Efficacy of transvenous embolization of CSF-venous fistula in spontaneous intracranial hypotension: Case-series

Abstract

Background

Spontaneous intracranial hypotension is a debilitating neurological condition which can be caused by a cerebrospinal fluid-venous fistula. Transvenous embolization is a promising technique to provide minimally invasive yet durable treatment.

Methods

A retrospective single-center case series was performed on all patients who underwent transvenous embolization of a cerebrospinal fluid (CSF)-venous fistula. Clinical and radiographic parameters, including Bern score, were reported preoperatively and at 3-month follow-up.

Results

Six patients underwent embolization of a CSF-venous fistula. All fistulae were located in the thoracic spine and technical success was achieved in all cases. Three patients had symptom resolution, two had significant improvement, and one had stable symptoms on follow-up. The mean Bern score was 6.83 (SD = 1.47) preoperatively and 1.83 (SD = 1.64) postoperatively with a mean improvement in Bern score of 5.0 (SD = 1.9, p = 0.0013).

Conclusions

CSF-venous fistulas are an increasingly recognized clinical entity which may be treated with transvenous embolization. This case series serves to further validate this technique and suggests it can be performed with similar outcomes in lower volume centers.

Introduction

Spontaneous intracranial hypotension (SIH) is a clinical neurological condition characterized most commonly by refractory orthostatic headaches. Diagnosis is supported by magnetic resonance imaging (MRI) features of intracranial hypotension, including diffuse pachymeningeal enhancement, subdural fluid collections, venous sinus distension, and decreased suprasellar, mamillopontine, and prepontine distances. ¹ Cerebrospinal fluid (CSF)-venous fistula is an increasingly recognized cause of SIH, and localization can be identified via digital subtraction myelography or computed tomography (CT) myelography. Treatment options have previously been limited to blood patch, which is limited in its durability, or open surgery, typically requiring nerve root ligation. ²

More recently, in 2021 a novel endovascular technique for embolization of a CSF-venous fistula using liquid embolics has been described. ³ However, only a few series exist and are primarily limited to a handful of institutions with a high volume of referrals.^4–6 Consequently, it remains understudied whether CSF-venous fistula embolization can be performed safely and effectively in “real world” lower-volume centers thereby further validating high-volume center published results. Likewise, endovascular embolization of CSF-venous fistulas remains a new technique, and additional studies are warranted to generally validate these outcomes.

Materials and methods

Patient selection

Institutional review board approval was obtained for this study. A retrospective review of a prospectively maintained database was performed to identify patients who underwent transvenous embolization of previously diagnosed CSF-venous fistulas between November 2021 and July 2023. All patients were previously diagnosed with SIH based on both clinical symptomatology and MR imaging findings. Following a diagnosis of SIH, each patient underwent digital subtraction myelography which confirmed and localized a CSF-venous fistula.

Digital subtraction myelography

Each patient underwent digital subtraction myelography to diagnose and localize the CSF-venous fistula. Our digital subtraction myelography (DSM) technique is similar to the previously described techniques by Brinjikji et al. and others.^4,7–9 Each DSM procedure was performed in a biplane angiography suite. Patients underwent either monitored anesthesia care or general anesthesia and were positioned in the lateral decubitus position. A lumbar puncture was performed near the level of L3–4, with a 20-gauge spinal needle. Opening pressure was measured and recorded. Because of the daily limits of intrathecal contrast administration, two shots were planned during which 6 cc of Omnipaque 300 was then injected followed by 10 mL of saline at a rate of approximately 1 mL per second. The table was tilted 3° trendelenberg for a first shot of the thoracolumbar spine and 5° trendelenberg for a second shot of the thoracic spine. CSF was drained to a normal closing pressure and the spinal needle was removed. The DSM procedure was then repeated the following day with the patient positioned with the other side down to evaluate for a contralateral fistula.

