrand – Olivecrona – Tönnis (1936) 1. Angioma cavernosum 2. Angioma racemosum a. Telangiectasias b. Angioma capillare et venosum calcificans (Sturge-Weber disease) c. Angioma racemosum arteriale d. Angioma racemosum venosum e. Aneurysma arteriovenosum 3. Angioblastoma, angioreticuloma or Lindau tumors 4. Angioglioma (!) Turner – Kernohan (1941) (spinal cord) 1. Vascular malformations a. Telangiectasia b. Angioma or hamartoma α angioma venosum β angioma arteriovenosum or γ angioma arteriale 2. Vascular neoplasms a. Capillary α capillary hemangioma β hemangioendothelioma γ capillary hemangioblastoma b. Cavernous α cavernous hemangioma β cavernous hemangioblastoma c. Hemangiosarcoma Wyburn-Mason – Holmes (1943) (spinal) 1. True tumors a. Hemangioblastoma α angioreticuloma β extradural hemangioblastoma 2. Malformations a. Telangiectasia b. Venous malformation α secondary venous anomalies β venous angioma c. Arteriovenous angioma d. Arterial anomalies Manuelidis (1950) 1. Telangiectasia a. Primary b. Secondary 2. Cavernous hemangioma 3. Venous hemangioma 4. Arteriovenous hemangioma Zülch (1951) 1. Angioreticuloma 2. Malformation a. Cavernous angioma b. Racemous capillary angioma (telangiectasia) c. Capillar et venous angioma (Sturge-Weber) d. Venous angioma e. Arteriovenous aneurysmatic angioma Asenjo (1953) I. Congenital lesions A. Expansive malformation a. Arteriovenous aneurysm b. Arterial racemous aneurysm c. Venous racemous aneurysm B. Angiosis d. Congenital arterial aneurysm e. Meningeal varix f. Sinus pericranii II. Acquired lesions A. Aneurysms a. Arteriosclerotic b. Mycotic c. Syphilitic B. Carotid-cavernous fistula C. Traumatic aneurysms III. Tumors A. Hemangioblastoma a. Benign b. Malignant B. von Hippel-Lindau disease C. Angiomatous meningioma Pluvinage (1954) I. Angioreticuloma II. Angioma 1. a. Cavernous angiomab. Telangiectasia 2. Sturge-Weber 3. Venous angioma a. Cerebral varix b. Racemous venous angioma c. Peleton de veines (!) 4. Arterial angioma a. Racemous arterial angioma b. Arteriovenous aneurysm Olivecrona – Ladenheim (1957) Etiology 1. Acquired 2. Congenital a. Anomalous arteriovenous b. Angiomatous arteriovenous Pathology 1. Cavernous 2. Racemous a. Telangiectasia b. Sturge-Weber c. Venous racemous d. Arterial racemous e. Angiomatous arteriovenous Russel – Rubinstein (1963) 1. Hemangioblastoma 2. Vascular malformation a. Capillary telangiectasia b. Cavernous angiomas c. Venous and arteriovenous malformation McCormick (1985) (in Fein and Flamm) I. Angioblastoma II. Angiomas 1. Venous angiomas 112 cases 2. Capillary angiomas (telangiectasias) 41 cases 3. AVM 11 cases 4. Cavernous angiomas 5 cases 5. Transitional 4 cases Classification of Plastic Surgeons Kaplan (1983) A. Stage 1 (undifferentiated capillary network) 1. Capillary hemangioma 2. Cavernous hemangioma B. Stage 2 (retiform plexus) 1. Diffuse microfistula 2. Localized macrofistula C. Stage 3 (mature vascular malformation) 1. Venous hemangioma 2. Venous hypoplasia (Klippel-Trénaunay syndrome) 3. Hemangiolymphangioma (vascular hamartoma) Spira (1983) A. Benign hemangiomas 1. Typical a. Capillary hemangioma b. Cavernous hemangioma c. Mixed-combined hemangioma d. Port-wine stain – nevus flammeus e. Angioma racemosum f. Angiokeratoma (Mibelli) 2. Atypical a. Sclerosing hemangioma b. Pyogenic granuloma c. Spider telangiectasia (nevus araneus) d. Glomus tumor e. Hemangiopericytoma f. Juvenile nasopharyngeal angiofibroma g. Venous lakes B. Syndromes – diseases 1. Rendu-Osler-Weber syndrome 2. Sturge-Weber-Dimitri syndrome 3. von Hippel-Lindau disease 4. Maffucci syndrome 5. Blue Rubber Bleb syndrome 6. Kasabach-Merritt syndrome 7. Klippel-Trenaunay syndrome C. Malignant hemangiomas 1. Angiosarcoma 2. Kaposi sarcoma 3. Dermatofibrosarcoma protuberans Classification of Neuroradiologists Merland et al. (1983) 1. Pure arterial dysplasia (2 cases) 2. A-V dysplasia (macroscopic shunt) a. Simple direct A-V fistulavertebro-vertebral, vertebro-jugularcarotido-cavernous, carotido-jugular b. A-V malformation (60 cases) 3. Capillary and capillary-venous malformation (26 cases) a. Pure capillary (Rendu-Osler) b. Capillary-venous malformation 4. Venous and cavernous ectasias (100 + 4 cases) 5. Additional types a. Unmature angioma of the newborn b. Portwine stain angioma c. Unusual angiomas Hemodynamic Classification 1. Active (large blood flow, direct A-V fistula) high flow 2. Inactive vascular Huang et al. (1984) I. Those that involve feeding arteries and draining veins (easily demonstrable angiographically) 1. Superficial type (pial or superficial AVM): involving mostly the cortical gray matter (and subjacent white matter) 2. Deep or central type (deep or central AVM): involving the subcortical (or central) gray matter and the adjacent white matter 3. Medullary type (AVM with a medullary component): involving primarily the medullary arteries and veins Classical pyramid-shaped AVMs are mostly a combination of the superficial type and the medullary type II. Those that primarily involve capillaries 1. Cavernous capillary malformation 2. Rendu-Osler-Weber disease 3. Louis-Bar syndrome III. Those that primarily involve veins 1. MVM a. Without an arterial component. Sturge-Weber disease should also be included here b. With an arterial component. (This should not be confused with an AVM with medullary component) 2. Cavernous venous malformation 3. Phlebectasia or varix (most of these cases, if not all, are MVMs) IV. Any combination of the above

