Spinal Intradural Arteriovenous Shunt: Angiographic Characteristics and Treatment Outcomes

Abstract

Introduction

Spinal intradural arteriovenous shunt (SIAVS) is a rare entity of vascular diseases, classified into spinal cord arteriovenous malformation (SCAVM), spinal cord arteriovenous fistula (SCAVF), filum terminale arteriovenous fistula (FTAVF), and radicular arteriovenous malformation (AVM). Clinical presentations commonly include progressive myelopathy due to spinal venous congestion and hemorrhagic events.

Objective

This study aimed to describe the clinical presentations, angiographic characteristics, and treatment outcomes.

Materials and Methods

This retrospective study was conducted between 2010 and 2023. The collected data included demographic characteristics, clinical presentations, radiological findings, treatment modalities, treatment outcomes, and complications. Functional outcomes were assessed pre- or posttreatment using the Modified Aminoff and Logue Disability Scale.

Results

A total of 36 patients (22 males) were included, with a mean age at diagnosis of 30.7 years. Seventeen patients were diagnosed with SCAVM, 15 with SCAVF, and 4 with FTAVF. Progressive myelopathy was significantly found in patients with the fistulous type (SCAVF and FTAVF; p  = 0.008), whereas hematomyelia was common in the nidus-type group (SCAVM; p  = 0.006). Eight patients had spinal arteriovenous metameric syndrome (SAMS), and radicular AVM was observed in five cases of SCAVM and SCAVF. Multiple intradural shunts were identified in seven patients. Endovascular treatment using glue embolization was performed in 26 patients. Four patients were treated with surgery alone, and four received conservative management. Two patients underwent combined surgery and glue embolization. Angiographic outcomes showed complete obliteration in 12 (33.3%) patients and partial obliteration in 20 (55.5%) patients. Significant functional outcome improvements were found in all groups of disease and overall patients. One SCAVM patient had progressive enlargement of the venous pouch, which required repeated embolization. No recurrent hemorrhages were observed throughout the follow-up period.

Conclusion

Transarterial glue embolization is an effective treatment for SIAVS, providing significant neurological improvement, even in cases of partial embolization. The potential risk of anterior spinal artery occlusion must be carefully evaluated. Surgical intervention remains a viable alternative.

Introduction

Spinal vascular shunts (SVSs) are rare lesions comprising arteriovenous fistula (AVF) and arteriovenous malformation (AVM). They account for 3 to 4% of intradural spinal lesions and 5 to 9% of central nervous system vascular malformations. ^{1 2 3} Various classification systems for SVSs have been proposed, ^{4 5} but the most widely accepted are based on their anatomical location and relationship to the spinal cord and surrounding spinal structures. These are categorized into spinal intradural arteriovenous shunt (SIAVS), spinal dural arteriovenous shunt, spinal epidural arteriovenous shunt, and parachordal arteriovenous fistula.

Regarding SIAVS, spinal cord arteriovenous malformation (SCAVM) refers to an SIAVS in which a nidus is embedded within the spinal cord, typically supplied by the anterior spinal artery (ASA) or posterior spinal artery (PSA). Spinal cord arteriovenous fistula (SCAVF), also known as perimedullary AVF or fistulous-type SCAVM, is a direct arteriovenous shunt (AVS) without an intervening nidus, usually located on the surface of the spinal cord. ^{6 7 8} Filum terminale arteriovenous fistula (FTAVF) is an intradural extramedullary AVF located along the filum terminale, supplied by the artery of the filum terminale. Radicular AVM is situated on the spinal nerve roots and is supplied by radicular arteries. ⁹

Clinical presentations of SCAVM and SCAVF vary and may include back pain, radicular pain, progressive myelopathy due to spinal venous congestion, bowel or bladder dysfunction, or acute hemorrhage events, such as hematomyelia, subarachnoid hemorrhage (SAH), or intraventricular hemorrhage (IVH). Early and accurate diagnosis is essential to prevent disease progression, reduce the risk of permanent neurological damage, and restore neurological function. ¹⁰ Treatment options for SIAVS include surgical disconnection, endovascular embolization, combined surgical-endovascular approaches, and conservative management. Due to the rarity of these diseases, few studies have comprehensively evaluated the clinical presentations, angiographic features, and treatment strategies of SIAVS. Therefore, this research study presents our single-center experience in the treatment of SIAVS, which includes SCAVM, SCAVF, FTAVF, and radicular AVM.

Materials and Methods

Patient Population

We retrospectively reviewed all patients diagnosed with SIAVS and treated at Siriraj Hospital between January 2010 and December 2023. This study was approved by the institutional ethics committee prior to data collection (certificate of approval no. Si 074/2024). Inclusion criteria encompassed all patients diagnosed with SIAVS, including SCAVM, SCAVF, FTAVF, and radicular AVM. Patients lost to follow-up, and those without posttreatment outcomes were excluded. A total of 36 patients met the inclusion criteria. Clinical and angiographic evaluations were conducted by neurosurgeons and interventional neuroradiologists.

Clinical Evaluation

Pretreatment and last follow-up clinical statuses were collected from medical records and evaluated using the Modified Aminoff and Logue Disability Scale (mALS), ¹¹ as shown in Table 1 . This scale assesses functional outcomes in three domains: gait, micturition, and defecation disturbances, with scores ranging from 0 to 11. Higher scores indicate worse neurological outcomes.

