CSF-venous fistulas associated with traumatic spinal pseudomeningoceles

Abstract

CSF-venous fistulas (CVFs) are a common cause of spontaneous intracranial hypotension. These fistulas usually occur without any preceding major trauma, surgery, or other iatrogenic cause. Occasionally, patients have a history of minor trauma, though such cases are usually still considered spontaneous. Little is known about predisposing factors that cause patients to develop spontaneous CVFs. Most patients with CVFs have multiple meningeal diverticula on spine imaging, and fistulas usually arise in association with a diverticulum. In the vast majority of cases, the culprit diverticulum from which the CVF arises is atraumatic in origin, presumably on the spectrum of normal variation in spinal anatomy. Here, we present two cases of CVFs that arose in association with posttraumatic pseudomeningoceles. To our knowledge, this phenomenon has not yet been reported, and it potentially represents a novel etiology for CVFs that furthers understanding of their pathogenesis.

Introduction

CSF-venous fistulas (CVFs) are a common and increasingly recognized type of spinal CSF leak. ¹ CVFs result in unregulated egress of CSF from the spinal subarachnoid space, leading to typical symptoms of intracranial hypovolemia/hypotension such as orthostatic headaches, tinnitus, brain fog, and occasionally more severe presentations.^2,3 Variably accompanying the clinical syndrome are particular brain imaging features of CSF hypotension, such as pachymeningeal enhancement, brain sagging, and venous sinus distention. ⁴

The overwhelming majority of CVFs are spontaneous, thus being categorized as etiologies for spontaneous intracranial hypotension (SIH). ⁵ Occasionally, patients offer a history of minor trauma prior to the onset of symptoms. In most cases, this minor trauma is not thought to be causative of the fistula. ⁶ CVFs typically occur in patients with multiple meningeal diverticula on spine imaging. ⁷ Such diverticula are present in many asymptomatic patients, and they usually represent normal variations in spinal anatomy. Nonetheless, CVFs most commonly arise in association with one or more of these diverticula, though not necessarily from the largest or most irregular diverticulum.

Although spine MRI and conventional CT myelography can demonstrate the presence of meningeal diverticula, these modalities almost never reveal CVFs. Rather, advanced myelographic techniques such as decubitus CT myelography or decubitus digital subtraction myelography (DSM) are usually needed to localize CVFs. ⁸ These exams are not universally available and can be resource intensive. It is important, therefore, to be aware of the full spectrum of clinical presentations and conventional spinal imaging findings that way warrant further investigation for a CVF.

Here, we present two cases of CVFs arising in association with posttraumatic pseudomeningoceles. To our knowledge, these are the first two reported cases of this phenomenon.

Case reports

Patient 1

A 43-year-old woman presented with 3 years of progressive orthostatic headaches that were exacerbated by coughing and other Valsalva maneuvers. Her history was remarkable for a motor vehicle accident 20 years earlier, which resulted in right-sided cervical nerve root avulsions and partial loss of right upper extremity motor and sensory function. She had a normal range body mass index and no history of connective tissue disorders. Based on her newer orthostatic headaches, she underwent contrast-enhanced brain MRI. This revealed multiple findings of intracranial hypotension, including brain sagging, distention of the dural venous sinuses, and pituitary engorgement (Figure 1). Minimal superficial siderosis was present over the cerebellar vermis (Figure 1). Spine MRI showed multiple right-sided cervical pseudomeningoceles, the largest of which was at C7-T1, involving the C8 nerve root (Figure 2). Right lateral decubitus CT myelography demonstrated a right C8 CVF (Figure 2). The patient underwent transvenous Onyx embolization of the CVF. At 3-month follow-up, her orthostatic headaches had resolved. The patient experienced symptoms of rebound intracranial hypertension, which were treated with acetazolamide. Brain MRI continued to show some evidence of brain sagging, though improved from the pretreatment scan.

Figure 1.

Sagittal T1W postcontrast brain MRI (A) from Patient 1 demonstrates multiple findings of intracranial hypotension, including pituitary engorgement, brain sagging, venous sinus distention, and inferior cerebellar tonsillar ectopia (A, arrows). Axial susceptibility weighted image shows mild superficial siderosis over the cerebellar vermis (B, arrows).

Figure 2.

Axial T2W MRI (A) from Patient 1 shows a right C8 pseudomeningocele (A, dashed arrow). Axial (B) and coronal (C) images from a right lateral decubitus CT myelogram demonstrate opacification of this pseudomeningocele (B and C, dashed arrows) with an adjacent CSF-venous fistula (B and C, solid arrows).

Patient 2

A 43-year-old man presented with 15 years of episodic headaches and tinnitus. Five years prior to the onset of his headaches he had a snowmobiling accident resulting in upper thoracic nerve root avulsion injuries and complete right T1 paraplegia. He had a normal range body mass index and no history of connective tissue disorders. At the time of presentation to our institution, brain MRI revealed infratentorial superficial siderosis, marked brain sagging, pachymeningeal enhancement, and venous sinus distention (Figure 3). Spine MRI showed a right T1 pseudomeningocele, though no evidence of extradural CSF. The patient did not have any prominent meningeal diverticula elsewhere. A right lateral decubitus DSM was performed, demonstrating a right T1 CVF associated with the pseudomeningocele (Figure 4). The patient underwent transvenous Onyx embolization of the CVF but presented twice more over the next year with persistent symptoms. On both occasions, a recurrent right T1 CVF was evident on repeat DSM (Figure 4). After two additional rounds of Onyx embolization, the patient was ultimately headache free. Repeat brain MRI demonstrated marked improvement in findings of intracranial hypotension (Figure 3). In this case, we suspected that it was challenging to fully treat the nidus of the fistula due to the anatomically difficult location at T1.

