Spinal dural arteriovenous fistulas (SDAVFs) typically represent abnormal shunts between a radiculomeningeal artery and radicular vein with retrograde drainage into perimedullary veins. SDAVFs are rare entities, with an estimated incidence rate of 5–10 per million per year and represent the most frequent vascular malformation of the spine accounting for approximately 70% of all spinal arteriovenous shunts.1–4 SDAVFs are more often found in men, being affected five times more commonly than women, with a mean age of 55–60 years at diagnosis.5
It is presumed that SDAVFs are acquired diseases, although their exact etiology remains unknown. The increase in spinal cord venous pressure, due to arterialization of the perimedullary veins, diminishes the arteriovenous pressure gradient leading to decreased drainage of normal spinal cord draining veins and therefore venous congestion.6,7 This congestion can lead to intramedullary edema, chronic hypoxia, and progressive myelopathy.8
The fistulous point of an SDAVF is most commonly located within the dura mater close to the spinal nerve root, where the arterial blood from the segmental radiculomeningeal artery (i.e., the artery supplying the nerve root and meninges and occasionally the spinal cord) directly shunts into a radicular vein, at the point where the latter traverses the dura on the dorsal aspect of the dural root sleeve in the intervertebral foramen. Consequently, the fistulous point classically presents directly underneath the pedicle of the vertebral body supplied by the injected segmental artery on digital subtraction angiography (DSA) (Fig. 1).9 Typically, a single feeding dural branch of the radiculomeningeal artery connects to the radicular vein to form the shunt.
Features of a typical SDAVF. A and B: Sagittal MR images of the spine demonstrating a T2 hyperintense signal from the midthoracic spine to the conus medullaris, with flow voids dorsal to the cord representing engorged perimedullary veins (A) and patchy cord enhancement on a postcontrast T1-weighted sequence (B). C: Selective angiogram of the right T8 segmental artery revealing the dural arteriovenous shunt fed by a single radiculomeningeal artery. D: Unsubtracted image showing the exact point of fistulization (white asterisk) beneath the right T8 pedicle. E: Schematic drawing showing the feeding radiculomeningeal artery (red arrow) forming the fistula (green asterisk) within the dura on the dorsal aspect of the nerve root sleeve (dotted orange line) directly beneath the pedicle, with the draining radicular vein (blue arrow). Figure is available in color online only.
While this shunt location and angiographic morphology are well known, dural arteriovenous shunts can also present with more than one arterial feeder, a so-called bimetameric supply. In a recent series, Kiyosue et al. reported that an SDAVF with multiple feeding arteries arising from two ipsilateral segmental arteries at adjacent spinal levels occurred in 13% of cases, while an SDAVF with multiple arterial feeders from bilateral segmental arteries was seen in 4.6% of cases.10
In addition, we recently recognized an SDAVF with a variant location of the fistulous point, distinct from the typical location below the pedicle of the vertebral body. To the best of our knowledge, case reports on this newly described entity are lacking, and their morphology is distinct from spinal epidural AVFs which are better described.11 In this case series, we present patients with a so-called variant type of SDAVF and report on their incidence, epidemiology, angiographical findings, and treatment.
Methods
Following institutional ethics review board approval, a retrospective analysis of our prospectively maintained SDAVF database was conducted. We evaluated all cases of SDAVF treated at our institution between January 1, 2008, and June 15, 2020.
Spinal DSA was reviewed by a fellowship-trained interventional neuroradiologist to identify patients with a variant location of the spinal dural arteriovenous shunt. Data for analysis were anonymized. Age, sex, clinical presentation, angiographic findings, treatment modality, and clinical outcomes were reviewed.
Patients older than 17 years of age who had spinal DSA demonstrating an SDAVF with a variant type of location (i.e., the fistulous point not beneath the pedicle) and relevant clinical information and imaging available were included.
Results
We identified 67 patients with SDAVFs treated at our institution, 8 of whom were subsequently excluded because of lack of available imaging/clinical information. Of the remaining 59 patients, 55 demonstrated a typical SDAVF with the point of fistulization presenting beneath the pedicle. In the remaining 4 patients (6.8%), the point of fistulization was found in a variant location, more posterior along the dura of the spinal canal, at the level between two adjacent vertebral bodies (Fig. 2). In 3 (75%) of these patients, a bimetameric arterial supply was also demonstrated. Of these, 2 patients required two treatment sessions to achieve definitive treatment. In contrast, typical SDAVFs demonstrated bimetameric supply in just 12 (21.8%) of 55 patients in our series, with 7 patients (12.7%) having a contralateral supply via retrocorporeal anastomoses.