Table 1. Modified Aminoff and Logue Disability Scale ¹¹ .

Domain	Score	Description
Gait disturbance	0	Normal gait and activity
	1	Leg weakness or abnormal gait, no restricted activity
	2	Leg weakness or abnormal gait with restricted activity
	3	Requires a cane or similar support for walking
	4	Requires a walker or crutches for walking
	5	Unable to stand, confined to bed or wheelchair
Micturition disturbance	0	Normal
	1	Hesitance, urgency, or frequency
	2	Occasional urinary incontinence or retention
	3	Total urinary incontinence or retention
Defecation disturbance	0	Normal
	1	Slight constipation, react to laxation
	2	Occasional fecal incontinence or severe constipation
	3	Total fecal incontinence

Radiological Evaluation

Pretreatment spinal magnetic resonance imaging (MRI)/magnetic resonance angiography (MRA) was available for 33 (91.7%) patients, while one patient with SCAVM and two patients with SCAVF were initially diagnosed using computerized tomography (CT). MRI was assessed for signs of spinal cord edema and hemorrhage (including hematomyelia or SAH) as well as the presence of extradural AVS, such as vertebral AVMs or cutaneous AVMs. Spinal angiography was performed in all patients to confirm the diagnosis and to evaluate the angiographic features of each type of SIAVS. Angiographic features assessed included the number of AVS, AVS location, arterial feeders, pattern of venous drainage, presence or absence of nidus, evidence of angiographic spinal cord congestion, and angiographic risk features, such as intranidal aneurysms or venous pouches ( Figs. 1 2 3 4 ). Congestive myelopathy was diagnosed based on clinical symptoms in conjunction with MRI findings, such as spinal cord enlargement with focal or diffuse T2 hyperintensity and angiographic evidence of delayed venous washout.

Fig. 1.

Spinal angiography of SCAVM in anteroposterior view. ( A ) Digital subtraction spinal angiography and ( B ) 3-dimensional spinal angiography showing the anterior spinal artery (single arrowhead) as an arterial feeder, AVM nidus (arrow), and a large tortuous draining vein (double arrowheads).

Fig. 2.

Digital subtraction spinal angiography of SCAVF in anteroposterior view. ( A ) The enlarged anterior spinal artery (single arrowhead). ( B ) An arteriovenous fistula (arrow) is located at the surface of the spinal cord and is supplied by the anterior spinal artery (arrowhead). ( C ) Dilated draining veins (double arrowheads). ( D ) Subsequent draining into the dilated perimedullary veins (double arrows).

Fig. 3.

Digital subtraction spinal angiography of FTAVF in anteroposterior view. ( A ) The artery of the filum terminale (arrowhead), the most caudal continuity of the anterior spinal artery (asterisk), supplies an arteriovenous fistula (arrow) located at the L3 vertebral level. ( B ) Dilated vein of the filum terminale as the draining vein (double arrowheads). ( C ) Puff of dilated veins along the filum terminale (double arrows).

Fig. 4.

Angiographic findings in a patient with concomitant SCAVF and radicular AVM. ( A ) Digital subtraction spinal angiography in anteroposterior view showing SCAVF (arrowhead) supplied by the anterior spinal artery, and radicular AVM (arrow) supplied by the radicular artery. Three-dimensional spinal angiography co-registered with vertebral images in axial ( B ) and coronal ( C ) views showing SCAVF (arrowhead) and radicular AVM (arrow).

A follow-up MRI was performed on all patients to assess for improvement or recurrence. Follow-up spinal angiography was also performed, except when complete obliteration was confirmed on immediate posttreatment angiography, accompanied by clinical improvement and a reduction in abnormal signal intensity on follow-up MRI.

Treatment

SIAVs were managed using surgery, endovascular therapy, a combination of both, or conservative treatment. At our medical institute, the treatment strategy was determined based on the patient's neurological symptoms, the presence of spinal cord congestion, and angiographic risk features, such as intranidal aneurysms or venous pouches. The primary treatment goal was complete obliteration of the fistula (in cases of SCAVF and FTAVF) or the nidus (in cases of SCAVM) without complications. If it was unlikely to complete obliteration without complications, the secondary objectives were to reduce spinal cord congestion through flow reduction and to eliminate angiographic risk factors.

Endovascular therapy was the preferred first-line treatment at our institute ( Fig. 5 ). In all patients undergoing endovascular treatment, N-butyl cyanoacrylate (NBCA) was used as the sole embolic agent after superselective microcatheterization transarterially. Glue embolization using NBCA was performed only when the microcatheter could be safely advanced to the distal arterial feeder, sufficiently distant from the ASA axis. Embolization was considered successful when NBCA glue reached the proximal draining vein via the fistula or nidus. The embolization technique is demonstrated in Fig. 6 .

Fig. 5.

