A case of filum terminale arterial venous fistula needed a long arterial access for trans-arterial shunt obliteration

Abstract

A 78-year-old man was referred to our institution with a predominantly progressive numbness of both legs, and bladder dysfunction with urinary retention. He was diagnosed as the symptomatic arteriovenous fistula of the filum terminale (AVFFT). A trans-arterial embolization (TAE) of the arteriovenous shunt was planned for his symptomatic AVFFT. The long distance between the origin of the radiculo meningeal artery (Th8) and the site of the fistula (S1) resulted in the first TAE having a feeder occlusion. The length of accessible feeder in the first TAE was the longest (about 40 cm) as the past reports of the endovascular therapy. However, complete shunt occlusion was accomplished at a second session two weeks after the initial TAE because a more accessible feeder was developed by the initial feeder occlusion.

Introduction

Arteriovenous fistula (AVF) of the filum terminale (AVFFT) is an extremely rare type of spinal arteriovenous shunt disease that is defined as an abnormal communication of the artery of the filum terminale (the continuation of the anterior spinal artery (ASA)) and the spinal vein (perimedullary vein).^1,2 This direct communication leads to an increase in venous pressure in the perimedullary veins that may result in spinal cord congestion and edema. Clinical presentation is usually the result of progressive myelopathy or conus medullaris syndrome.

Both endovascular treatment and surgery have been used to manage this condition. Endovascular treatment is usually technically difficult, due to the long and tortuous course of the ASA and the artery of the filum terminale supplying the AVFFT.^3,4 We report on an elderly patient presenting with symptomatic AVFFT with supply from the artery of the ASA. The length of accessible feeder in the first trans-arterial embolization (TAE) was the longest according to the past reports of the endovascular therapy. Complete shunt occlusion was accomplished at a second session two weeks after the initial TAE because a more accessible feeder was developed after the initial feeder occlusion.

Case report

A 78-year-old man was referred to our institution with a predominantly progressive numbness of both legs, and bladder dysfunction with urinary retention. These symptoms were first observed several years earlier. The patient reported a fluctuating but essentially progressive course. The medical history was dyslipidemia.

The patient had undergone a neurological workup. Lumbar magnetic resonance imaging (MRI) showed an edema of the conus medullaris and the thoracic spinal cord as well as dilated and tortuous spinal perimedullary veins (Figure 1(a)). Digital subtraction angiography (DSA) of the thoracic, lumbar, and sacral segmental arteries was performed, and showed an arteriovenous shunt at S1 level, supplied mainly by the left radicular medullary artery at Th 8–9 level (a branch from Adamkiewicz artery) (Figure 1(b),(c)) and fed by a tiny branch of the left lateral sacral artery (LSA) (Figure 1(d)), which was not an accessible feeder. The length of the accessible feeder from the Adamkiewicz artery was about 40 cm. Therefore, we considered the pathology of the patient was AVFFT with dural components, occurring secondarily.

Figure 1.

A T2-weighted image of the thoracic spine in a patient with progressive numbness of both legs, and bladder dysfunction with urinary retention showing an edema arising from the middle to the lower thoracic spinal cord, and demonstrating the abnormally dilated and tortuous vessels situated on the surface of the spinal cord (a). Digital subtraction angiography (DSA) with selective injection of the ninth thoracic segmental artery on the right side (b) showing a hypertrophic radiculo medullary artery supplying the anterior spinal artery that directly shunted into a dilated draining vein anterior to the filum terminale at the level of the first sacral vertebral disc, representing an arteriovenous fistula of the filum terminale (c), in addition, fed by a tiny branch of left lateral sacral artery (LSA) (d: Shunt point is shown as yellow arrow)

The feeder was very long at 40 cm and direct shunt obliteration was an adequate option for his AVFFT; however his cardiopulmonary function was not adequate for general anesthesia. The TAE of the arteriovenous shunt via the feeder from the Adamkiewicz artery was planned for the symptomatic AVFFT.

