Cavernous Malformations of the Cervicomedullary Junction

10.1055/b-0034-84448

Cavernous Malformations of the Cervicomedullary Junction

Joseph M. Zabramski, John R. Robinson, Jr., and Robert F. Spetzler

Cavernous malformations are low-flow, hemorrhagic vascular lesions that affect 0.4 to 0.5% of the population. Because they are “angiographically occult” and poorly visualized by computed tomography (CT), they were once considered rare. With the introduction of magnetic resonance imaging (MRI), the importance of these lesions as a surgically treatable cause of seizures and neurological impairment has been well documented. For the most part, the excision of cavernous malformations is not difficult; however, the location of a specific lesion may raise a significant risk for perioperative complications. This risk is particularly true for lesions that involve the craniovertebral junction (CVJ).

This chapter reviews the epidemiology, presentation, and management of cavernous malformations with a particular emphasis on the surgical decision making for lesions located at the CVJ. As elsewhere in this text, the term CVJ is used to refer to the posterior fossa and upper cervical spine. Other chapters describe the technical details of the specific surgical approaches to this region.

Pathology

Grossly, cavernous malformations are well-defined dark red or purple, mulberry-like masses, surrounded by a characteristic brown or dark yellow-stained, gliotic border ( Fig. 17.1 ). They may range from a few millimeters to several centimeters in size. In one study in which dimensions were measured by MRI, the average lesion was 1.7 cm in diameter, with a range of 3 mm to 4 cm.1 There was no significant difference in the size of lesions in the supra- and infratentorial compartments, but posterior fossa lesions were more commonly symptomatic, particularly when located in the lower brain stem.

Microscopically, cavernous malformations are characterized by a complex of markedly dilated vascular channels (caverns) arranged in a back-to-back pattern with little or no intervening brain parenchyma ( Fig. 17.2 ).2–7 A loose collagenous matrix may separate the channels at the periphery. The dilated channels frequently contain thrombus of various ages and degrees of organization. Histologically, the vascular channels are thin-walled and lined by a single layer of vascular endothelium. The walls contain no elastin or smooth muscle and characteristically have no basement membrane ( Fig. 17.2 ).8 The lack of a supporting matrix for the vascular structures appears to make the lesions susceptible to repeated episodes of focal intralesional hemorrhage and thrombosis. Focal areas of calcification are also relatively common and may be visualized on CT.

Gross pathologic specimen demonstrating the typical appearance of a cavernous malformation. Hemosiderin from recurrent episodes of intralesional hemorrhage and thrombosis accumulates by diapedesis in macrophages and glia around these lesions, producing a characteristic brown or dark yellow-stained border that surrounds the mass (*arrowheads*). (From Zabramski JM, Spetzler RF. Cavernous malformations. In: Aminoff M, Daroff R, eds. Encyclopedia of the Neurological Sciences. San Diego, CA: Academic Press; 2003:532–538. Reprinted with permission from Elsevier.)

Masson′s trichrome stain of a characteristic cavernous malformation. Note the back-to-back vessel arrangement of markedly dilated capillary vessels (caverns) that compose the lesion. The vessel walls are lined by a single layer of vascular endothelium with an abundance of collagen (*blue stain*) and an absence of elastin (*black stain*) or a basement membrane.

Epidemiology

Cavernous malformations account for 5 to 10% of all central nervous system (CNS) vascular malformations.7,9 In surgical series they are the second most common vascular malformation, outnumbered only by arteriovenous malformations (AVMs). Cavernous malformations have been reported from almost all corners of the world, although the incidence may be slightly lower in African and Asian cultures.