Endovascular treatment of SCAVM using glue embolization in a 22-year-old female patient presenting with sudden neck pain, left hemiparesis, and hemiparesthesia. ( A1–A3 ) Sagittal spinal MRI in T1W ( A1 ), T2W ( A2 ), and T1W images after gadolinium injection ( A3 ) showing hematomyelia (arrowhead) at the C4 vertebral level with spinal cord edema. ( B1, B2 ) Digital subtraction ( B1 ) and 3-dimensional vertebral angiography ( B2 ) in anteroposterior view, revealing the anterior spinal artery (arrowhead) supplying SCAVM nidus (arrow) with intranidal aneurysm (double arrowheads). ( C ) The AVM nidus (arrow) is also supplied by the ascending cervical artery (arrowhead). ( D ) Right vertebral angiography in anteroposterior view after targeted embolization showing complete obliteration of the intranidal aneurysm.

Fig. 6.

Targeted glue embolization of spinal cord arteriovenous malformation in an 18-year-old male presented with radicular pain in the right leg, progressive weakness of the right foot, and impaired perianal sensation for 5 months. ( A1, A2 ) Sagittal spinal MRI in T1W image ( A1 ) showing hematomyelia (arrowhead) at the T12 vertebral level, and T2W MRI ( A2 ) showing spinal cord edema and perimedullary flow void dilatation (arrowhead). ( B1, B2 ) Digital subtraction spinal angiography in anteroposterior view showing spinal cord arteriovenous malformation (SCAVM) at the T12 vertebral level. ( B1 ) The right posterior spinal artery (PSA; arrowhead) supplying the SCAVM arises from the right ninth intercostal artery. ( B2 ) The anterior spinal artery (ASA; arrowhead) and enlarged ASA (double arrows) arising from the left twelfth intercostal artery also supply the SCAVM; a venous pouch (arrow) is demonstrated. ( C1–C5 ) The first attempt at glue embolization. ( C1 ) The tip (arrow) of a 1.2 Fr microcatheter (MAGIC) is placed into the venous pouch via the ASA (arrowhead). ( C2 ) Superselective injection reveals the tip of the microcatheter (arrow) positioned in the venous pouch (arrowhead). ( C3 ) Embolization with 25% glue concentration (arrowhead). ( C4 ) Glue cast in the venous pouch (arrowhead). ( C5 ) Obliteration of the superior part of the venous pouch with preservation of the ASA after the embolization (arrowhead). ( D1–D5 ) The second attempt at glue embolization. ( D1 ) The tip (arrow) of a 1.2 Fr microcatheter (MAGIC) is placed into the inferior part of the venous pouch via the ASA (arrowhead). ( D2 ) Superselective injection reveals the tip of the microcatheter (arrow) positioned in the venous pouch (arrowhead). ( D3 ) Embolization with 25% glue concentration (arrowhead). ( D4 ) Old glue cast after the first embolization (arrowhead) and a new glue cast after the second embolization (double arrows). ( D5 ) Obliteration of the venous pouch with preservation of the ASA after the embolization (arrowhead), and residual AVM is also seen. ( E1, E2 ) Sagittal spinal MRI in T1W ( E1 ) and T2W ( E2 ) images at 1 year after the treatment, showing markedly decreased spinal cord edema and perimedullary flow void appearance, without recurrent hematomyelia; clinically, the patient had improvement of motor function and significantly reduced radicular pain.

Surgical intervention was reserved for cases in which the endovascular approach was not feasible or when there was an insufficient safety margin from the ASA axis to safely perform embolization ( Fig. 7 ). Conservative management was considered in patients for whom both embolization and surgery were deemed high-risk ( Fig. 8 ), particularly if symptoms were benign.

Fig. 7.

Preoperative ( A, B ) and postoperative ( C, D ) digital subtraction vertebral angiography in a 45-year-old male patient with C1 radicular AVM and SCAVM at the C1 vertebral level. The patient presented with a sudden onset of a severe headache. Cranial CT revealed a subarachnoid hemorrhage in the region of the craniocervical junction. ( A ) Preoperative left vertebral angiography showing left C1 radicular AVM (arrow). ( B ) Preoperative right vertebral angiography showing the anterior spinal artery (arrowhead) supplying the SCAVM nidus (arrow) with an intranidal aneurysm (double arrowheads). Because of the inadequate safety margin of embolization, surgical treatment was considered. The patient underwent suboccipital craniectomy, C1 laminectomy, and AVM resection. ( C ) Postoperative left vertebral angiography showing complete obliteration of the AVMs with preservation of the anterior spinal artery (arrowhead). ( D ) Postoperative right vertebral angiography showing complete obliteration of the SCAVM and intranidal aneurysm.

Fig. 8.

Radiographic findings of spinal arteriovenous metameric syndrome (SAMS) in a 26-year-old female patient presenting with sudden paraparesis and urinary retention. ( A1–A3 ) Sagittal spinal MRI in T1W ( A1 ), T2W ( A2 ), and T1W images after gadolinium injection ( A3 ) showing hematomyelia (arrow) at the T4 vertebral level with extensive spinal cord edema, T5 and T6 vertebral AVMs (arrowhead), and enhancement of epidural venous plexus (double arrowheads). ( B ) Digital subtraction spinal angiography in anteroposterior view showing two arteriovenous malformations, including left radicular AVM (arrow) and vertebral AVM (arrowhead) in the same metameric segment. This finding is defined as SAMS. These lesions are encountered in two consecutive (T5 and T6) segments. ( C1–C3 ) Three-dimensional spinal angiography co-registered with vertebral images in coronal ( C1 ), sagittal ( C2 ), and axial ( C3 ) views showing T5 and T6 vertebral AVMs (arrowhead) and left radicular AVMs (arrow) at the corresponding neural foramens. ( D ) Digital subtraction spinal angiography in anteroposterior view showing another SCAVF (arrowhead) at the T3 vertebral level supplied by the anterior spinal artery (arrow). Because of a great risk of complications in embolization and surgical treatment, the patient underwent conservative treatment. ( E ) Sagittal spinal MRI in T2W image at 6 months after the hemorrhage showed spontaneous resolution of the hematoma and spinal cord edema. Her neurological functions had fully recovered at 10 months (mALS = 0) following the event.