Imaging evaluation

Patients with a potential diagnosis of SIH underwent MR imaging of the brain with and without gadolinium. These studies were independently reviewed by an attending neuroradiologist in addition to the senior author neurointerventionalist. As described by Dobrocky et al., the Bern score was calculated using the major (presence of pachymeningeal enhancement, engorgement of venous sinuses, effacement of the suprasellar cistern (≤4.0 mm)), and minor (subdural fluid collection, prepontine cistern ≤5.0 mm, mamillopontine distance of ≤6.5 mm) imaging criteria. ¹⁰ Major criteria (2 points each) and minor criteria (1 point each) were then totaled to give a score out of a possible 9 points. Preoperatively, patients also had noncontrast imaging of the spinal axis to evaluate for spinal epidural longitudinal CSF collection, which if present would suggest the presence of a ventral leak from discogenic bone spur or meningeal diverticulum. A repeat contrasted MRI brain was obtained approximately 3 months following treatment, and a postoperative Bern score was again calculated.

Clinical evaluation

Each patient was seen by both headache neurologist and the neurointerventionalist prior to their procedure. Patients were asked about the presence of headache, cognitive dysfunction, tinnitus, vision changes, gait disturbances, hearing abnormalities, dizziness, and movement disorders. Medical management and blood patches were usually attempted prior to digital subtraction myelogram and transvenous embolization. Postoperatively patients were kept in the hospital overnight to monitor for rebound headaches, and also assessed at 1 to 3 months following treatment to assess for sustained symptom improvement.

Endovascular technique

Transvenous embolization was performed in a biplane angiography suite. All patients underwent general anesthesia. This technique is similar to those described previously.^4,6,7 Femoral venous access was obtained with a 6 French sheath. A Benchmark guide catheter was then advanced over a Berenstein select catheter and 035 guidewire into the azygous vein. Once access was obtained into the azygous vein, a Scepter XC microcatheter was navigated over an 0.014″ Synchro Select microwire into the appropriate paraspinal vein. Venography was performed as needed to confirm the correct location. Onyx 18 or 34 was then injected through the Scepter XC microcatheter under balloon inflation via a “pressure cooker” technique, to achieve a robust Onyx cast of the adjacent foraminal venous plexus while minimizing reflux into the paraspinal vein. In one case a single coil was placed within the paraspinal vein to further minimize reflux. CT was performed immediately postoperatively to further confirm the extent of the Onyx cast.

Statistical analysis

Differences in mean continuous variables before and after intervention were calculated using a paired two-tailed t-test. A p-value of <0.05 was used to indicate statistical significance. All statistics and graph construction were completed using GraphPad Prism (Version 10).

Results:

Case illustration

The patient was a 67-year-old female who presented with 10 months of symptoms, including orthostatic headaches, brain fog, tinnitus, and sensorineural hearing loss. Contrasted brain MRI showed a Bern score of 8, lacking only subdural collections (Figure 1). Her headaches improved for 3 days with a blood patch. She had numerous perineural cysts on the spine MRI. Sedated DSM was performed, identifying early opacification of a venous network adjacent to the perineural cyst underneath the T9 pedicle on the left. Transfemoral transvenous embolization was performed through the 6F Benchmark guide catheter, Scepter 4 × 11 mm extra compliant balloon, and Onyx 18, casting the foraminal vein. Postembolization, the patient experienced rebound headache requiring prolonged Diamox taper over 2 months. A contrast brain MRI performed 3 months postembolization showed a Bern score of 1.

Figure 1.

Imaging findings for patient no. 1. (A) Preoperative magnetic resonance imaging demonstrating pachymeningeal enhancement. (B) Preoperative magnetic resonance imaging demonstrating decreased suprasellar cistern, prepontine space, and mamillopontine distance. (C): Digital subtraction myelography demonstrating the left T9 CSF-venous fistula. (D): Digital subtraction angiography demonstrating Onyx cast within the foraminal vein and intercostal vein following fistula embolization. (E) Postoperative magnetic resonance imaging demonstrating improvement in pachymeningeal enhancement. (F) Postoperative magnetic resonance imaging demonstrating improvement in suprasellar cistern, prepontine space, and mamillopontine distance.

Patient population

Six patients underwent transvenous embolization of a CSF-venous fistula. Demographic information and outcomes are shown in Table 1. The average age was 66.8 years old and four of the six patients were female (67%). The fistula was located in the thoracic spine in every case, with four patients having right-sided lesions and two having left-sided lesions. Technical procedural success was achieved in each case.

Table 1.