Case 1. Example of variant SDAVF anatomy. A and C: Injections of the right T5 (A) and T6 (C) segmental arteries, revealing a dural arteriovenous shunt refluxing into perimedullary vessels, with a bimetameric supply. B and D: Respective unsubtracted images showing that this shunt is located superomedial to the right T6 pedicle (white asterisk), at the level of the intersegmental disc space. E: Schematic drawing of the variant SDAVF, showing the point of fistulization (green asterisk) on the dorsal dura mater (dotted orange line) more medially within the spinal canal, in the region of the intervertebral disc space. As a consequence of their intervertebral location, the variant type of SDAVF will have a propensity to recruit arterial supply from more than one metameric level (red arrows). Figure is available in color online only.
The mean age at the time of diagnosis of a variant SDAVF was 80 years (range 59–88 years) and that at the time of diagnosis of a typical SDAVF was 65 years (range 34–87 years). The mean time from symptom onset to diagnosis was 19.5 months (range 6–48 months) in patients with variant SDAVFs and 14 months (range 0–64 months) in patients with typical SDAVFs.
Symptoms encountered in the patients with variant SDAVFs were similar to those expected in patients with typical SDAVFs, with progressive myelopathy characterized by lower-limb paresthesia, gait disturbance, and sphincter dysfunction (Table 1).
Patient characteristics and treatment results
Case No. | Age (yrs), Sex | Presenting Symptoms | Time From Symptom Onset to Diagnosis (mos) | Location of Fistula | Bimetameric Supply | Treatment Year | Treatment Modality | Treatment Result |
---|---|---|---|---|---|---|---|---|
1 | 88, F | Progressive lower-limb weakness, urinary & bowel incontinence | 6 | Rt T5–6 | Yes | 2020 | Surgery | Complete occlusion |
2 | 86, M | Progressive lower-limb weakness | 48 | Lt T5–6 | Yes | 2013 | 1) Surgery 2) Surgery | 1) Incomplete occlusion 2) Complete occlusion |
3 | 86, M | Progressive lower-limb weakness, urinary incontinence | 6 | Rt T10 | No | 2010 | Endovascular | Complete occlusion |
4 | 59, M | Progressive lower-limb weakness, urinary & bowel incontinence, erectile dysfunction | 18 | Lt T4–5 | Yes | 2008 | 1) Endovascular 2) Surgery | 1) Incomplete occlusion 2) Complete occlusion |
Discussion
The main findings of our study are: 1) a variant location of the arteriovenous shunt in patients with SDAVF is a rare but not uncommon entity, representing 6.8% of all SDAVFs in our series; 2) variant SDAVFs are unusual in their preponderance for demonstrating a bimetameric arterial supply; and 3) these SDAVFs may be more refractory to treatment.
For the first time, we describe a rare variation in the location of SDAVFs that are located more posteromedially along the dura, between two adjacent vertebrae, rather than laterally on the nerve root sleeve underneath the pedicle (Fig. 3). Given their location, these shunts have a high propensity to receive blood supply from two adjacent segmental levels.
Case 4. Example of variant SDAVF anatomy. A and C: A dural arteriovenous shunt is seen after injection of the left T4 (A) and T5 (C) segmental arteries. B and D: Unsubtracted images revealing the point of fistulization (white asterisk) superomedial to the left T5 pedicle. E–G: Intraoperative photographs demonstrating the arterialized radicular vein as it exits the dura (E), which, when ligated at the dura (F), immediately decreased in caliber, signifying successful disconnection of the fistula (yellow arrowhead, G). Figure is available in color online only.
A previously proposed classification system for craniospinal dural arteriovenous shunts described three main venous drainage patterns according to anatomical location (toward the ventral, lateral, or dorsal epidural venous spaces), with SDAVFs representing dural arteriovenous shunts of the lateral epidural group.12 Unlike radicular arteries, which closely follow the peripheral nerves, radicular veins do not necessarily do so, with up to 40% of spinal radicular veins not draining in the expected location alongside the nerve root.13,14 This variation in spinal radicular venous drainage therefore predicts the variant locations of arteriovenous shunts seen in variant SDAVFs.