Outcome Assessment

The primary outcomes included both angiographic and functional results following treatment. Angiographic outcomes were evaluated immediately after treatment or during follow-up spinal angiography and classified into three categories: complete obliteration (no visible early venous drainage observed during the arterial phase of spinal angiography), partial obliteration (partial surgical or embolization involving parts of the nidus, arterial feeders, or targeted embolization to angiographic risk features), and unchanged. Functional outcomes were categorized according to mALS as follows: asymptomatic (mALS = 0), improved (decrease in mALS score), unchanged (no change in mALS score), or deteriorated (increase in mALS score). The secondary outcomes included clinical presentations, angiographic findings, and treatment-related complications.

Statistical Analysis

All data analyses were performed using IBM SPSS Statistics 29.0 (IBM Armonk, New York, United States: IBM Corp). Categorical data were presented as frequencies and percentages. Continuous data were expressed as mean, standard deviation (SD), or median with interquartile range (IQR).

Comparisons of independent categorical variables were performed using the chi-square test or Fisher's exact test. The independent sample t -test was used to compare independent numerical parameters with a normal distribution between two independent groups. The Mann–Whitney U test was employed to compare independent numerical variables without normal distributions between two independent groups. Comparison of dependent numerical data was performed using the Wilcoxon signed-rank test. A p -value of less than 0.05 was considered statistically significant.

Results

Demographic Characteristics and Clinical Features

A total of 36 patients were included in the study, including 22 males (61%) with a mean age at diagnosis of 30.7 ± 16.3 years. The median follow-up duration was 38.5 months (IQR: 15.3, 71.5). Among the 36 patients diagnosed with SIAVS, 17 (47.2%) had SCAVMs, 15 (41.7%) had SCAVFs, and 4 (11.1%) had FTAVFs.

In the comparison of variables between patients with SIAVS, all patients were classified into two groups: nidus-type SIAVS and fistulous-type SIAVS. The nidus-type group consisted of patients with SCAVM only, whereas SCAVF and FTAVF were included in the fistulous-type group ( Table 2 ). There was no statistical gender preference between the two groups ( p  = 0.791). The FTAVF group had the oldest mean age at diagnosis (mean: 46.2 ± 20.1 years), followed by the SCAVF (31.3 ± 15.5) and SCAVM (26.6 ± 14.8) groups, but there was no statistically significant difference in mean age between the nidus-type and fistulous groups ( p  = 0.157). The majority of patients in the fistulous-type group presented with progressive myelopathy (16/19, 84.2%) and bowel-bladder dysfunction (13/19, 68.4%). No statistically significant difference in the duration between the groups was observed ( p  = 0.194).

Table 2. Demographic characteristics, clinical presentation, and angiographic findings.

Variable	All patients	Spinal intradural arteriovenous shunt
		Nidus type (SCAVM)	Fistulous type (SCAVF and FTAVF)	p -Value
	n  = 36	n  = 17	n  = 19
Demographic characteristic
Male gender, n (%)	22 (61)	10 (58)	12 (63.2)	0.791
Age (y), mean ± SD	30.7 ± 16.3	26.6 ± 14.8	34.4 ± 16.7	0.157
Clinical presentation, n (%)
Progressive myelopathy	23 (63.8)	7 (41)	16 (84.2)	0.008 a
Bowel and bladder dysfunction	24 (66.7)	11 (64.7)	13 (68.4)	0.806
Hematomyelia	11 (30.5)	9 (53)	2 (10.5)	0.006 a
SAH or IVH	2 (5.5)	1 (5.8)	1 (5.3)	1.000
Duration between symptom onset and diagnosis (mo), median (IQR)	5 (0.6, 21)	1 (0.2, 24)	6 (2, 12)	0.194
Pretreatment mALS score, median (IQR)	7.5 (3, 10)	7 (1.5, 9.5)	8 (6, 10)	0.484
Angiographic features, n (%)
Multiple arterial feeders	20 (55.5)	11 (64.7)	9 (47.4)	0.296
Intranidal aneurysm	5 (13.9)	5 (29.4)	0 (0)	0.016 a
Venous pouch	9 (25)	8 (47)	1 (5.3)	0.006 a
Location of AVS, n (%)
Craniocervical junction (C0–C2)	2 (5.5)	2 (11.7)	0 (0)	0.216
Cervical level (C3–C6)	2 (5.5)	1 (5.9)	1 (5.3)	1.000
Cervicothoracic junction (C7–T2)	1 (2.8)	1 (5.9)	0 (0)	0.472
Thoracic level (T3–T11)	19 (52.8)	10 (58.8)	9 (47.4)	0.493
Thoracolumbar level (T12–L1)	1 (2.8)	1 (5.9)	0 (0)	0.472
Lumbar level	1 (2.8)	0 (0)	1 (5.3)	1.000
Conus medullaris	5 (13.9)	1 (5.9)	4 (21.1)	0.342
Cervical and conus medullaris	1 (2.8)	1 (5.9)	0 (0)	0.472
Filum terminale	4 (11.1)	0 (0)	4 (21.1)	0.106
Comparison of the location of AVS, n (%)
Thoracic spinal regions	20 (55.5)	11 (64.7)	9 (47.4)	0.296
Other spinal regions	16 (44.5)	6 (35.3)	10 (52.6)	0.296
Multiple intradural AVS, n (%)	7 (19.4)	4 (23.5)	3 (15.8)	0.684
Spinal arteriovenous metameric syndrome, n (%)	8 (22)	5 (29.4)	3 (15.8)	0.434
Follow-up (mo), median (IQR)	38.5 (15.3, 71.5)	47 (33.5, 93)	22 (11, 64)	0.090