Patient demographics and characteristics before and after transvenous embolization.

Patient no.	Age	Symptoms	CVF location	Pre-Bern	Post-Bern	Rebound headache
1	67	Headache, cognitive dysfunction, tinnitus, gait disturbance	Left T9	8	1	1
2	58	Headache, hearing dysfunction, dizziness	Right T10	8	4	0
3	66	Headache, tinnitus	Right T12	5	3	0
4	74	Headache, cognitive dysfunction, gait disturbance	Left T8	8	3	0
5	66	Headache	Right T12	5	0	0
6	70	Headache	Right T6	7	0	1

Imaging outcomes

The mean Bern score preoperatively was 6.83 (SD = 1.47) and ranged from 5 to 8. The mean Bern score postoperatively was 1.83 (SD = 1.64) and ranged from 0 to 4. The mean improvement in Bern score was 5.0 (SD = 1.9, p = 0.0013, Figure 2). There was an improvement in the Bern score following embolization in every patient.

Figure 2.

Preoperative and postoperative BERN scores following transvenous embolization in patients with cerebrospinal fluid-venous fistula. ** = p < 0.01.

Clinical outcomes

On postoperative follow-up, three patients were noted to have complete resolution of their prior symptoms. Two patients had significant improvement of their symptoms on follow-up (defined as >75% improvement), but with persistence of at least some of their preoperative symptoms. One patient denied any significant improvement in their symptoms on follow-up. Despite having a two-point reduction of their Bern score on follow-up, this patient also had the lowest preoperative Bern score and the least amount of improvement. Two patients had rebound headaches requiring treatment with Diamox for a short period of time. In both cases, these headaches had resolved by 3 month follow-up.

Discussion

CSF-venous fistula is an increasingly recognized pathology which was historically treated with a blood patch or open surgical ligation. More recently, transvenous embolization has been described with low risk and excellent clinical and radiographic outcomes.^4–6 The presented case series demonstrated “real-world” efficacy of CSF-venous fistula embolization, further validating the findings of studies published by high-volume practices.

The workup of SIH involves several protocolized steps on which we have developed two subtle variations. The first step is a contrast brain MRI, on which numerous stigmata of SIH have been described, not limited to the six incorporated in the “Bern score.” ¹⁰ Next we obtain noncontrast images of the spinal axis to evaluate for extradural CSF collection, which can implicate a ventral discogenic bone spur and is better evaluated with prone myelography. ¹¹ MRI of the spinal axis often reveals perineural cysts, which can be a source of CSF-VF, but our institutional experience reveals are physiologic in the vast majority of occasions. ¹² This insight has been useful to reinforce the diagnostic algorithm among referring physicians who might be inclined to refer patients with an isolated finding of perineural cysts on MRI. Spine MRI for Case no 6 revealed numerous perineural cysts; DSM with lateral fluoroscopy would later identify one of these as the source of a high-flow CSF-venous fistula (Figure 3). Lumbar puncture opening pressure is explicitly not a part of the diagnostic algorithm for SIH patients, but we nevertheless find it relevant context with no downside at the beginning of a DSM. ¹³

Figure 3.

Imaging findings for patient no. 6. (A) Preoperative magnetic resonance imaging demonstrating pachymeningeal enhancement consistent with intracranial hypotension. (B) CT myelography demonstrating multiple perineural cysts. (C) Digital subtraction myelography demonstrating the right T6 CSF-venous fistula. (D) Axial CT myelography demonstrating the Onyx cast following embolization of the CSF-venous fistula. (E) Sagittal CT myelography demonstrating the Onyx cast following embolization of the CSF-venous fistula. (F) Postoperative magnetic resonance imaging demonstrating improvement in pachymeningeal enhancement.

CT: computed tomography; CSF: cerebrospinal fluid.