The overall incidence of cavernous malformations in the general population is estimated to be between 0.4 and 0.5%. In two large reviews of nonselected autopsies cases, Sawar and McCormick9 reported an incidence of 0.4% in 4069 patients, whereas Otten and colleagues10 reported 0.53% in a series of 24,535 cases. These results closely agree with those from two large MRI reviews: Del Curling and colleagues11 cited an incidence of 0.39% in a series of 8131 patients undergoing MRI, whereas Robinson and colleagues1 found a 0.47% incidence in their review of 14,035 sequential MRIs. These findings were recently confirmed by Vernooij and colleagues12 in a study describing the incidental findings on brain MRI in the general population: The authors reported the results of MRI studies in 2000 adult patients over the age of 45 (mean 63.3 years) and found that 0.4% of this population had incidental cavernous malformations. Although various authors have reported a slight male or female preponderance for cavernous malformations, overall both sexes appear to be affected equally.13

Cavernous malformations are distributed throughout the CNS in rough relationship to the volume of the various compartments; 70 to 80% occur supratentorially, 10 to 20% are in the posterior fossa, and 5 to 10% are in the spine.14–21 In the posterior fossa, the lesion preferentially involves the pons and cerebellum. Table 17.1 lists the distribution of cavernous malformations at the CVJ in 142 patients from the three largest available series.14,22,23

Cavernous malformations occur in two forms: spontaneous and familial. The spontaneous form occurs as isolated cases and is characterized by the presence of a single lesion, whereas the familial form is characterized by multiple lesions and an autosomal dominant mode of inheritance. Multiple lesions and a family history of seizures are nearly pathognomonic for the familial form of cavernous malformations. With careful screening we find that more than 80% of patients with three or more lesions have a history consistent with the familial form of this disease.22 Indeed, many individuals with multiple lesions but lacking a clear family history have been shown to harbor a de novo germline mutation.24–27

The introduction of MRI, with the resulting ability to readily detect cavernous malformations, combined with an explosion of genetic technology provided the tools necessary to identify the genetic mutations that cause these lesions. The onset of cavernous malformations occurs either sporadically or in an inherited autosomal dominant form due to mutations in one of three genes, CCM1/KRIT1,28,29 CCM2/malcavernin,24,30 or CCM3/PDCD10.25 Based on available data, at least two of these loci (CCM1 and CCM2) appear to be involved in the regulation of β₁ integrin signal transduction, which plays a crucial role in the regulation of cell adhesion and migration during angiogenesis. The third locus (CCM3) involves an apoptotic pathway. The role of the latter gene in angiogenesis and the formation of cavernous malformations remains to be investigated.

**Distribution of 142 Cavernous Malformations Involving the Craniovertebral Junction**
Location	No. of Lesions (%)
Pons	56 (39)
Pontomedullary	10 (7)
Medulla	23 (16)
Cerebellum	49 (35)
Cervicomedullary	4 (3)

Diagnosis

Prior to the introduction of modern imaging techniques, cavernous malformations were considered rare curiosities. In 1976 Voigt and Yasagil31 described their experience with one case and reviewed the world literature, finding only 126 reported cases. Because cavernous malformations are low-flow, capillary lesions, angiography is of little use in their diagnosis.32 Occasionally, larger lesions will produce some degree of mass effect or late-phase venous pooling. For the most part, however, these lesions are not visualized by angiography and in the pre-CT/MRI era were commonly referred to as “angiographically occult AVMs,” or “thrombosed AVMs.”

Although CT scanning was a significant advance in neuroimaging, it lacked the sensitivity and specificity necessary to be diagnostic for cavernous malformations. The averaged, mixed-tissue density of many cavernous malformations is almost identical to that of the normal brain on CT. In addition, mass effect, contrast enhancement, and edema, which are the key diagnostic features on CT for other pathologies, are not characteristic of cavernous malformations. In general, CT identifies only ~50% of the lesions demonstrated by MRI studies.33 Nevertheless, CT is a useful screening test in patients presenting with the new onset or exacerbation of focal neurological deficits, as it is exquisitely sensitive to acute blood and may identify new episodes of hemorrhage.