Abbreviations: AVS, arteriovenous shunt; C, cervical; FTAVF, filum terminale arteriovenous fistula; IQR, interquartile range; IVH, intraventricular hemorrhage; L, lumbar; mALS, modified aminoff and logue disability scale; mo, month; n , number of patients; SAH, subarachnoid hemorrhage; SCAVF, spinal cord arteriovenous fistula; SCAVM, spinal cord arteriovenous malformation; SD, standard deviation; T, thoracic; y, year.

^a Indicates statistically significant difference level ( p  < 0.05).

Hemorrhagic presentations, such as hematomyelia, SAH, or intraventricular hemorrhage (IVH), were more common in the nidus-type group (10/17, 58.8%) than in the fistulous-type group (3/19, 15.8%). Conversely, progressive myelopathic symptoms were significantly found in the fistulous-type group (16/19, 84.2%) compared with the nidus-type group (7/17, 41%; p  = 0.008). With regard to hemorrhagic events, hematomyelia was significantly prevalent in the SCAVM group (53%; p  = 0.006), while SAH and IVH occurred at similar rates between both groups ( p  = 1.000). Prevalence of bowel and bladder dysfunction was not different between the groups of disease ( p  = 0.806). No asymptomatic patients were found in this study.

There was no statistically significant difference in pretreatment mALS scores between the two groups ( p  = 0.484). In particular, two patients, one with SCAVM and one with SCAVF, presented with severe headache or back pain secondary to SAH or IVH and had an initial mALS score of 0.

Radiological Features

The majority of patients in the nidus-type (11/17, 64.7%) and fistulous-type (9/19, 47.4%) groups tend to have the lesions in the thoracic spinal region compared with other regions, but there was no statistically significant difference between the two groups ( p  = 0.296). Multiple arterial feeders were more commonly encountered in the nidus-type group (11/17, 64.7%) compared with the counterpart (9/19, 47.4%). Nevertheless, this difference between the two groups did not achieve a statistically significant level ( p  = 0.296). Angiographic risk features were found in 13 of 17 SCAVMs (76.5%), including five intranidal aneurysms and eight venous pouches. Only one SCAVF case (6.7%) demonstrated a venous pouch. The prevalence of intranidal aneurysm was obviously higher in the nidus-type group (5/17, 29.4%) when compared with the fistulous-type group (0/19, 0%); the difference reached a statistically significant level ( p  = 0.016). In the same way, the angiographic presence of venous pouches was significantly dominant in the nidus-type group (8/17, 47%) compared with the fistulous-type group (1/19, 5.3%; p  = 0.006).

Additionally, eight patients (22%) were diagnosed with SAMS. They consisted of five with SCAVM, three with SCAVF, and none with FTAVF ( p  = 0.434). Prevalence of multiple SIAVSs was not significantly different between the two groups of disease ( p  = 0.684). Also, there was no statistically significant difference in follow-up duration between the groups ( p  = 0.084). The demographic data, clinical presentations, and radiographic findings are summarized in Table 2 .

Multiple SIAVSs were identified in seven patients (19.4%), as summarized in Table 3 . These included five cases of SCAVM or SCAVF associated with radicular AVM, one case involving both cervical and conus medullaris AVM, and one case with coexisting SCAVM and SCAVF.

Table 3. Multiple spinal intradural arteriovenous shunts in seven patients.

Location of AVS	n
Multiple SCAVMs (cervical cord and conus medullaris AVM)	1
SCAVM and SCAVF	1
SCAVM and radicular AVM	2
SCAVF and radicular AVM	3

Abbreviations: AVM, arteriovenous malformation; AVS, arteriovenous shunt; n , number of patients; SCAVF, spinal cord arteriovenous fistula; SCAVM, spinal cord arteriovenous malformation.

Treatment Outcomes

Treatment modalities and outcomes, including angiographic and functional results, are summarized in Table 4 . Among the 17 SCAVM patients in the nidus-type group, 13 underwent embolization, two were managed conservatively, one underwent surgical resection, and one received combined surgery and embolization. Of the 15 SCAVF patients in the fistulous-type group, 11 underwent embolization, two underwent surgery, and two were treated with conservative management. Among the four patients with FTAVF in the fistulous-type group, two underwent embolization, one underwent surgery, and one required a combination of treatment (surgery followed by embolization) due to unsuccessful superselective catheterization during the first treatment attempt. There was no significant difference in treatment modalities between the groups of disease.

