Neurosurgical versus endovascular treatment of craniocervical junction arteriovenous fistulas: a multicenter cohort study of 97 patients

Keisuke Takai MD, PhD1, Toshiki Endo MD, PhD2,3, Toshitaka Seki MD, PhD4, Tomoo Inoue MD, PhD2,3, Izumi Koyanagi MD, PhD5, Takafumi Mitsuhara MD, PhD6, and the Neurospinal Society of Japan CCJ AVF Study Investigators
View More View Less
  • 1 Department of Neurosurgery, Tokyo Metropolitan Neurological Hospital, Tokyo, Japan;
  • | 2 Department of Neurosurgery, Tohoku University, Graduate School of Medicine, Sendai, Miyagi, Japan;
  • | 3 Department of Neurosurgery, Kohnan Hospital, Sendai, Miyagi, Japan;
  • | 4 Department of Neurosurgery, Hokkaido University, Graduate School of Medicine, Sapporo, Hokkaido, Japan;
  • | 5 Department of Neurosurgery, Hokkaido Neurosurgical Memorial Hospital, Sapporo, Hokkaido, Japan; and
  • | 6 Department of Neurosurgery, Hiroshima University Hospital, Hiroshima, Japan
Open access

OBJECTIVE

Craniocervical junction (CCJ) arteriovenous fistulas (AVFs) are treated using neurosurgical or endovascular options; however, there is still no consensus on the safest and most effective treatment. The present study compared the treatment results of neurosurgical and endovascular procedures for CCJ AVFs, specifically regarding retreatment, complications, and outcomes.

METHODS

This was a multicenter cohort study authorized by the Neurospinal Society of Japan. Data on consecutive patients with CCJ AVFs who underwent neurosurgical or endovascular treatment between 2009 and 2019 at 29 centers were analyzed. The primary endpoint was the retreatment rate by procedure. Secondary endpoints were the overall complication rate, the ischemic complication rate, the mortality rate, posttreatment changes in the neurological status, independent risk factors for retreatment, and poor outcomes.

RESULTS

Ninety-seven patients underwent neurosurgical (78 patients) or endovascular (19 patients) treatment. Retreatment rates were 2.6% (2/78 patients) in the neurosurgery group and 63% (12/19 patients) in the endovascular group (p < 0.001). Overall complication rates were 22% and 42% in the neurosurgery and endovascular groups, respectively (p = 0.084). Ischemic complication rates were 7.7% and 26% in the neurosurgery and endovascular groups, respectively (p = 0.037). Ischemic complications included 8 spinal infarctions, 2 brainstem infarctions, and 1 cerebellar infarction, which resulted in permanent neurological deficits. Mortality rates were 2.6% and 0% in the neurosurgery and endovascular groups, respectively (p > 0.99). Two patients died of systemic complications. The percentages of patients with improved modified Rankin Scale (mRS) scores were 60% and 37% in the neurosurgery and endovascular groups, respectively, with a median follow-up of 23 months (p = 0.043). Multivariate analysis identified endovascular treatment as an independent risk factor associated with retreatment (OR 54, 95% CI 9.9–300; p < 0.001). Independent risk factors associated with poor outcomes (a postoperative mRS score of 3 or greater) were a pretreatment mRS score of 3 or greater (OR 13, 95% CI 2.7–62; p = 0.001) and complications (OR 5.8; 95% CI 1.3–26; p = 0.020).

CONCLUSIONS

Neurosurgical treatment was more effective and safer than endovascular treatment for patients with CCJ AVFs because of lower retreatment and ischemic complication rates and better outcomes.

ABBREVIATIONS

AVF = arteriovenous fistula; CCJ = craniocervical junction; mRS = modified Rankin Scale; SAH = subarachnoid hemorrhage.

OBJECTIVE

Craniocervical junction (CCJ) arteriovenous fistulas (AVFs) are treated using neurosurgical or endovascular options; however, there is still no consensus on the safest and most effective treatment. The present study compared the treatment results of neurosurgical and endovascular procedures for CCJ AVFs, specifically regarding retreatment, complications, and outcomes.

METHODS

This was a multicenter cohort study authorized by the Neurospinal Society of Japan. Data on consecutive patients with CCJ AVFs who underwent neurosurgical or endovascular treatment between 2009 and 2019 at 29 centers were analyzed. The primary endpoint was the retreatment rate by procedure. Secondary endpoints were the overall complication rate, the ischemic complication rate, the mortality rate, posttreatment changes in the neurological status, independent risk factors for retreatment, and poor outcomes.