Figure 3.

Axial susceptibility weighed image (B) from Patient 2 shows marked infratentorial superficial siderosis over the cerebellar folia (A, arrows). Sagittal T1W postcontrast image (B) shows extensive findings of intracranial hypotension, including pituitary engorgement, brain sagging, inferior cerebellar tonsillar ectopia, and venous sinus distention (B, arrows). Posttreatment sagittal T1W postcontrast image (C) after repeated Onyx embolization of a right T1 CSF-venous fistula demonstrates resolution of these intracranial abnormalities (C, arrows). Axial T2W image from the patient's spine MRI (D) shows the appearance of the pseudomeningocele at T1 on the right (D, white arrow), with adjacent traumatic changes to the spinal cord (D, black arrow).

Figure 4.

Right lateral decubitus digital subtraction myelogram (A) in Patient 2 demonstrates a right T1 CSF-venous fistula (A, arrow) arising from a pseudomeningocele (A, dashed arrow). The patient had persistent symptoms 6 months after initial Onyx embolization of the fistula, and repeat digital subtraction myelogram showed a persistent fistula (B, arrow) arising from the same pseudomeningocele (B, dashed arrow). Milder symptoms persisted after a second Onyx embolization procedure. Precontrast (C) and postcontrast (D) unsubtracted images from a third DSM 6 months later show subtle persistent venous opacification (D, arrow) associated with the pseudomeningocele (D, dashed arrow). After a third round of Onyx embolization, the patient was headache free, obviating the need for surgical management.

Discussion

We have described two cases of CVFs associated with posttraumatic pseudomeningoceles. Both patients presented with typical symptoms of intracranial hypovolemia/hypotension, as well as brain MRI abnormalities consistent with that diagnosis. As is the case with most CVFs, decubitus CTM or decubitus DSM were necessary to localize the fistula. ⁹ Both patients were successfully treated with transvenous Onyx embolization of the CVF, with the first patient only requiring one treatment and the second requiring three. Both patients had symptomatic resolution, and the second patient also had complete resolution of his brain MRI abnormalities.

The overwhelming majority of CVFs are spontaneous. ² One recent report described a posttraumatic CVF involving the clivus after major skull base trauma. ¹⁰ To our knowledge, no prior reports have described CVFs in association with posttraumatic pseudomeningoceles. Interestingly, both of our patients developed symptoms of intracranial hypotension years after their traumatic injuries. We speculate that the patients’ injuries led to formation of a pseudomeningocele, with the CVFs developing spontaneously from these pseudomeningoceles at a later time.

The precise pathogenesis of CVFs, even in typical cases of SIH, remains uncertain. Current theories suggest that spinal arachnoid granulations, which are thought to represent normal CSF resorption pathways during fetal development, persist into adulthood in some patients and permit the formation of CVFs. ¹¹ Since the vast majority of CVFs occur at spinal levels harboring meningeal diverticula, it is likely that these outpouchings of the subarachnoid space are mechanistically important in CVF development. ¹² It may be that persistent spinal arachnoid granulations are necessary, though not always sufficient for CVF formation, and an additional lesion such as a meningeal diverticulum is needed for CVFs to occur. Specifically, we would postulate that diverticula bring the subarachnoid space in closer proximity to paraspinal veins, facilitating CVF formation. Traumatic pseudomeningoceles could act in a similar fashion. CVFs are also more common in patients with elevated BMI, suggesting that some patients initially have intracranial hypertension before developing a CVF. ¹³ The two presented cases are interesting because they demonstrate that traumatic pseudomeningoceles, too, can potentially elicit CVFs. Intracranial hypertension and traumatic pseudomeningoceles may represent two separate risk factors for CVFs.

Another point of interest is the presence of superficial siderosis in both of our patients, the cause of which is unclear. Superficial siderosis is most classically associated with ventral dural tears, where chronic bleeding into the subarachnoid space is thought to occur at the margins of the dural tear. ¹⁴ One study showed superficial siderosis in a small percentage of patients with CVFs, thought to be due to bidirectional blood flow in the fistulas. ¹⁵ While this theoretically could be the cause of siderosis in our two patients, there is no definitive evidence in the literature that blood can flow from a paraspinal vein into the subarachnoid space through CVFs. Since our two cases were complicated by posttraumatic pseudomeningoceles, it is also possible that chronic subarachnoid bleeding from the sites of nerve root avulsion/pseudomeningocele caused siderosis.

Overall, these cases demonstrate three main lessons. First, CVFs can occur in association with post-traumatic pseudomeningoceles, representing a novel mechanism for CVF formation separate from that of typical cases. These CVFs may still be “spontaneous,” with the pseudomeningocele serving as a predisposing risk factor. Second, given that CVFs can occur in this unique setting, it is critical to consider decubitus myelography in patients with intracranial hypotension who have posttraumatic pseudomeningoceles. If symptoms or brain MRI findings are falsely attributed to the mere presence of pseudomeningoceles, a treatable CVF may be missed. Finally, these patients should be scrutinized for superficial siderosis on brain MR imaging. If present, it is important to follow them with repeat brain MRI to ensure that siderosis does not progress, in which case surgical repair of the pseudomeningoceles may also be warranted. Alternatively, CSF analysis to assess for active bleeding into the subarachnoid space could be considered in these patients.

It has been 10 years since the initial reports of CVFs, and our understanding of their unique pathophysiologies continues to evolve. ¹