Table 4. Treatment modalities, angiographic and functional outcomes.

Variables	All patients	Spinal intradural arteriovenous shunt
		Nidus type (SCAVM)	Fistulous type (SCAVF and FTAVF)	p -Value
	n  = 36	n  = 17	n  = 19
Treatment modality, n (%)
Conservative treatment	4 (11.1)	2 (11.7)	2 (10.5)	1.000
Endovascular treatment	26 (72.2)	13 (76.5)	13 (68.4)	0.717
Surgical treatment	4 (11.1)	1 (5.9)	3 (15.8)	0.605
Combined surgery and endovascular treatment	2 (5.6)	1 (5.9)	1 (5.3)	1.000
Angiographic outcome, n (%)
Complete obliteration	12 (33.3)	3 (17.6)	9 (47.4)	0.059
Partial obliteration	20 (55.5)	12 (70.5)	8 (42.1)	0.086
Unchanged	4 (11.1)	2 (11.7)	2 (10.5)	1.000
Functional outcome, n (%)
Completely recovered	10 (27.7)	5 (29.4)	5 (26.3)	1.000
Improved	20 (55.5)	9 (53)	11 (57.9)	0.764
Unchanged	5 (13.9)	2 (11.8)	3 (15.8)	1.000
Deteriorated	1 (2.7)	1 (5.8)	0 (0)	0.472
mALS score, median (IQR)
Pretreatment	7.5 (3, 10)	7 (1.5, 9.5)	8 (6, 10)	0.484
Posttreatment	2 (0, 6)	3 (0, 6.5)	2 (0, 6)	0.889
Change in mALS score	−2 (−6.5, 0)	−2 (−5.5, 0)	−2 (−7, −1)	0.435
Comparison between pretreatment and posttreatment mALS score in each group
p -Value	<0.001 a	0.015 a	<0.001 a

Abbreviations: FTAVF, filum terminale arteriovenous fistula; IQR, interquartile range; mALS, Modified Aminoff and Logue Disability Scale; n , number of patients; SCAVF, spinal cord arteriovenous fistula; SCAVM, spinal cord arteriovenous malformation.

^a Indicates statistically significant difference level ( p  < 0.05).

Four patients were managed conservatively; they included two with SCAVM and two with SCAVF. None with FTAVF underwent conservative treatment. The first individual with SCAVM presented with pain and hematomyelia without neurological deficit (mALS score = 0). The spinal angiography revealed that the arterial feeder was exclusively the ASA, and a venous pouch was identified as an angiographic risk. The attempt at first embolization failed due to unsuccessful catheterization, and the patient refused a second attempt because of spontaneous improvement in pain. During an 8-month follow-up period, complete resolution of pain without neurological deficit was achieved. The second SCAVM patient managed conservatively was diagnosed as Parkes Weber syndrome. She presented with hematomyelia in the cervical spinal cord, and spinal angiography revealed SCAVM at the cervical and conus medullaris levels. Neither angiographic risk nor spinal cord congestion was observed at both levels. A single branch from the ASA was identified as the arterial feeder supplying the cervical SCAVM. Conservative management was chosen due to a lack of an adequate safety margin for embolization via the ASA. Throughout the 42-month follow-up period, partial improvement of neurological symptoms was observed (reduced four scores of mALS).

The remaining two with SCAVF, who were managed conservatively, were diagnosed as SAMS. Conservative treatment was given because each lesion had a solitary arterial supply from the ASA, no angiographic risk, no spinal cord congestion, and a scant safety margin for embolization. One experienced partial neurological improvement (reduced four scores of mALS) at 14 months of posttreatment follow-up, while the other achieved complete resolution of neurological symptoms (reduced nine scores of mALS) at 10 months of follow-up.

Complete angiographic obliteration was greater in the fistulous-type group (9/19, 47.4%), compared with the nidus-type group (3/17, 17.6%). However, there was no statistically significant difference in the rate of complete angiographic obliteration between the two groups ( p  = 0.059). Partial obliteration was not significantly noted in any groups ( p  = 0.086).

Regarding functional outcomes at the last follow-up, five out of 17 (29.4%) patients with the nidus-type lesions and five out of 19 (26.3%) patients with the fistulous-type lesions had complete recovery of neurological function (posttreatment mALS score = 0) and no pain. Notably, four of these patients (three SCAVM, one SCAVF) initially presented with pain but no neurological deficits (pretreatment mALS score = 0). Most patients showed neurological improvement after treatment: 9 out of 17 (53%) SCAVM patients, eight out of 15 (53.3%) SCAVF patients, and three out of four (75%) FTAVF patients. However, there was no significant difference in functional outcome between the groups of disease.

Median posttreatment mALS scores showed clinical improvement (score reduction) across all groups: the fistulous-type group (from a pretreatment score of 8 to a posttreatment score of 2) and the nidus-type group (7–3). One SCAVM patient deteriorated due to progressive spinal cord congestion and enlargement of the venous pouch, which required the second embolization, resulting in a 6-point increase in the mALS score at final follow-up.