The procedure was performed under local anesthesia. After systemic heparinization, a 5Fr Cobra (Terumo, Tokyo, Japan) was placed to the distal portion, having backup force, in the right segmental artery (intercostal artery) at Th 9 and introduced to the orifice of the radicular artery via the right femoral artery (Figure 2(a)). We attempted to use 1.2 Fr Magic (Pur Sil 45 cm; Balt Extrusion, Montmorency, France) microcatheter at first; however we could not access the feeder of the AVF via the ASA with 1.2 Fr Magic microcatheter by the flow-guided method, and in addition, 0.07 size microguide wire is not available in Japan at present. Therefore, we accessed the feeder with a 1.5 Fr flow-dependent Magic microcatheter (Pur Sil 45 cm; Balt Extrusion, Montmorency, France) without micro-guidewire by the saline flash method. A 1.5 Fr Magic MP (Pur Sil 45 cm) with Chikai 0.008 (ASAHI Intec, Tokyo, Japan) was introduced into the distal portion of the ASA at the L1 level. Provocation test by the injection of 2% xylocaine at L1 level was negative. Heated 12.5 % n-butyl 2-cyanoacrylate (NBCA) was injected via the Magic catheter (Figure 2(b),(c)). DSA after TAE showed no shunt flow and the patency of ASA (Figure 2(d),€). We could not introduce the Magic catheter sufficiently proximal to the shunt point due to the very long feeder, so we performed a feeder occlusion. After the first procedure, post-procedural DSA showed no shunt. Post-procedural MRI showed no feeder, however the draining vein was the same as at the pre-procedural state. However, the symptoms of the patient persisted, and the recurrence of AVFFT was found on DSA two weeks after the TAE (Figure 3(a)). DSA showed an arteriovenous shunt at the S1 level fed by a branch of the left LSA, which became a more accessible feeder than was obtained at the first DSA. Fortunately, a more accessible feeder, which was not found on the first TAE, was developed by the initial feeder occlusion. TAE of the arteriovenous shunt via the branch of LSA was planned for the recurrent AVFFT.

Figure 2.

Trans-arterial shunt embolization was performed under local anesthesia. A guiding catheter was placed in right segmental artery at Th 9 and introduced to the orifice of radicular artery (a). Heated 12.5 % NBCA was injected via a microcatheter whose tip was placed at L1 vertebral level (b) red arrow shows the tip of microcatheter and pink arrow shows ends of casts of NBCA in DSA at L1-L5 level; (c) DSA at L5 level. DSA after TAE showed no shunt flow and the patency of ASA (d) DSA at L1 level, (e) DSA at L5 level.

ASA: anterior spinal artery; DSA: digital subtraction angiography; NBCA: n-butyl 2-cyanoacrylate; TAE: trans-arterial embolization

Figure 3.

Recurrence of AVFFT was found two weeks after the TAE (a). DSA showed an arteriovenous shunt at S1 level fed by a branch of left LSA. TAE of the arteriovenous shunt via the branch of LSA was planned for the recurrent AVFFT. TAE was performed under local anesthesia. A guiding catheter was introduced to the branch of left LSA (b), then a microcatheter was introduced close to the shunt point (c). Heated 12.5 % NBCA was injected via the microcatheter. DSA after TAE showed no shunt flow (d).

AVFFT: arteriovenous fistula of the filum terminale; DSA: digital subtraction angiography; LSA: lateral sacral artery; NBCA: n-butyl 2-cyanoacrylate; TAE: trans-arterial embolization

The procedure was performed under local anesthesia. After systemic heparinization, a 4.2 Fr RC 2 (Terumo, Tokyo, Japan) was placed in the left internal iliac artery, and a 4 Fr Headhunter (Terumo, Tokyo, Japan) introduced, having backup force, to the branch of left LSA (Figure 3(b)). A 1.5 Fr Magic MP (Pur Sil 45 cm; Balt Extrusion, Montmorency, France) with Chikai 0.008 (ASAHI Intec, Tokyo, Japan) was introduced close to the shunt point (Figure 3(c)). Provocation test by the injection of 2% xylocaine via Magic catheter was negative. Heated 12.5 % NBCA was injected via a Magic catheter. DSA after TAE showed no shunt flow (Figure 3(d)). This time, NBCA was penetrated into the shunt point to the draining vein. Post-procedural MRI showed the complete occlusion of the shunt. MRI after this procedure demonstrated a complete disappearance of the dilated perimedullary veins (Figure 4(a)–(f)). The patient's clinical condition including urinary disturbance and gait disturbance improved and he was discharged on foot. The patient's condition 3 months after TAE was at 0 on the modified Rankin Scale; disappearance of high signal intensity lesion and flow void of the conus medullaris and the thoracic spinal cord on T2 weighted lumber MRI. No recurrence has been observed to date for a year.

Figure 4.

Follow up T2-weighted MRI at 1 year after the second TAE, confirming regression of the spinal cord edema and flow void. (a) sagittal image at Th7-9, (b) sagittal image at Th10-12, (c) sagittal image at L1-5, (d) axial image at Th8, (e) axial image at Th10, F: Axial image at Th12/L1.