Although small in number, these variant SDAVFs appear to be more common in older patients, with a mean age at diagnosis of 80 years compared with 65 years for typical SDAVFs. Despite differences in anatomy, the clinical presentation demonstrated in this group was similar to the classic symptoms of progressive myelopathy.
Of the variant SDAVFs, 50% required two treatment sessions to achieve complete occlusion of the fistula. One of these patients presented with a recurrence after unsuccessful surgical disconnection at an outside institution, and the other patient had a failed glue embolization and underwent surgery during the same admission. In this endovascular case, the glue did not penetrate into the venous pedicle of the fistula. We postulate that this might have occurred due to wash-in from the second arterial feeder, causing early glue polymerization. By contrast, definitive treatment was achieved in a single treatment session in all but 7 (12.7%) of the 55 typical SDAVFs.
At our center, we have moved toward surgery as the first-line treatment for SDAVF disconnection, except in cases in which the lumbosacral region is affected. In centers that still employ glue embolization as the first-line therapy, recognition of these variants is important, as our experience might suggest that achieving definitive treatment with glue could prove more difficult. In this setting, on recognition of a variant SDAVF with bimetameric supply, surgeons may wish to adjust their treatment algorithm and consider adopting a surgical approach instead.
In centers that primarily perform embolization with the aid of ethylene vinyl alcohol copolymers, decisions regarding treatment modality would likely be unaffected, as wash-in from a second arterial feeder is unlikely to pose a risk to achieving the desired progression of embolic material.
Previous reports have described an unfavorable natural history of SDAVFs with progression of neurological symptoms with time.15 Aminoff and Logue found that 19% of patients were severely disabled within 6 months of the onset of their motor disturbances, and by 3 years, this number had grown to 50%.16 Furthermore, less than 10% of patients will be able to walk independently after 3 years.17 Therefore, treatment with endovascular or surgical disconnection of the fistula is indicated without delay. The goal of treatment of an SDAVF is to arrest neurological deterioration and, perhaps, improve symptoms in selected cases. Improvement in motor function after treatment is more likely to occur than micturition dysfunction. Studies have also revealed a correlation with the duration of symptoms and treatment outcome.18,19
Understanding the radiological features found on MRI and spinal angiography is fundamental to successfully treat SDAVFs. Often the diagnosis is suggested on MRI, with spinal angiography allowing confirmation and localization of the fistula for definitive management, either by endovascular or surgical means. Precise localization of the fistulous point is of critical importance for the surgical treatment and exposure, as the variant type of SDAVF described in this case series may pose an additional challenge for surgical treatment if the atypical fistula location is not recognized preoperatively.
In 3 of the 4 variant SDAVFs reported in this study, there were multiple feeders present, involving two ipsilateral adjacent segmental arteries. The fistulous points were located between two vertebrae at the disc space level and were more centrally located in the canal. The specific anatomical characteristics of this variant SDAVF should be well understood and differentiated from the typical SDAVF, which can sometimes present with multiple ipsilateral feeders from anastomosis between two segmental arteries, but in which the location of the fistula will classically remain underneath the pedicle, in distinction from the variant cases presented in this article.
This study is limited by the fact that variant SDAVFs are rare, accounting for just 4 cases or 6.8% of the SDAVF cases managed at our center during the study inclusion period. Interpretations drawn from our results should therefore take this small sample size into account.
Conclusions
Variant types of SDAVFs are located at the intervertebral level, more medially within the spinal canal, in contrast to the typical SDAVF shunting zone encountered directly beneath the pedicle. Recognition of this newly described anatomical subtype of SDAVF is important, as localization of the point of fistulization is critical for successful surgical disconnection, thereby preventing delay in achieving definitive treatment.
Disclosures
Dr. Krings: consultant for Stryker, Penumbra, MicroVention, and Cerenovus; direct stock ownership in Marblehead; and royalties from Thieme.
Author Contributions
Conception and design: O’Reilly, Brunet, Krings. Acquisition of data: O’Reilly. Analysis and interpretation of data: O’Reilly. Drafting the article: O’Reilly, Hendriks, Krings. Critically revising the article: O’Reilly, Hendriks. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: O’Reilly. Statistical analysis: O’Reilly. Study supervision: Radovanovic, Krings.