The steepest part of the learning curve in managing SIH patients with CSF-VF is obtaining excellent diagnostic images. Among high-volume centers, there are programs that favor digital subtraction myelography ¹⁴ and others that favor CT myelography. ¹⁵ We prefer DSM but keep the patient in a lateral position for immediate CT of the cervical, thoracic, and lumbar spine to assess for slower-flow fistula and/or fistula located outside the frame (for instance, in the upper cervical spine). ¹⁴ Whether to sedate, intubate, and paralyze patients to minimize motion is controversial. Motion artifact can render an appearance similar to venous structures on digital subtraction, which resolves on unsubtracted imaging. There are compelling case reports of awake patients performing “resisted inspiration” through a straw to increase CSF-VF conspicuity by reducing intrathoracic pressure and increasing venous return.^16,17 We have found this challenging to execute in practice. Our patient no 4 initially underwent DSM under monitored anesthesia care with resisted inspiration, which did not demonstrate a fistula. Due to a high Bern score and persistent symptoms, repeat DSM was performed two months later, this time under general anesthesia. A fistula was then identified and subsequently embolized at the left T8 level, resulting in complete resolution of symptoms and improvement of Bern score from 8 to 3 on follow-up (Figure 4). Rotating the B-plane detector maximizes the field of view, increasing the number of levels that can be captured in two runs and overlap in the low thoracic region which is most commonly the source of CSF-VF. The use of biplane also allows for detection of fistulae that may be less apparent on a single plane, such as in patient no 6 (Figure 3) in our series.

Figure 4.

Imaging findings for patient no. 4. (A) Preoperative magnetic resonance imaging demonstrating decreased suprasellar cistern, prepontine space, and mamillopontine distance consistent with intracranial hypotension. (B) Preoperative magnetic resonance imaging demonstrating pachymeningeal enhancement consistent with intracranial hypotension. (C) Initial digital subtraction myelogram with no visualized fistula. (D) Repeat digital subtraction myelogram demonstrating left T8 CSF-venous fistula. (E) Digital subtraction angiography demonstrating coils and Onyx cast within the intercostal vein, foraminal vein, and adjacent venous plexus following fistula embolization. (F) Postoperative magnetic resonance imaging demonstrating improvement in suprasellar cistern, prepontine space, and mamillopontine distance. (G) Postoperative magnetic resonance imaging demonstrating improvement in pachymeningeal enhancement.

CSF: cerebrospinal fluid.

All CSF-VF treated with transvenous embolization at our institution to date were in the lower thoracic spine. According to Berg et al., approximately 70% of CSF-VF originate from these levels with a right-sided predominance. ⁷ Although upper extremity venous access has been reported, ⁶ we have exclusively used a transfemoral approach. We favor the 6F benchmark sheath, which provides adequate support while tracking easily over the 5F Berenstein select catheter and 035 glidewire which can reliably find the azygous vein over the right main stem bronchus. To ensure penetration of the paravertebral plexus lateral to the foraminal vein, we typically use a dual-lumen balloon microcatheter. We recently began placing a coil in the segmental vein to eliminate this sink and ensure better craniocaudal penetration of the paravertebral plexus (Figure 4). This so-called “pressure cooker” technique has also been described by the Mayo Clinic group in Jacksonville, ¹⁸ Additional onyx can be injected after taking the balloon down to ensure penetration of the epidural venous plexus and casting of the intervening foraminal vein. Based on reports of recurrence and the need for repeat embolization at the same level, ⁴ the mistake would appear to be insufficient embolization rather than excessive spread. We have not seen or read about the mass effect or venous congestion from excessive embolic material.

Our experience supports others’ observation of an indirect relationship between the duration of preoperative symptoms and the likelihood of relief following CSF-venous fistula embolization. Hani et al. surveyed 86 patients who underwent embolization of a CSF-venous fistula, and found that shorter symptom duration was associated with improved outcomes. ¹⁹ In our series, patients with symptoms present for at least five years prior to their embolization, appeared to have lower rates of improvement on postoperative MRI, with Bern score reduction from 8 to 3 and 8 to 4. In contrast, those presenting with less than 1 year of symptoms improved from a Bern score of 5 to 0 and 7 to 0 postoperatively (Table 1).

Conclusions

We present a single-center, retrospective “real-world’ case series of six patients who underwent endovascular embolization of CSF-venous fistula. These fistulas are an increasingly recognized clinical entity associated with intracranial hypotension and debilitating symptoms, and our findings validate the efficacy of this technique in lower-volume centers. Our experience provides an important addition to the literature and further studies are necessary to further explore the long-term durability of this technique.