Do brainstem cavernous malformations have a higher rate of hemorrhage?

Abstract

The natural history of brainstem cavernous malformation is particularly complex. Currently, there is a wide range of reported annual rates of hemorrhage. Many studies report a significantly higher rate of hemorrhage and/or rehemorrhage in cavernous malformations located in the brainstem, but this is quite controversial. It is possible that brainstem cavernous malformations have a higher risk of hemorrhage and rehemorrhage, but reported variance in the literature may also be due to study limitations along with selection, patient and disease-specific; follow-up; and recall bias. An accurate assessment of hemorrhagic risk along with evaluation of patient and lesion-specific characteristics is critical in the decision-making process for potential intervention, as microsurgical intervention can significantly decrease the risk of future hemorrhage, but may be associated with significant complications.

Keywords:

• angioma
• brainstem
• cavernous
• cavernous malformation
• cerebral
• hemangiomas
• hemorrhage
• intracranial
• natural history
• neurosurgery
• surgery

In the July 2015 Expert Reviews of Neurotherapeutics, Gross and Du [1] provide an eloquent review of the literature regarding ‘The Natural History and Clinical Management of Cerebral Cavernous Malformations.’ Due to widespread use of MRIs, incidental cavernous malformations are being diagnosed with increasing frequency [2,3]. Neurological alterations are still common and are most attributable to either hemorrhage or seizures. Cavernous malformations located in the brainstem may lead to severe morbidity and mortality due to their intimate relationship with critical neural structures. The natural history of these lesions is particularly complex. Currently, there is a wide range of reported annual rates of hemorrhage across studies for brainstem cavernous malformations. Many studies report a significantly higher rate of hemorrhage and/or rehemorrhage in cavernous malformations located in the brainstem, but this is quite controversial. An accurate assessment of hemorrhagic risk along with evaluation of patient and lesion-specific characteristics is critical in the decision-making process for potential intervention as microsurgical intervention for brainstem cavernous malformations can significantly decrease the risk of future hemorrhage, but may be associated with significant complications.

Brainstem cavernous malformation are quite uncommon, comprising approximately 13% of vascular malformations of the posterior fossa [4–16]. Among cavernous malformations, location within the brainstem reported in the literature varies significantly from 4 to 35% [4,5,10,17]. Rates of hemorrhage are even more variable. In natural history studies, the reported annual hemorrhage rates range from 0.7 to 4.1% for all cerebral cavernous malformations [4–16,18,19]. Reported annual hemorrhage rates are lower for incidental cavernous malformations (0.08–0.2%) and unruptured cavernous malformations (0.3–0.6%) [4–16,18,19]. Rehemorrhage rates for all cerebral cavernous malformations are even higher and range from 3.8 to 22.9% [4–16,18,19]. Similarly, annual rates of hemorrhage among brainstem cavernous malformations range from 2.3 to 6.8% [5,16–18,20]. Rehemorrhage rates reported in the literature for brainstem cavernous malformations also vary from 5 to 21.5% [5,16–18,20]. At first glance, rates of hemorrhage and rehemorrhage between supratentorial and brainstem cavernous malformations appear to overlap. A number of studies have noted a higher risk of hemorrhage in brainstem lesions [4,8,13,14], but this has not been found in a number of well-conducted prospective studies [21,22]. It is possible that brainstem cavernous malformations have a higher risk of hemorrhage and rehemorrhage, but reported variance in the literature may also be due to study limitations along with selection, patient and disease specific; follow-up; and recall bias.

Deep-seated cavernous malformations, including brainstem lesions, may have an inherently higher risk of hemorrhage and rehemorrhage [13,23]. Many overall studies reporting overall rates of hemorrhage included brainstem, thalamic and basal ganglia cavernous malformations in their assessment of risk, which may inflate the overall range of hemorrhage for overall cerebral lesions. It is not always possible to partition out the specific hemorrhage rates based on specific locations in many studies.