Cushing and Bailey (1928) were the first to separate two groups:

I. Angioblastoma (true neoplasm),

II. Angiomatous malformation.

They did not consider cavernous angioma as a separate entity and listed it under angioblastoma. Their venous angiomas would be called AVMs today.

Bergstrand et al. (1936) added to the neoplastic group the angioglioma of Roussy and Oberling. These are to a large degree still not accepted, yet appear to have been occasionally identified (Bonnin et al. 1983). Bergstrand doubted the existence of a true arterial aneurysm as described and illustrated by Simmonds (1905) (Figs 3.5, 3.6).

Fig 3.6 Coloured artistic drawing of Simmonds (1905) unique case of a pure arterial malformation in a 53 year old man who died of an acute intracerebral hematoma. At autopsy a medium-sized arterial convolution was found. It was histologically verified to be purely arterial.

Historical review shows that 4 or 5 different vascular diseases (angioblastoma, telangiectasia, venous, cavernous and arteriovenous angiomas) were recognized in the very earliest studies and then gradually identified as being separate entities.

McCormick’s classification (Table 3.6b) is similar to earlier works and he no longer appears to regard “varix” as a special entity. His contribution was important in that he provided statistics regarding the numbers of different types of malformation and reopened the discussion on the relationship of transitional types of angiomas to vascular tumors as originally mentioned by Virchow.

Huang et al. (1984) noted the wide acceptance of Russel and Rubinstein’s (1963) classification. However, they pointed out the disadvantages of attempting to differentiate histologically between many cavernous venous malformations and venous angiomas and showed that some areas within venous angiomas may be similar to capillary malformations or even an AVM. The classification of Huang et al. has put forward new and important elements for consideration.

We include the classification of Merland et al. (1983) as a very stimulating view of the external angiomas, seen from the perspective of the interventional neuroradiologist.

The classification of Kaplan (1983), Spira (1983) show up the similar problems experienced by plastic surgeons in describing cutaneous malformations.

The Author’s Classification

Our own classification is based on the relative preponderance and contribution of the various vascular elements, arteries, veins, capillaries, and abnormal channels (Table 3.6c). There may run a spectrum from theoretically completely arterial lesions to completely venous lesions and from large fistulae to extensive convoluted vessels. While the lesions can conveniently be grouped into four primary headings, there are at each level examples of transitional lesions, e. g. AVMs with slow flow or venous malformations with increased flow. Part of the definition of the lesion must rest with dynamic properties related to flow and shunting, which cannot be examined by the pathologist in the resected specimen or at autopsy. We propose the following classification more for practical use in neuroradiology, neurology and neurosurgery but hope that neuropathologists will be stimulated to undertake further investigation of these lesions.