MRI is the study of choice for the diagnosis of cavernous malformations. It is both highly sensitive and selective. The characteristic MRI appearance is considered nearly pathognomonic. To provide some idea of the impact of MRI on the recognition of this disease, we reviewed the MEDLINE data base and found only five articles published on cavernous malformations during the first 10 years of the CT era (1976 to 1985), compared with 110 articles during the subsequent 10 years when MRI became generally available (1986 to 1995).

The classic magnetic resonance imaging appearance of cavernous malformations is demonstrated on these heavily T2-weighted **(A)** spin echo and **(B)** gradient echo images in this patient with the familial form of the disease. The lesions typically contain a variegated core of mixed signal intensity produced by the presence of hemorrhage and thrombus of widely differing ages and degrees of organization. The core of the lesion is surrounded by a ring of low signal intensity that is characteristic of chronic hemorrhage, produced by the deposition of iron and hemosiderin in the surrounding tissue (Fig 17.1). **(B)** This metallic artifact is significantly greater on gradient echo images.

The classic appearance of cavernous malformations on MRI—that of a mass with a core of mixed signal intensity on T1- and T2-weighted spin echo images surrounded by a ring of low signal intensity—is most apparent on heavily T2-weighted spin echo and gradient echo images ( Fig. 17.3 ). This halo of signal loss is diagnostic for remote hemorrhage and is a result of magnetic field distortions produced by the deposition of hemosiderin in the tissue adjacent to the cavernous malformation. Signal loss and distortion are minimized on heavily T1-weighted images, which are recommended for surgical planning.

Despite their size, the lesions produce little mass effect on the surrounding brain ( Fig. 17.3A ). Extensive edema is rare, and there is little or no contrast enhancement after the administration of gadolinium. Lesions containing a large component of hyperacute hemorrhage may appear dark on both T1- and T2-weighted MRI but are readily identified by CT. Subacute hemorrhage may produce areas of high or low signal intensity depending on its age and stage of resorption ( Table 17.2 ).

Other lesions with a propensity for hemorrhage may occasionally masquerade as cavernous malformations. The presence of extensive edema in the surrounding tissue should bring the diagnosis of cavernous malformation into question. Repeat imaging after a short period of observation (2 to 3 weeks) will resolve the issue in difficult cases. Mixed forms of vascular malformations, containing cavernous malformations in combination with an AVM or other vascular malformations, may be suggested by CT or MRI findings.2 The addition of angiography is helpful in such cases.

**Changes in Imaging Characteristics of Hemorrhage with Time**
Diagnostic Study	Acute (1–2 days)	Subacute Early (1–2 weeks)	Subacute Late (1–2 months)	Chronic (>6 months)
CT	+ +	±	−	−
T1-weighted MRI	−	+ +	+ +	−
T2-weighted MRI	−	+ +	±	− (+ halo)
Abbreviations: CT, computed tomography; MRI, magnetic resonance imaging; +, hyperintense to surrounding brain; −, hypointense to surrounding brain; ±, mixed intensity; + halo, a ring of low signal intensity surrounding the core of the lesion.

When other lesions are found in association with cavernous malformations, they are most commonly venous malformations. Venous malformations, also called developmental venous anomalies (DVAs), are benign vascular lesions found in ~2% of the population. They are identified in association with cavernous malformations in 20 to 30% of cases.2,18,34 On angiography, venous malformations are readily visualized as radial collections of multiple small veins draining into a large central trunk to produce a characteristic caput medusa pattern ( Fig. 17.4 ). On CT and MRI, the large central draining vein can frequently be recognized as a linear-enhancing flow void ( Fig. 17.4 ). It is critical to understand that these lesions provide the normal venous drainage of the surrounding brain. Larger branches must be carefully preserved. Their sacrifice, particularly of the central draining vein, may lead to venous infarction with associated swelling, hemorrhage, and even death.