In an analysis of the mALS score, no statistically significant differences were found in the comparison of the pretreatment score ( p  = 0.484), posttreatment score ( p  = 0.889), and change in the score ( p  = 0.435) between both groups. Nevertheless, when the posttreatment mALS score was compared with the pretreatment score in an individual group, we found a statistically significant reduction of mALS score (neurological function improvement) in the nidus-type ( p  = 0.015) and fistulous-type ( p  < 0.001) groups, as well as in all patients ( p  < 0.001).

Functional outcomes stratified by treatment modality are summarized in Table 5 . In the conservative management group, spontaneous neurological improvement was observed in two out of four (50%) patients. Complete symptom resolution occurred in two out of four (50%) patients: one initially presented with back pain but no neurological deficit (pretreatment mALS = 0), and the other with hematomyelia and bowel-bladder dysfunction (pretreatment mALS = 9). In the endovascular treatment group, neurological improvement was achieved in 61.5% of patients (16 out of 26), while 15.4% (four out of 26) had no residual symptoms at follow-up. Notably, most of them (69.2%, 18 out of 26) underwent partial embolization. All patients who underwent complete surgical resection experienced complete symptom resolution. One patient who underwent partial surgical resection showed improvement in neurological symptoms at the last follow-up. Importantly, no cases of recurrent hemorrhage were reported throughout the follow-up period.

Table 5. Functional outcomes stratified by treatment modalities.

Treatment modality	n	Functional outcome
		Completely recovered (%)	Improved (%)	Unchanged (%)	Deteriorated (%)
Conservative treatment	4	2 (50)	2 (50)	0 (0)	0 (0)
Endovascular treatment	26	4 (15.4)	16 (61.5)	5 (19.2)	1 (3.9)
Partial embolization	18	3 (16.7)	11 (61.1)	3 (16.7)	1 (5.5)
Complete embolization	8	1 (12.5)	5 (62.5)	2 (25)	0 (0)
Surgery	4	3 (75)	1 (25)	0 (0)	0 (0)
Partial resection	1	0 (0)	1 (100)	0 (0)	0 (0)
Complete resection	3	3 (100)	0 (0)	0 (0)	0 (0)
Combined surgery and endovascular treatment	2	1 (50)	1 (50)	0 (0)	0 (0)

In a comparison of functional outcomes between treatment modalities ( Table 6 ), we did not encounter statistically significant differences in functional outcome between the treatment groups ( p  = 0.807).

Table 6. Comparison of functional outcomes between treatment modalities.

Functional outcome	Treatment modality
	Conservative treatment	Endovascular treatment	Surgery	Combined surgery and endovascular treatment	p -Value
	n  = 4	n  = 26	n  = 4	n  = 2	0.807
Completely recovered (%)	2 (50)	4 (15.4)	3 (75)	1 (50)
Improved (%)	2 (50)	16 (61.5)	1 (25)	1 (50)
Unchanged (%)	0 (0)	5 (19.2)	0 (0)	0 (0)
Deteriorated (%)	0 (0)	1 (3.9)	0 (0)	0 (0)

Abbreviation: n , number of patients.

Complications

One patient with a history of traumatic complete spinal cord injury, who was diagnosed with SCAVF and presented with progressive paraparesthesia, experienced glue reflux into the ASA during embolization. However, no worsened neurological function was observed after the procedure. Early postprocedural neurological worsening was noted in two patients due to progressive venous thrombosis. However, both patients showed clinical improvement following intravenous heparinization for the treatment of venous thrombosis.

Discussion

SVSs are rare lesions comprising AVF and AVM. ^{1 2 3} Despite no consensus about a classification system, ^{2 4 5} the most widely accepted are based on their anatomical location and relationship to the spinal cord and surrounding spinal structures. These are categorized into SIAVS, spinal dural arteriovenous shunt, spinal epidural arteriovenous shunt, and parachordal arteriovenous fistula. SIAVS is classified into SCAVM, SCAVF, FTAVF, and radicular AVM. Clinical presentations of SCAVM and SCAVF vary and may include back pain, radicular pain, progressive myelopathy due to spinal venous congestion, bowel or bladder dysfunction, or acute hemorrhage events, such as hematomyelia, SAH, or IVH. Incidental findings may also be identified in cases of radicular AVM, which can coexist with SCAVM as part of SAMS.

In our study, age at onset varies according to the type of SIAVS. Patients with the nidus-type lesions tended to be younger (mean: 26.6 years), and those with the fistulous-type lesions had an older age (mean: 34.4 years); however, this difference did not reach a statistically significant level ( p  = 0.157). Patients with the nidus-type lesions more frequently presented with hemorrhagic events, particularly hematomyelia ( p  = 0.006). In contrast, patients with fistulous-type lesions more often presented with progressive myelopathy ( p  = 0.008), resulting in a longer delay between symptom onset and diagnosis compared with patients with the nidus-type lesions. These findings indicated that SCAVF and FTAVF in the fistulous-type group had a gradually progressive time course of the disease caused by congestive myelopathy, whereas a rapid progressive onset due to hemorrhagic manifestation was commonly found in patients with SCAVM in the nidus-type group. We did not find differences in the location of AVS and the severity of neurological deficits evaluated by mALS between the groups.

In terms of treatment outcome, the rate of complete obliteration tended to be higher in the fistulous-type group compared with the counterpart, but there was no statistically significant difference ( p  = 0.059). This result indicated that characteristics of vascular lesions in SCAVM were more complex and treatment was more difficult compared with patients with arteriovenous fistular lesions. Although a slight difference in the rate of complete obliteration was found between the groups, there was no difference in functional outcome assessed by mALS.