Table 3.6c Authors classification

I. Vascular neoplasms 1. Hemangioblastoma a. Cystic b. Solid 2. Angioglioma (mixed hemangioblastoma and glioma) 3. Angioblastic meningioma 4. Hemangiopericytic meningioma (hemangiopericytoma of the meninges) 5. Angiosarcoma II. Malformations 1. Telangiectasia 2. Cavernous malformation a. Intrinsic b. Extrinsic 3. Venous malformation a. Cortical b. Subcortical (medullary) α superficial β deep 4. Arteriovenous malformation a. Plexiform (dilated, tortuous pathological vessels with thickened or thinned (or combined) walls, arteriectasia, aneurysms, phlebectasia, varices; they can be cryptic, occult, micro, moderate, large or giant in size.They may be uni- or multilocular.They may have a mono-nidus with mono- or multi-compartments) b. A-V Fistula (direct communication between arteries and venous channels (veins and sinuses) without the interposition of a convolue. α simple: – carotid = cavernous, carotid = jugular, – MCA = v. Labbé or v. Trolard or Sylvii, – ACA = inferior sagittal sinus, – pericallosal artery = v. Galen, – PCA = v. Galen, PCA – transverse or sigm. sinus, – vertebra = vertebral, vertebra = jugular, – AICA = lateral rec. vein or petrosal sinus, – SCA = transverse sinus, – basilar artery = galenic vein β complex: – pericallosal + PCA + MCA = v. Galeni, – MCA + dural branches = hero-philic sinus or SSS, – PCA + dural branches = hero-philic sinus or transverse sinus. c. Transitional type between a–c α more fistula > less plexiform (network) β more network > less fistula 5. Transitional malformations Combinations 1+2, 1+3, 2+3, 1+2+3, 3+4 (Huang) 1+3+4 (Huang) III. Vascular malformation and vascular tumor associated with phacomatosis (Phacomatotic angiomatous diseases) Neurocutaneous syndromes 1. Angioblastoma (angioreticuloma) (von Hippel-Lindau)(angioblastomatosis) 2. Encephalofacial angiomatosis = neuro-oculo-cutaneous(Sturge-Weber-Krabbe-Dimitri) 3. a. Hereditary hemorrhagic telangiectasia (Rendu-Osler-Weber) (cutaneous, mucosal and visceral capillary malformation) b. Ataxia telangiectasia (Louis-Bar) facial naevus, cerebellar angioma, angioma of the choroid of the eye, defective immune globulin system of the IgA class 4. Encephaloretinofacial angiomatosis } (Wyburn – Mason)(Bonnet–Dechaume-–Blanc) 5. Orbitothalamoencephalic angiomatosis (Brégeat syndrome) 6. Diffuse corticomeningeal capillovenous familial angiomatosis (non calcifying) (Divry-Van Bogaert) 7. Cutaneomeningospinal angiomatosis (Cobb) 8. Congenital venous dysplasia (extremities, spinal) (Klippel-Trénaunay-Weber syndrome) 9. Glomangiomatosis (glomus tumor) (Bailey) 10. Dyschondroplasic hemangioma (Maffucci-Kost) 11. Angiokeratosis naeviformis (?) 12. Extensive cavernous hemangioma + thrombocytopenia and purpura (Kasabach-Merritt syndrome) 13. Blue Rubber Bleb syndrome 14. Malignant hemangioma

Location of AVMs

As the disorder occurs quite early in development, it would be anticipated that a vascular malformation may be present in any tissue layer from skin to ependyma. The primitive vascular plexus forms from cephalad to caudad with the meninges forming in the 30 to 60 mm period cleaving the vascular plexus into vessels destined to supply the dura and skull and those destined to supply the brain and choroid plexuses. As pointed out by Padget, the superficial vascular plexus supplying the scalp does not appear until after the membranous skull is complete. Nevertheless, the vascular plexus supplying the face is in continuity with the brain and at times connections exist with the scalp through the skull especially in the venous system. It would follow that association of scalp and brain vascular malformations would be rare, association of facial, retinal, and brain malformations more common, and dural and brain malformations even more frequent.

If the vascular malformations do indeed represent the result of an early deep fistula with remote effects, it is surprising that the lesions are not multiple more often, as many areas of the vasculature would be at risk.