References
- 1↑
Kendall BE, Logue V. Spinal epidural angiomatous malformations draining into intrathecal veins. Neuroradiology. 1977;13(4):181–189.
- 2
Merland JJ, Riche MC, Chiras J. Intraspinal extramedullary arteriovenous fistulae draining into the medullary veins. J Neuroradiol. 1980;7(4):271–320.
- 3
Behrens S, Thron A. Long-term follow-up and outcome in patients treated for spinal dural arteriovenous fistula. J Neurol. 1999;246(3):181–185.
- 4↑
Krings T, Geibprasert S. Spinal dural arteriovenous fistulas. AJNR Am J Neuroradiol. 2009;30(4):639–648.
- 5↑
Jellema K, Tijssen CC, van Gijn J. Spinal dural arteriovenous fistulas: a congestive myelopathy that initially mimics a peripheral nerve disorder. Brain. 2006;129(Pt 12):3150–3164.
- 6↑
Kataoka H, Miyamoto S, Nagata I, Ueba T, Hashimoto N. Venous congestion is a major cause of neurological deterioration in spinal arteriovenous malformations. Neurosurgery. 2001;48(6):1224–1230.
- 7↑
Hurst RW, Kenyon LC, Lavi E, Raps EC, Marcotte P. Spinal dural arteriovenous fistula: the pathology of venous hypertensive myelopathy. Neurology. 1995;45(7):1309–1313.
- 8↑
Krings T, Lasjaunias PL, Hans FJ, et al. Imaging in spinal vascular disease. Neuroimaging Clin N Am. 2007;17(1):57–72.
- 9↑
Krings T, Lasjaunias PL, Geibprasert S, et al. Classification of spinal vascular malformations. Neuroradiol J. 2009;22(1 suppl):97–106.
- 10↑
Kiyosue H, Matsumaru Y, Niimi Y, et al. Angiographic and clinical characteristics of thoracolumbar spinal epidural and dural arteriovenous fistulas. Stroke. 2017;48(12):3215–3222.
- 11↑
Nasr DM, Brinjikji W, Clarke MJ, Lanzino G. Clinical presentation and treatment outcomes of spinal epidural arteriovenous fistulas. J Neurosurg Spine. 2017;26(5):613–620.
- 12↑
Geibprasert S, Pereira V, Krings T, et al. Dural arteriovenous shunts: a new classification of craniospinal epidural venous anatomical bases and clinical correlations. Stroke. 2008;39(10):2783–2794.
- 13↑
Lasjaunias P, Berenstein A, ter Brugge KG. Spinal and spinal cord arteries and veins. In: Lasjaunias P, Berenstein A, ter Brugge KG. Clinical Vascular Anatomy and Variations. Springer-Verlag Berlin Heidelberg; 2001:154–158.
- 14↑
Berenstein A, Lasjaunias P, ter Brugge KG. Spinal dural arteriovenous fistulae. In: Berenstein A, Lasjaunias P, ter Brugge KG. Surgical Neuroangiography. Vol 2: Clinical and Endovascular Treatment Aspects in Adults. Springer-Verlag Berlin Heidelberg; 2004:851-852.
- 15↑
Gemmete JJ, Chaudhary N, Elias AE, et al. Spinal dural arteriovenous fistulas: clinical experience with endovascular treatment as a primary therapy at 2 academic referral centers. AJNR Am J Neuroradiol. 2013;34(10):1974–1979.
- 16↑
Aminoff MJ, Logue V. The prognosis of patients with spinal vascular malformations. Brain. 1974;97(1):211–218.
- 17↑
Dehdashti AR, Da Costa LB, terBrugge KG, Willinsky RA, Tymianski M, Wallace MC. Overview of the current role of endovascular and surgical treatment in spinal dural arteriovenous fistulas. Neurosurg Focus. 2009;26(1):E8.
- 18↑
Song JK, Viñuela F, Gobin YP, et al. Surgical and endovascular treatment of spinal dural arteriovenous fistulas: long-term disability assessment and prognostic factors. J Neurosurg. 2001;94(2 Suppl):199–204.
- 19↑
Safaee MM, Clark AJ, Burkhardt JK, Winkler EA, Lawton MT. Timing, severity of deficits, and clinical improvement after surgery for spinal dural arteriovenous fistulas. J Neurosurg Spine. 2018;29(1):85–91.