Discrepancies over hemorrhage risks may also be due to inconsistencies in the definition of hemorrhage and rehemorrhage reported in the literature [24]. Studies reporting rehemorrhage rates based only on clinical alterations may be over-inflating hemorrhage rates that are actually due to associated thrombosis. Studies that define hemorrhage based only on symptomatic alterations along with MRI-confirmed hemorrhage understandably report smaller hemorrhage rates. Furthermore, recent use of high-field 7 Tesla MRIs and the advent of hemorrhage-specific MRI sequences are able to further define hemorrhages that may not have resulted in significant clinical alterations [25–28]. In addition, there is a known phenomenon of ‘cluster’ rehemorrhages once a lesion is ‘unstable’, which may be counted individually or as a single event depending on temporal proximity [4,10,13,18,19,29–31]. Whether a study incorporates microhemorrhages or intralesional hemorrhage further confounds the literature. Unfortunately, not all studies provide clear definitions of hemorrhage and rehemorrhage [24].

Rehemorrhage rates in supratentorial cavernous malformations may be particularly underestimated as they may not result in significant clinical alterations in either signs or symptoms. Cavernous malformations of the brainstem and other eloquent locations may specifically result in an increased reported risk of hemorrhage and rehemorrhage rates as even small microhemorrhages and intralesional hemorrhages can result in significant clinical alterations due to their proximity to integral and sensitive neural structures.

Hemorrhage rates may also be widely altered by patient-specific bias. Aside from location, patient- and lesion-specific characteristics have been shown to increase the risk of bleeding. Female gender, male gender, patient age, associated edema, size, deep location, and multiplicity of cavernous malformations have been found to increase the risk of bleeding and these factors are variable dispersed and reported among studies [4,5,10,13,18,19,21,22,29–31]. Furthermore, the risks in familial and radiation-induced cavernous malformations may differ from congenital and spontaneous lesions [10,17,32].

Selection bias can also have a significant effect on reported hemorrhage rates. In surgical series, the reported rates of hemorrhage and rehemorrhage are higher than in natural history studies. This may be due to the fact that these patients are pre-selected for surgery based on both known and unknown risk factors. For brainstem cavernous malformations, hemorrhage rates in natural history studies range from 2.3 to 4.1% compared with 2.7 to 6.8% for surgical series [5,16–18,20].

Selection and patient bias may not fully account for these differences. There are few prospective studies of brainstem cavernous malformations and prospective studies with planned clinical and radiographic follow-up are limited. These drawbacks may result in recollection bias along with bias due to loss or inadequate follow-up. Even well-planned prospective, collaborative studies may have been limited by patient enrolment and inadequate power due to how uncommonly brainstem cavernous malformations are encountered [21,22].

Furthermore, calculation of hemorrhage rates may be flawed or based on incorrect presumptions. Many studies calculate hemorrhage rates based on the notion that cavernous malformations are congenital lesions with a uniform hemorrhage risk throughout life. Many of these lesions may have formed spontaneously after birth or occur as secondary phenomena following radiation. In addition, many studies have shown that the risk of hemorrhage is not constant and likely follows variable patterns [4,8–16,21,22,33]. In some patients, the incidence of hemorrhage is often saltatory with periods of quiescence and clusters of hemorrhages [29–31]. For some patients, there is a period of instability followed by a decreased risk over time. In a collaborative, prospective study of 292 patients with 2035 patient years of follow-up, patients who presented with hemorrhage were at significantly higher risk of hemorrhage but this risk decreased over time [21]. In the prospective study of 139 patients with cavernous malformations with 1177 person-years of follow-up, Al-Shahi Salman et al. found that the annual risk intracranial hemorrhage or focal neurological deficits (not including epileptic seizure) declined from 19.8% (95% CI: 6.1–33.4) in year 1–5.0% (0·0–14.8) in year 5. Although there were small numbers of patients, the authors did not find that brainstem location significantly increased the risk of recurrent hemorrhage [22].

It remains unclear whether cavernous malformations located in the brainstem have an increased risk of hemorrhage or recurrent hemorrhage. Natural history studies may be limited by patient and disease-specific, selection, follow-up, and recall bias. These lesions are not necessarily uniform and the risk of hemorrhage may differ over time. Analysis with generalized estimating equations or other time-dependent regression analysis that allows for multivariate intra- and inter-patient analysis of risk while accounting for each individual hemorrhagic event may be beneficial. Future studies will depend on planned and meticulous neurological and radiographic follow-up and require concrete definitions of hemorrhage to allow comparison between studies and. Improved imaging with high-field MRIs, novel imaging sequences, and antibody-mediated imaging may better allow for prediction of patients at risk for hemorrhage. As these are rare lesions, future studies will benefit from multi-institution or international collaboration.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.