Garretson (1985) has categorized cerebral AVMs (which, incidentally are seen most frequently involving branches from the MCA, less often the ACA, and least frequently the PCA) as those involving the epicerebral, transcerebral and subependymal circulations (see also section on microcirculation). He notes that AVMs involving only the transcerebral (long perforating) arteries are not visible on the surface of the hemisphere yet arterialized veins may frequently be seen owing to the anastomoses between the transcerebral and epicerebral veins. A rare group of AVMs remains confined to the pial surface of the brain stem.

Neuropathological studies have shown that brain tissue around an AVM is frequently gliotic and exhibits cystic changes. Besides these subcortical changes which may extend very deep into the white matter there may occasionally be seen cases of severe encephalomalacia with atrophy of gyri or lobules in the neighbourhood of the AVM and also changes of the arachnoid-pialayer. In some cases this change is combined with hemorrhage, but in others there is no sign of bleeding. The superficial and deep changes are not clearly related to the site or size of the AVM. Neurosurgical observations do not confirm that gliotic change around the convolutions is, as often stated, a “pseudocapsule or matrix” in every case.

The gliotic area surrounding an AVM may represent the brain reaction to pulsation of the AVM, to ischemia, to microhemorrhage or to a primary developmental phenomenon (lack of an astrocytic layer around vessels, therefore diffusion of metabolites).

It is assumed that because of the dysplasia of a capillary bed, there is no functioning brain tissue within the AVM itself, at least in compact lesions. However, we do not know at what distance from the lesion normal cerebral architecture is preserved and this must vary in each case. A better understanding of the pathophysiology of the lesion thus awaits a more comprehensive embryological and histological analysis.

Localization

Besides the ‘pure’ AVM occupying a single intracranial compartment we have seen examples of AVMs involving multiple anatomical layers including skin, muscle, bone, dura, arachnoid, brain, and ventricle.

The following combinations have been described:

Localization of the AVM Within the Brain

AVMs do not always involve all layers of the brain from surface arachnoid down to the ventricle (Figs 3.7A–B, 3.8). From the surgical perspective they may be divided into the following groups and subgroups:

I. Surface Lesions (visible on exploration on the surface of the brain)

II. Deep Lesions (invisible at exploration on the surface)

Fig 3.7A–B A Typical relationship of cortical-subcortical AVMs to the ventricular system. 1 = frontal, 2 = temporal dorsal, 3 = temporal (amygdala + hippocampus), 4 = parietooccipital dorsal, 5 = parietal paramedian, 6 = callosal.

Fig 3.7B Infratentorial AVMs with extension to the IVth ventricle, with (1) and without (2) ventricular obstruction.

I 1–2–3 Surface Lesions

Surface lesions may involve only the cortex or the white matter and the subependymal layer to extend into the white matter or extend through reach the ventricular system (Figs 3.8, 3.9A–D).

Fig 3.8 Artistic drawing of the possible locations and extensions of hemispheric AVMs:

A with connection to ventricular system

B without connection to ventricular system.

Fig 3.9A–D Cortical-subcortical AVM (A–B) on the visible surface, with (D) and without (C) subependymal extension.

II 1 Within the Sulci

An AVM may appear to be superficial on angiography but at operation only a red draining vein is present, or there may be no evidence of the AVM at all as it is deep within the sulci. The lesion is still a cortical + subcortical one but hidden deep within a sulcus which may be two or three centimeters or more below the actual surface of the brain. Dissection through the arachnoid of the sulcus will expose the lesion (Fig 3.10A–D).

Fig 3.10A–D Cortical-subcortical AVM located in the depth of sulci (with [D] and without [C] subependymal extension) (B) and therefore not visible on the surface (A).

II 2 Within the Deep Fissures

Similarly, an AVM may lie deep within the interhemispheric (median surface of frontal, parietal, occipital lobe), Sylvian (insula and adjacent operculum) or transverse fissures respectively or in the sulci of these fissures. These AVMs are deeply located but actually are still superficial in relation to the corpus callosum, cingulate gyrus, insula, parahippocampus, and pulvinar thalami (see Fig 3.13B).

Exposure of these lesions can be gained with minimal retraction of the adjacent lobuli and by opening the subarachnoid cisterns.

II 3 Within the Deep White Matter

(see Fig 3.8B)

These often small lesions may be found in every part of the brain, but more frequently around the paraventricular area and within the internal capsule. An approach can be made through the fissures or sulci, or sometimes transcallosally. In some cases it is necessary to employ a transcortical incision or stereotactic or ultrasound techniques or to use stereotactic irradiation.