Presentation

Cavernous malformations appear to grow and to produce symptoms as a result of recurrent episodes of intralesional hemorrhage and thrombosis. More rarely, they present with extralesional hemorrhage. These episodes are asymptomatic until the lesion reaches a critical size that irritates or compresses adjacent structures.

Headache is a relatively common but nonspecific symptom in patients with cavernous malformations. It is present in approximately one-third of patients regardless of whether lesions are supra- or infratentorial, or whether there is evidence of acute hemorrhage.1,11,18,22 The mechanism by which these lesions produce headache is unknown. Significant mass effect and the obstruction of cerebrospinal fluid (CSF) pathways are rare, and hemorrhage rarely reaches pial surfaces. Whatever the etiology, headache is the complaint that initiates the diagnostic evaluation in a significant number of patients.

Although seizures are the most common presenting symptom when lesions are located supratentorially, focal neurological deficits predominate when lesions involve the CVJ. Combinations of long tract signs and cranial nerve deficits vary, depending on the exact location of the lesion and its size. The most common presentation ( Table 17.3 ) is acute onset of oculomotor motor dysfunction (47%), followed by ataxia/dizziness (37%), sensory disturbances (30%), headache (30%), and motor weakness or spasticity (28%).18

This 52-year-old woman had a long history of headache. An exacerbation of her normal headache pattern led to this magnetic resonance imaging (MRI) study. The MRI findings are consistent with subacute hemorrhage from a cavernous malformation in the left cerebellum. The conversion of hemoglobin to methemoglobin in subacute hemorrhage produces a bright signal on both **(A,B)** the heavily T2-weighted spin echo images and **(C,D)** T1-weighted images. (C, *curved arrow*) The source of the hemorrhage, a small cavernous malformation, is best seen on the nonenhanced T1-weighted coronal image. **(B)** Note the linear flow void immediately adjacent to the lesion (*arrow*) on the T2-weighted axial image; **(E)** this finding is characteristic for a venous malformation and is confirmed by subsequent catheter angiography. **(D)** The venous malformation is also apparent on the gadolinium-enhanced coronal, T1-weighted image draped over the superior aspect of the hemorrhage (*straight arrows*). It is important to recognize that the venous malformation is not the source of the hemorrhage in this case. Resection or obliteration of the venous malformation, which drains most of the left cerebellar hemisphere, would result in massive venous infarction. The surgical approach chosen for this lesion must allow resection of the cavernous malformation while avoiding injury to the associated venous malformation.

The development of symptoms in patients with hemorrhage from cavernous malformations of the CVJ is characteristically acute and maximal at onset. Neurological deficits from the first symptomatic episode of hemorrhage tend to resolve completely as the hemorrhage is organized and absorbed, whereas recurrent hemorrhages are likely to be associated with progressively more severe deficits and an increased risk of permanent neurological impairment. Before the advent of MRI, this stuttering clinical course was frequently mistaken for multiple sclerosis, particularly when lesions involved the lower brainstem.

**Symptom Profile of 62 Cavernous Malformations Involving the Cervicomedullary Junction**
Symptom	Number
Ocular motility	47
Ataxia	37
Sensory deficits	31
Headache	31
Motor deficits	28
Facial pain and hyperesthesia	26

Natural History

The natural history of cavernous malformations has become a topic of increasing interest as physicians struggle with management decisions, particularly in patients with minimally symptomatic or deeply seated lesions of the CVJ. The decision of whether to proceed with operative intervention should be based on a clear understanding of the risks of the surgical procedure compared with those of the natural history of the disease.

Numerous studies have been published on the natural history of cavernous malformations. Hemorrhage rates vary widely from series to series depending on the authors’ definition of hemorrhage and the population being studied. Not surprisingly, hemorrhage rates tend to be higher in surgical series, as patients with symptomatic lesions are more likely to be referred for neurosurgical intervention. The literature is further complicated by the use of four different methods of calculating hemorrhage rates including retrospective and prospective methods—either of which can be reported as risk of hemorrhage per patient or as risk of hemorrhage per lesion ( Table 17.4 ).