MRI: magnetic resonance imaging; TAE: trans-arterial embolization

Discussion

Spinal vascular malformations can be classified according to the location of the arteriovenous shunt.³ AVFFTs are small and single AVFs supplied by the terminal portion of a thin ASA and the shunt, which is located on the anterior aspect of the conus medullaris or of the filum terminale, slows drainage in the ascending perimedullary veins.^4–6 The radiculo medullary artery that contributes to the ASA and continues as the artery of the filum terminale represents the arterial supply in AVFFTs.⁷

Vascular anatomy of the filum terminale is unique. A single artery, the artery of the filum terminale, arises from the termination of the anterior spinal axis, as a caudal continuation of the ventral spinal artery. This artery travels in front of the filum and rapidly diminishes in caliber, mostly nourishing the filum itself. It is argued that there are no radicular nerves originating from the filum; therefore, no other artery can supply the filum, including the sacral arteries. However, there was a report which illustrated two small arteries on a microscopic examination of a cross-section of the filum.⁸ In addition, there was an another report of the presence of a small artery supplying a coccygeal nerve, which is adherent to the filum in its proximal portion.⁹ These observations suggest that it is possible to fill the anterior spinal axis from the LSA, which is the arterial supply to the filum terminale externum. We consider these arterial supplies to the filum terminale externum leading to the dural supply in the AVFFT.

The goal of treatment of an AVFFT is the interruption of the arteriovenous shunt in order to eliminate the venous congestion of the spinal cord.^10–14 The endovascular approach has the advantage of being less invasive, and there have been some reports of endovascular treatment of AVFFT in recent years. However, the need to catheterize the ASA, especially the Adamkiewicz artery may be associated with a risk of ischemic injury of the spinal cord and conus medullaris. In addition, in the case of recurrent AVFFT after the feeder occlusion only for the initial endovascular therapy, there could be new complicated shunts or feeders found on the next DSA. However, we assume there would be the possibility that an accessible feeder can be found in the case of AVFFT, since the shunt point is in the more caudal portion than other types of spinal dural arteriovenous fistulae. Severe atherosclerosis of the aorta, small caliber of the ASA, and the long distance between the origin of the radiculo meningeal artery and the site of the fistula are arguments in favor of a surgical excision. Even in the case of AVFFT, if there was a single feeder on DSA, after the feeder occlusion for the initial endovascular therapy, the recurrence of AVFFT should be observed. On the other hand, surgical shunt obliteration will be the most curative procedure. There has been a recent report of a surgical procedure using indocyanine green (ICG) for spinal dural arteriovenous fistula.¹⁴ However, there are cases of difficulty in performing general anesthesia or the surgical procedure as with our case. In our case, the accessible feeder at the first DSA was very long at 40 cm and the surgical shunt obliteration was an adequate option for this AVFFT; however the patient was of an advanced age, and was in favor of the treatment under local anesthesia. Therefore, we chose TAE of the feeder from ASA as the first treatment in our case, even if this treatment was disrupted by feeder occlusion, there would still be the possibility that an accessible feeder could be found in the case of AVFFT (due to the shunt point being located in the more caudal portion than other types of spinal dural arteriovenous fistulae).

We consider the heated diluted contrast would be well tolerated, according to a paper on long feeder embolization¹⁵. This paper reported that the low concentration of NBCA induced a tissue response similar to that of the high-concentration form. Embolized vessels exhibited no greater incidence of recanalization. Therefore, embolization of arteriovenous shunts with diluted NBCA is well tolerated. In addition, heated diluted contrast of NBCA has low viscosity which advances more than non-heated diluted contrast. The adhesiveness of NBCA to microcatheters tends to be low in diluted contrast NBCA. In addition, warming histoacryl and lipiodol mixtures to a temperature that is close to 40℃ decreases the mixture's viscosity significantly and makes the injection easier to manage.¹⁶