II 4 Within the Deep Gray Matter (see Fig 3.8)

These AVMs are localized within the amygdala, putamen, pallidum, nucleus caudatus, thalamus, hypothalamus, nucleus ruber and substantia nigra, nucleus dentatus and other nuclei. They are supplied primarily by perforating arteries of the ACA, MCA, PCA, PcoA, anterior and posterior choroidal arteries, AICA, PICA, SCA, vertebral and basilar arteries.

II 5 Cisternal (Subarachnoidal)

(see Fig 3.13B)

Although angiography of an AVM of the vein of Galen shows the lesion in the very center of the brain, surgical experience has shown that they lie entirely within the cisternal system. Surgical explorations have further demonstrated that there exist pure cisternal (subarachnoidal) AVMs which may be paramesencephalic (ventral or dorsal), parapontine (ventral or ventrolateral within the cerebellopontine angle) and parabulbar (around the medulla oblongata).

These paramedullary superficial AVMs of the brain stem seem to be an intracranial equivalent to some paramedullary spinal AVMs.

II 6 Intraventricular (see Fig 3.11)

A few totally intraventricular AVMs of the choroid plexus within the trigonum are known. These lesions may be approached through the corpus callosum. We have seen 2 cases with an AVM of the tela chorioidea of IIIrd ventricle, three cases of lateral ventricle, and two cases of an AVM of the plexus chorioideus of the IVth ventricle.

Fig 3.11A–C Surgically observed locations of para- and intraventricular AVMs with obstruction of ventricular system. A Anterior callosal and septal (1), posterior callosal (2), choroid plexus of Illrd ventricle (3), IVth ventricle (4).

Fig 3.11B, C B Paraventricular. C Varix, intraventricular.

As can be proven angiographically, AVMs in groups I 1a, 2a, 3a and II 1a, 2a, 3a (see page 64) are supplied by cortical arteries rather than perforators whereas in groups I 1b, 2b, II 1b, 2b, 3, 4, 5, 6 the reverse is true. Cisternal AVMs, particularly vein of Galen malformations, appear to be supplied about equally by cortical and perforating vessels (Figs 3.12–13).

Fig 3.12A–D The cerebral and cerebellar AVMs are similarly composed. Angiographically the flow to the AVM can be visualized from mainly 1 (A), 2 (B) or 3 (C) sources (ACA, MCA, PCA or PICA, AICA, SCA) or only from perforators (D). a = anterior cerebral artery, m = middle cerebral artery, p = posterior cerebral artery, d = deep perforators.

Fig 3.13A-D A Cerebral convexial AVM supplied mainly by cortical branches of ACA, MCA, PCA. Possible participation of dural branches. The perforating feeders very often participate in the supply of the AVM even though they may be invisible angiographically.

B AVM of vein of Galen is usually supplied by both cortical and perforating branches.

Fig 3.13C Deep, e.g. thalamic, parathalamic AVMs, are mainly supplied by perforating arteries.

Fig 3.13D Infratentorial convexial AVMs are composed similarly to cerebral convexial AVMs. These are mainly supplied by cortical (SCA, PICA, AICA) and perforating feeding arteries, from basilar artery and its branches.

The Nidus

The term “nidus” was introduced by Doppman (1971) when describing the structure of spinal AVMs as demonstrated by his innovative techniques of selective spinal angiography.

This terminology has been subsequently adopted by interventional neuroradiologists. They are now able to study the exact composition of extracranial AVMs with the help of superselective angiography. They are also able to describe the different compartments of an AVM, information which is extremely useful for the neurosurgeon in understanding the angioarchitecture of an AVM as well as in determining if and how it can be treated microsurgically. Despite this significant process, it still remains difficult to identify the actual nidus or the core of an AVM as one cannot always distinguish between the hemodynamic effects upon normal vascular channels (whether they are arteries or veins) and true embryonic remnants. Since the nidus represents the area of arteriovenous shunting within the AVM, it is probably best to consider it as that part of the malformation which is interposed between the recognizable feeding arteries and the larger terminal draining veins.

The nidus (epicenter) of the AVM is composed of a conglomerate of vascular loops, whose precise origin remains a source of controversy. Some feel these represent abnormal vascular channels, others that they are embryonic veins or normal veins arterialized by high blood flow and pressure. It has been noted at operation that the nidus of an AVM contains two types of connections. The first is a tangle of loops which appear to have some interconnections although the number and extent of these is generally not clear.

Histologic sections through AVMs do not show definite connections between loops. It is therefore possible that these represent branches of the arterial feeders which eventually reanastomose in order to form the draining vein or veins. A second type of connection is the direct (small to large) arteriovenous fistula which has long been recognized (Dandy 1928, Olivecrona 1957).