Comparison of 3 Different Types of Spinal Arteriovenous Shunts below the Conus in Clinical Presentation, Radiologic Findings, and Outcomes

Comparison of Angioarchitecture

The angioarchitecture of the 3 types of spinal arteriovenous lesions below the conus share some common features. Our study showed that most²⁰ of these shunts were fistulous in nature, with nidus-type shunts being very rare and indicating rAVS if present. In addition, most shunts had a relatively slow flow and thus only mildly dilated arteries and veins. Furthermore, the venous drainage pattern was similar because in all shunts, the veins of the shunts drained upward to the perimedullary veins and thus caused venous congestion of the spinal cord. This drainage pattern explains that progressive myelopathy was the main symptom of all 3 types and hemorrhage occurred only rarely. In our series, it only occurred in 1 patient with an rAVS. Despite the above similarities, the 3 types of lesions have distinctive features to differentiate.

The filum terminale extends downward from the apex of the conus medullaris to the cul de sac of the dura at S2, so the FTAVF should be fed mainly by the filum terminale artery as the caudal continuation of the anterior spinal artery below the arcade of the cone. Whether the filum can have other blood supplies besides the filum terminale artery is still debatable^13,14; as Djindjian et al⁵ reported, the coccygeal nerve with a radicular artery supply was adherent to the filum in its proximal portion. Thus, an explanation for FTAVFs localizing in the distal portion of the filum having an additional blood supply could be a radicular artery or dural artery, as shown in our series and in 1 previous case report.¹⁰ In some circumstances, a radicular artery may adhere to the distal portion of the filum and provide an extra vascular supply. The other possibility is a dural supply because the fistula localizes to the dural attachment point of the filum at around the S2 level, which then may include the dural arterial supply.

In our series, rAVSs were the type least often encountered, which is reflected by the literature because previously only 2 cases were reported.¹⁶ The arterial vascularization of the cauda equina nerve roots includes both distal and proximal radicular arteries with an anastomosis at the proximal one-third of the root.⁶ Ohtonari et al¹⁶ reported 2 cases of rAVSs fed by proximal radicular arteries, while in our study, all 7 rAVSs were fed by distal radicular arteries from segmental arteries. We believe that at least some of our cases may have also had proximal radicular artery feeders, which are suppressed by the main flow from their distal counterpart.

SDAVFs below the conus usually present with a single segmental feeding artery, which implies that the fistula is located within the dura directly adjacent to the nerve root (ie, where the radiculomeningeal artery pierces the dura). In cases in which the fistula is located between 2 adjacent nerve roots, dual segmental arterial supply from adjacent segmental arteries may be seen. This configuration was present in 3 patients in our series with SDAVFs supplied by 2 lumbar arteries or bilateral iliac arteries. During embolization, embolic agents should always approach the proximal venous end of the fistula to achieve complete obliteration, which may be more difficult to achieve with N-butyl-cyanoacrylate (glue) in cases of dual supply because the glue will harden on contact with blood from the second feeder.

Embryologic Consideration

Development of the human spinal cord involves both primary and secondary neurulation. Unlike the primary neurulation, which establishes the brain and spinal cord and is derived from the ectoderm, all neural elements of secondary neurulation are developed from pluripotent mesodermal cells. These cells, termed the “caudal cell mass,” coalesce and then epithelialize to form a separate neural tube that will connect with the neural tube formed by the primary neurulation.^21,22 The exact level where the 2 neural tubes meet is still debatable, ranging from the upper conus²³ to the lowest sacral level of the conus²¹; however, it is agreed that the caudal cell mass will give rise to the filum terminale and the cauda equina. Thus, the embryologic origin of FTAVFs and rAVSs below the conus is associated with secondary neurulation, differing from the remainder of intradural spinal vascular shunt origins that are associated with the primary neurulation. This may explain the interesting finding in our study that FTAVFs below the conus were often associated with spinal dysraphisms such as tethered cord, spinal lipoma, and spina bifida; this association was also shown in a previous case report.²⁴ One may hypothesize that patients with abnormal secondary neurulation are more prone to develop abnormal vascular shunts: If during embryologic development, pluripotent cells at the caudal end of the human embryo were abnormally triggered to form a lipoma that will cause tethering, these abnormal pluripotent cells may also be prone to develop abnormal vascular shunts.²⁵ The reason why SDAVFs were associated with dysraphisms may also be due to the abnormal pluripotent cells, which have a propensity to develop this acquired vascular lesion. However, why rAVSs were not associated with spinal dysraphisms in this series remains unclear.

The interesting finding that some patients demonstrated coexistence of rAVS with a conus AVM is presumably related to a low spinal arteriovenous metameric syndrome. We hypothesize that similar to thoracic, cervical, and cerebral metameric syndromes, abnormal cells that are prone to form arteriovenous shunts migrated along their segment and seeded daughter cells along their migrational path as previously described in cerebral arteriovenous metameric syndromes (Wyburn-Mason syndrome).²⁶

Differential Diagnosis and Treatment

Spinal arteriovenous shunts below the conus may not be as rare as was previously thought because they represented 13.3% of all spinal shunts in our data base. However, misdiagnosis is not uncommon according to the similar clinical manifestation and nonspecific radiologic findings among the different groups. Our study suggests some differential diagnostic clues. rAVSs show a younger age at first presentation, a larger female proportion, and a higher incidence of pain symptoms and hemorrhage. FTAVFs and SDAVFs share a similar age at onset, which may be related to the same pathophysiology of venous hypertension. However, FTAVFs present with a different feeding artery derived from the anterior spinal artery and they are more commonly associated with dysraphic malformations. A standard spinal angiography, including the bilateral internal iliac arteries and median sacral artery, is the criterion standard to diagnose a lesion in this region. Treatment outcomes are better compared with spinal vascular lesions in other regions. Although embolization resulted in only a 56.0% complete obliteration rate in our series (given the often tortuous vascular anatomy and small caliber of the feeding arteries), surgical resection achieved 100% cure because disconnection of the venous outflow could be easily achieved after accurate intraoperative identification of the draining vein.

Limitations

Our study has some limitations pertaining to its retrospective design and sample volume. Due to the rarity of the disease and the potential of referral basis, our data may not represent the situation in the general population. Because some statistical comparisons are based on small numbers, there is the possibility of spurious statistically significant associations. Additionally, some cases were associated with additional abnormalities (ie, conus AVMs or spinal dysraphic disorders); thus, patient symptomatology may be related to various sources.