RESULTS

Ninety-seven patients underwent neurosurgical (78 patients) or endovascular (19 patients) treatment. Retreatment rates were 2.6% (2/78 patients) in the neurosurgery group and 63% (12/19 patients) in the endovascular group (p < 0.001). Overall complication rates were 22% and 42% in the neurosurgery and endovascular groups, respectively (p = 0.084). Ischemic complication rates were 7.7% and 26% in the neurosurgery and endovascular groups, respectively (p = 0.037). Ischemic complications included 8 spinal infarctions, 2 brainstem infarctions, and 1 cerebellar infarction, which resulted in permanent neurological deficits. Mortality rates were 2.6% and 0% in the neurosurgery and endovascular groups, respectively (p > 0.99). Two patients died of systemic complications. The percentages of patients with improved modified Rankin Scale (mRS) scores were 60% and 37% in the neurosurgery and endovascular groups, respectively, with a median follow-up of 23 months (p = 0.043). Multivariate analysis identified endovascular treatment as an independent risk factor associated with retreatment (OR 54, 95% CI 9.9–300; p < 0.001). Independent risk factors associated with poor outcomes (a postoperative mRS score of 3 or greater) were a pretreatment mRS score of 3 or greater (OR 13, 95% CI 2.7–62; p = 0.001) and complications (OR 5.8; 95% CI 1.3–26; p = 0.020).

CONCLUSIONS

Neurosurgical treatment was more effective and safer than endovascular treatment for patients with CCJ AVFs because of lower retreatment and ischemic complication rates and better outcomes.

Craniocervical junction (CCJ) arteriovenous fistulas (AVFs) are rare vascular malformations of the central nervous system. Their diagnosis and treatment are challenging because they have a more complex angioarchitecture than thoracolumbar spinal AVFs.13 Advances in neuroimaging have facilitated the identification of narrower vessels, and studies on neurosurgical and endovascular treatments have been increasing in the past 2 decades.46 However, a standard treatment has not yet been established because of limited experience with small groups of patients. The present study was conducted on a large multicenter cohort of patients with CCJ AVFs to establish whether neurosurgical or endovascular treatment is more effective and safer, specifically regarding retreatment, complications, and outcomes.

Methods

Study Design and Ethics

The present study was a multicenter cohort study authorized by the Neurospinal Society of Japan. This study protocol was approved by the institutional review board at the Tokyo Metropolitan Neurological Hospital and participating centers. Since this was a noninvasive study, the requirement for written informed consent from patients was waived. A public notice that provided information on this study was instead given on the websites of individual centers.

Patient Selection

Inclusion criteria were consecutive patients with CCJ AVFs defined as dural, intradural, and epidural AVFs at the C1 or C2 level treated neurosurgically or endovascularly at 29 centers between 2009 and 2019. Exclusion criteria were patients with intracranial AVFs, spinal AVFs below the C3 level, iatrogenic AVFs, traumatic AVFs, AVFs in the pediatric population (age < 20 years), CCJ AVFs treated with a combination of neurosurgical and endovascular procedures, and CCJ AVFs treated conservatively. All patients underwent angiography, and the decision to perform surgery or endovascular treatment was made at individual centers. There were no standard indications for the selection of CCJ AVFs for neurosurgical or endovascular treatment because this was a retrospective study. Surgery involved the microsurgical obliteration of the fistula through suboccipital craniotomy and C1 laminectomy by neurosurgeons. Endovascular treatment consisted of embolization of the fistula with n-butyl cyanoacrylate and/or coils by endovascular neurointerventionalists.

Angiographic Characteristics

Angiographic data were reviewed by neuroradiologists at each center who were blinded to the purpose of the present study. Data were rereviewed and differential diagnoses were dural, intradural, or epidural AVFs by 4 board members (K.T., T.E., T.S., and T.I.) according to a previous angiographic study by Hiramatsu et al.1 Reviewers were certified by the Neurospinal Society of Japan and had > 18 years of experience in spinal surgery.

Primary and Secondary Endpoints

The primary endpoint was a comparison of the retreatment rate after initial treatment between the neurosurgery and endovascular groups. Secondary endpoints were comparisons between the 2 groups of 1) the overall complication rate, 2) the ischemic complication rate, 3) the mortality rate, 4) posttreatment changes in modified Rankin Scale (mRS) scores, 5) independent risk factors associated with retreatment, 6) independent risk factors associated with poor outcomes, 7) details of retreatments, 8) details of angiographic occlusion, and 9) details of complications.