There are some reports about the endovascular treatment for AVFFT.^{12–14,17–22} In an evaluation of the recent literature, we found 10 cases published in detail, summarized in Table 1. Arterial feeders of the shunts of AVFFT reported were mainly ASA, which go upwards to the top and turn to the shunt point of the fistulae. Distances from the top of arterial feeder to the location of fistulae should be considered for the TAE. In our patient, distance from the guiding catheter at the orifice of radicular artery to the shunt point was over 40 cm when we performed the first TAE. This distance was the longest case in Table 1. A 1.5 Fr Magic MP was longer than other types of Magic catheter, however the tip of the Magic catheter was best introduced at the L1 level. The first TAE ended as a feeder occlusion. The risk of NBCA embolization in ASA especially when over a long distance is also the high risk of tearing/dissecting the ASA especially at the ‘hair-pin’ curve level when withdrawing the system and in particular when there is minimal NBCA reflux around the tip of the microcatheter. For avoidance of the risk, we selected the longest flexible tip catheter available in Japan (Magic 1.5 Fr MP). The access region of the microcatheter, distal to the hair-pin curve of the ASA, was the flexible tip area of the Magic microcatheter, so we considered the risk of arterial injury by the microcatheter was low. In addition, we performed the arterial infusion of NBCA very slowly, and applied a negative pressure of the microcatheter during system removal, carefully confirming the live image during all the procedure.

Table 1.

Summary of angiographic findings in case series of endovascular treatment for AVFFT.

Reference	Sex	Age (years)	Arterial Feeder	Top of arterial feeder	Location of fistula
Chanthanaphak and colleagues20	F	70	ASA from left Th12	Th10-11	L5
	M	55	ASA from left Th10	Th9	L4
	F	39	ASA from left Th11	Th10	L2–3
	M	31	ASA from right Th10	Th9-10	L2–3
	M	67	ASA from right Th9	Th8-9	L3
	M	72	ASA from left L1	Th9-10	L5
Lim and colleagues10	F	53	ASA from right L1	Th10-11	L4–5
	F	63	ASA from right L4, left LSA	Th10	L3–4
Macht and colleagues15	M	57	SRA from left S2	L5	S3–4
Our case	M	78	ASA from right Th 9, left LSA	Th8	S1–2

ASA: Anterior spinal artery; AVFFT: Arteriovenous fistula of the filum terminale; F: female; LSA: lateral sacral artery; M: male; SRA: sacral radicular artery

A very important factor for the successful TAE, that will enable super-selective distal catheterization of the feeding vessel, is for the guiding catheter to be introduced in a more stable and distal position to the shunt point. In the case of recurrent AVFFT after the feeder occlusion and only for the initial endovascular therapy, there could be new shunts or feeders found on the next DSA. In this case, a more accessible feeder to the shunt point could be found during the relatively short period after the feeder occlusion. There have been no reports about the adequate interval between feeder occlusion and TAE after feeder occlusion; however, there was a report about the direct shunt obliteration after the endovascular feeder occlusion.²¹ A precise description of the challenges and traps included in such endovascular treatment has been discussed.²² We consider DSA should be performed at the early stage after a feeder occlusion. In our case, DSA at the recurrence of AVFFT showed, fortunately, a thick feeder (branch of LSA) to the fistula, which could be a stable access route for the Magic catheter for the TAE. There was a possibility that the branch of the LSA to the shunt was thicker than shown, for the result of the first feeder occlusion of TAE.

In addition, from a point of view of risk of TAE, the filum terminale does not carry any neurological function, closure of the artery of the filum does not result in neurological complications⁴, and the procedure thus should be well tolerated unless there is reflux of the embolic material upstream to the ASA system. The filum terminale is normally supplied by the artery of the filum terminale, which arises from the termination of the ASA at the inferior aspect of the conus medullaris, that is, it represents the continuation of the ASA.⁴ The vein of the filum terminale travels ventral to the filum terminale (but dorsal to the artery) and runs cephalad to join the anterior spinal venous system. When there is an abnormal communication between the artery and the vein of the filum terminale, an AVF, both vessels are enlarged. The fistula will then cause an increase in intravascular pressure in the veins around the spinal cord, resulting in a disturbance of spinal cord drainage. Consequently, the spinal cord becomes congested, and venous ischemia and infarction will occur if no treatment is provided.^1,3 In our case, the artery of the filum terminale, which arises from the termination of the ASA, was very long into the shunt of the fistula, which caused an increase of intravascular pressure in the veins around the spinal cord, resulting in a disturbance of spinal cord drainage.

Conclusion

In our case, the long distance between the origin of the radiculo meningeal artery and the site of the fistula resulted in a feeder occlusion in the first TAE. In the case of recurrent AVFFT after feeder occlusion, only for the initial endovascular therapy, there could be new complicated shunts found on the next DSA. However, there is be the possibility that an accessible feeder could be found in the case of AVFFT, due to the shunt point being located in a more caudal portion than other types of spinal dural arteriovenous fistulae.

Declaration of conflicting interests

The authors declare no potential conflicts of interest with respect to the research, authorship, or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.