**Methods of Calculating Hemorrhage Rates**
Method	Retrospective (assume present from birth)	Prospective (from time from first identification)
Rate per lesion	Retrospective per lesion	Prospective per lesion
Rate per patient	Retrospective per patient	Prospective per patient

The retrospective method assumes that all lesions have been present from birth. Using this assumption, Del Curling and colleagues11 calculated a symptomatic hemorrhage rate of 0.25% per patient per year. Kondziolka and colleagues35 reported a rate of 1.3% per patient per year, and Kim and colleagues36 calculated a rate of 2.3% per patient per year (1.4% per lesion per year). This method of calculation, which depends on the patient′s recall to define episodes of hemorrhage and assumes that all lesions are present from birth, is likely to underestimate the actual risk of significant bleeding. Once considered congenital in origin, there is increasing evidence that new lesions may appear de novo in both the sporadic and familial forms of the disease.22,37,38

Another confounding factor is the highly variable nature of these lesions. Zabramski and colleagues22 classified cavernous malformations into four subtypes based on MRI characteristics ( Table 17.5 ). These authors and others have found that the risk of hemorrhage appears greatest for type I and type II lesions, which are more likely to be symptomatic.1,39–43 Most clinical and surgical series are heavily biased toward these two subtypes of lesions, which are readily identified on MRI studies and are frequently symptomatic. These uncertainties emphasize the need to rely on prospective data for the natural history in these patients.

Robinson and colleagues1 prospectively followed a group of 57 patients with serial clinical examinations and MRI studies for a mean of 26 months and found a risk of symptomatic hemorrhage of 0.7% per lesion per year. Kondziolka and colleagues35 reported a slightly higher hemorrhage rate of 2.6% per year but noted that the risk of hemorrhage was strongly related to clinical presentation. They prospectively followed 122 patients for a mean of 34 months and found that the hemorrhage rate was significantly lower in patients who presented with incidental lesions: 0.6% per year (n = 61) compared with 4.5% per year in those with a history of previous symptomatic hemorrhage (n = 61). Aiba and colleagues39 followed 110 patients with cavernous malformations for a mean of 4.5 years and reported a 0% hemorrhage rate for patients with incidental lesions.

In general, the natural history of cavernous malformations involving the brainstem and the upper cervical cord at the CVJ parallels that of lesions elsewhere in the CNS. However, because of eloquence of the surrounding structures, episodes of hemorrhage (even from relatively small hemorrhages) are much more likely to be symptomatic. A conservative estimate, based on the assumption that lesions are present from birth to first symptom, places the risk of symptomatic hemorrhage in the range of 2.5 to 6.8% per year (mean, 4.5% per lesion-year).35,44–48 This estimate is similar to the 2% per year risk reported by Mathiesen and colleagues49 in a group of 11 patients with asymptomatic brainstem cavernous malformations who were followed prospectively for a mean of 4 years. For patients who present with a history of previous symptomatic hemorrhage, the risk of rebleeding ranges from 5.1 to 60% (mean, 28.7% per lesion-year),35,44–47,50 with higher rates reported for those presenting with evidence of recent hemorrhage on MRI.42,47