In our study, all radicular AVMs were discovered incidentally in association with SCAVM or SCAVF. Conservative treatment is recommended in the absence of perimedullary venous reflux. Focusing on four patients with FTAVF in our series, all of them were male, with a mean age of 46.2 years. All presented with progressive myelopathy and bowel/bladder dysfunction due to spinal cord congestion. The diagnostic delay was notable in this group (median duration: 9.5 months), relatively longer than that of SCAVM and SCAVF patients. Importantly, two of four (50%) patients had previously been misdiagnosed with degenerative spinal disease and had undergone spinal surgery without symptomatic relief. With early recognition and appropriate treatment, complete angiographic obliteration and neurological improvement, including recovery of bowel and bladder function, were achieved in all FTAVF patients.

The primary goal of treatment is to achieve complete obliteration of the AVS without inducing further neurological deficits. When complete obliteration is not feasible, partial obliteration, either through embolization or surgical resection, can still offer clinical benefit by reducing spinal cord congestion and eliminating angiographic risk features, such as intranidal aneurysms or venous pouches. The option of treatment is guided by initial clinical presentation and angiographic characteristics. At our medical institute, transarterial glue embolization is the preferred first-line approach, with strict precautions to prevent reflux of embolic material into the ASA axis, as previously reported. ^{7 8 10} The goal of embolization is the obliteration of fistulous connections as well as the proximal draining vein. Partial embolization of angiographic risk features, such as intranidal aneurysm or large venous pouch, may be acceptable in selected patients. This approach is particularly relevant when the vascular supply of the malformation is associated with the ASA. In those cases, the risk of spinal cord ischemia or worse neurological function with curative embolization may be prohibitive. ¹⁰ Surgical intervention is considered when the safety margin of embolization is inadequate or distal catheterization is not possible. ^{8 10} Such situations included feeding arteries that were too small in caliber or too tortuous to advance the microcatheter into the fistulas, and a reflux distance that was too short to prevent embolism of glue. ⁸ Conservative treatment is reserved for cases where treatment risks by embolization or surgery are high, symptoms are mild, and no angiographic risk features or spinal cord congestion are present. Notably, none of our patients in the conservative group experienced new neurological deficits, neurological deterioration, or recurrent hemorrhage after the treatment.

In evaluating functional outcomes (posttreatment mALS), we found no difference in posttreatment functional outcomes between the nidus and fistulous types of the disease. However, when comparing the functional scores between pre- and posttreatment in each group, we found significant improvement of mALS after treatment in all groups of disease and all patients. These results emphasized the usefulness of the treatment of AVS for neurological function improvement.

In comparison with previous studies, our findings were comparable to those of Lee et al ⁷ and Cho et al ⁸ in terms of clinical presentations, angiographic features of SCAVM and SCAVF, and the higher rate of complete angiographic obliteration in SCAVFs. However, functional outcomes differed. In our cohort, the rate of neurological improvement, including patients who were completely asymptomatic at last follow-up, was 82.4% for SCAVMs and 79.9% for SCAVFs, compared with 31 and 44%, respectively, from the study of Lee and coworkers. ⁷ Additionally, the clinical worsening rate in our study was lower (5.8% vs. 23% ⁷ ). We propose two main reasons for these differences. First, the shorter follow-up period in our study (median: 2.5 years vs. 9.6 years in the previous study ⁷ ) may limit the capture of long-term outcomes in some patients. The second reason is that differences in functional outcome assessment in the previous study only evaluated motor disability using mALS, whereas our study analyzed all aspects of mALS. As such, improvements in defecation or micturition were excluded in the previous study. One patient in our study experienced progression of neurological symptoms due to an increase in the size of the venous pouch and an increase in spinal cord edema on MRI, which required a second embolization. However, at the last follow-up, the functional outcome was the same as after the first embolization. No patients in our study experienced rebleeding after presentation and treatment, whereas the study by Lee and coworkers ⁷ reported a rebleeding rate of 4.8% per year in untreated lesions and 2.9% per year after partial treatment. The shorter follow-up period in our study may limit assessment of the long-term rebleeding rate. Nevertheless, in terms of functional outcome after partial embolization or partial surgical resection, our study and previous studies ^{2 8 10} demonstrated improvement in both SCAVM and SCAVF groups.

Limitations

This study has some limitations, including its retrospective design, small sample size, and lack of long-term follow-up in some patients. Importantly, there were no delayed posttreatment spinal angiographies in all patients; therefore, incompletely cured lesions could be missed. Additionally, the complete obliteration for our study was relatively lower than that found in other studies. This may reflect the technical surgical expertise. Nevertheless, we believe that treatment decisions should be individualized. Therapeutic options should be tailored to each patient's clinical presentation and angiographic characteristics.

Conclusion

Early and accurate recognition of SIAVS is crucial to minimizing permanent spinal cord damage. Treatment strategies should be tailored according to the initial neurological status and angiographic findings. While complete obliteration of the AVS through embolization or surgical resection remains the primary treatment goal, partial obliteration can still be effective in reducing spinal cord congestion and eliminating angiographic risks when complete treatment poses excessive procedural risk. In select cases, conservative management may also be considered.