**Magnetic Resonance Imaging Classification of Cavernous Malformations**
Lesion Type	MRI Signal Characteristic	Pathological Characteristics
Type I_A	T1: hyperintense focus of hemorrhage T2: hyper- or hypointense focus of hemorrhage extending through at least one wall of the hypointense rim that surrounds the lesion (Figs. 17.3 and 17.4) Focal edema* may be present (Fig. 17.8)	“Overt” extralesional focus of hemorrhage extending outside the lesion capsule
Type I_B	T1: hyperintense focus of hemorrhage T2: hyper- or hypointense focus of hemorrhage surrounded by a hypointense rim (Fig. 17.5)	Subacute focus of intralesional hemorrhage
Type II	T1: reticulated mixed signal core T2: reticulated mixed signal core surrounded by a hypointense rim (Fig. 17.1 and Fig. 17.2, straight arrows)	Loculated areas of hemorrhage and thrombosis of various ages surrounded by gliotic, hemosiderin-stained brain; in large lesions, areas of calcification may be seen
Type III	T1: iso- or hypointense T2: hypointense, with hypointense rim that magnifies the size of lesion GE: hypointense with greater magnification than T2 (Fig. 17.9)	Chronic resolved hemorrhage with hemosiderin staining within and around the lesion
Type IV	T1: poorly seen or not visualized at all T2: poorly seen or not visualized at all GE: punctate hypointense lesions (Fig. 17.9)	Two lesions in the category have been pathologically documented to be telangiectasias
* Focal edema may surround the extralesional portion of hemorrhage in type I_A lesions.
Abbreviations: T1 and T2, T1- and T2-weighted magnetric resonance images, respectively; GE, gradient echo sequences; MRI, magnetic resonance imaging.
Source: Adapted from Feiz-Efran I, Zabramski JM, Kim LJ, Klopfenstein, JD. Natural history of cavernous malformations of the central nervous system. In: Lanzino G, Spetzler RF, eds. Cavernous Malformations of the Brain and Spinal Cord. New York, NY: Thieme Medical Publishers; 2008:3–10.

Magnetic resonance imaging (MRI) in a 34-year-old woman with a 1-week history of sudden onset of left-sided weakness and sensory loss demonstrates findings consistent with a large subacute hemorrhage from a pontine cavernous malformation. Note that the bulk of the subacute hemorrhage—areas of high signal intensity on **(A)** T1-weighted and **(B)** T2-weighted images—lies outside the normal confines of the lesion capsule (ring of low signal intensity) on **(B)** the T2-weighted images. This pattern and the presence of edema on **(B)** the T2-weighted images are consistent with “gross,” or extralesional hemorrhage, and are associated with a high risk of recurrent symptomatic bleeding. The area of recent subacute hemorrhage reaches the pial surface on the ventral pons and provides a safe path for surgical excision of the lesion. The hematoma and associated cavernous malformation were removed via an orbitozygomatic approach without complications, and the patient was discharged home in good condition 1 week after surgery. The patient has been symptom-free with no evidence of recurrent hemorrhage or residual lesion during 2 years of follow-up. **(C)** The difficulty in interpreting early postoperative MRI studies is demonstrated by the T2-weighted, spin echo images obtained on the second day after surgery in this patient; edema and blood in the surgical bed produce an image readily confused with residual cavernous malformations. **(D)** The repeat MRI 6 months after surgery reveals only the characteristic low signal signature of hemosiderin associated with chronic resolved hemorrhage on T2-weighted, spin echo images. The absence of any high signal intensity areas within this region on this delayed image confirms the complete resection of the cavernous malformation.

Hemorrhage rates appear to be particularly high in patients who present with acute/subacute bleeding episodes that violate the lesion capsule, producing a so-called “overt,” extralesional hemorrhage (type I_A lesion, Table 17.5 ; Fig. 17.5A,B ) into the surrounding brain. Aiba and colleagues39 followed 62 such patients for a mean of 3.12 years and noted a risk of recurrent symptomatic hemorrhage of 22.3% per lesion-year. Barker and colleagues42 reported a similar experience with 141 patients selected for intervention who presented with “overt” hemorrhages. In this series, 63 patients experienced a second hemorrhage before treatment. Hemorrhages clustered around the initial event, with a rehemorrhage rate of 25.2% per year for the first 28 months. Comparable rates of rebleeding have been reported following incomplete resection of cavernous malformations, presumably due to interruption of the lesion capsule, stressing the importance of complete resection during the initial surgical procedure.42,44,50,51