Definition of Retreatment, Complications, Good Outcomes, and Complete Obliteration

A patient with a residual fistula that required additional neurosurgical or endovascular treatment after the initial treatment was defined as a case of retreatment. A patient with a complication that required additional treatment was not included as a case of retreatment. Only the initial treatment was registered for analyses; patients in whom more than one treatment was performed were regarded as a single case. Complications only included major or permanent adverse events that required additional invasive treatments or resulted in permanent morbidity or death within 30 days of the initial treatment as follows: ischemic complications (ischemic infarction in the cerebral, cerebellum, brainstem, or spinal cord), hemorrhagic complications (intraoperative vascular injury and postoperative hematoma), CSF leakage, hydrocephalus, new permanent neurological deficits, symptomatic vasospasm, and medical complications that required intensive care. The neurological status was evaluated using the mRS score, with improvement being defined as improvement in the postoperative mRS score of 1 or greater. Complete obliteration was defined as the disappearance of the intradural draining vein, aneurysm, and varix after the initial treatment on digital subtraction angiography, MR angiography, or CT angiography for patients with dural AVFs, intradural AVFs, and epidural AVFs with intradural drainage. Among patients with epidural AVFs without intradural drainage, the disappearance of the epidural draining vein was required for complete obliteration.

Sample Size Calculation and Statistical Analysis

To consider the feasibility of the statistical analysis of our null hypothesis for the retreatment rate, a power analysis was performed before the collection of CCJ AVF cases. Among patients with spinal dural AVFs in previous studies,79 the retreatment rate after the initial treatment was 0.69%–3.4% in the neurosurgery group and 28%–36% in the endovascular group. Therefore, the sample size for patients with CCJ AVFs in the present study assumed that the retreatment rate will be 5% in the neurosurgery group and 50% in the endovascular group. In order for a 5% 2-sided test with a neurosurgery and endovascular group ratio of 2:1 to have 90% power to detect this difference, 34 and 17 patients were required in the neurosurgery and endovascular groups, respectively.

To validate comparability between the neurosurgery and endovascular groups, baseline characteristics for clinical variables (age, sex, symptom at onset, symptom duration, pretreatment mRS score, misdiagnosis, and steroid use), angiographic variables (fistula level, fistula side, feeding artery, aneurysm/varix, draining vein, and differential diagnosis), and MRI variables (brainstem, cervical, and thoracic edema) were compared between the two groups using Fisher’s exact test or the chi-square test for categorical variables and the Student t-test for continuous variables. The retreatment rate, overall complication rate, ischemic complication rate, and posttreatment changes in mRS scores were compared using Fisher’s exact test or the chi-square test. Risk factors for retreatment or poor outcomes were identified via a univariate logistic regression analysis, from which variables with a p value < 0.1 were selected. A multivariate logistic regression analysis was then used after simultaneously controlling for potential confounders. The software used for the analysis was SPSS Statistics version 23 (IBM Corp.). A 2-sided p value < 0.05 was considered to be significant.

Results

Patient Selection

Among the 111 consecutive patients with CCJ AVFs, 97 were included in the present study after exclusion of the following patients: those with intracranial AVFs (n = 2), spinal AVFs below the C3 level (n = 2), iatrogenic AVFs (n = 1), traumatic AVFs (n = 2), pediatric AVFs (n = 1), CCJ AVFs treated by a combination of neurosurgical and endovascular procedures (n = 4), and CCJ AVFs treated conservatively (n = 2). Neurosurgery was performed in 78 (80%) of the 97 patients, and endovascular treatment was performed in 19 (20%).

Baseline Characteristics

Baseline characteristics did not significantly differ between the two groups, except for sex (Table 1). Regarding clinical examinations, patients with CCJ AVFs were diagnosed at an older age (mean 67 ± 12 years [± SD]) and had a male predominance (74% of the 97 patients). They frequently presented with subarachnoid hemorrhage (SAH; 49%) or myelopathy due to venous congestion (28%) but rarely with intramedullary hemorrhage (7.2%). Overall, 15% of patients were asymptomatic. They commonly showed a dependent functional status (38%) and were misdiagnosed at the onset (26%). Angiographic examinations showed that most CCJ AVFs developed at the level of C1 (68%) or C2 (21%). Multiple AVFs were observed in 11% of cases. CCJ AVFs were frequently fed by spinal arteries (35%), and an aneurysm was present on the feeding artery or varix on the draining vein (41%). Most CCJ AVFs had intradural venous drainage (94%), which was ventral to the medulla oblongata or spinal cord in 100% and was ascending in 51%. Some cases had epidural venous drainage (15%), which was mostly located at the C2 level. AVFs were located on the dura mater (57%), on the intradural spinal cord or nerve (18%), or on the epidural space (14%). MRI revealed that patients with CCJ AVFs frequently had venous congestion in the medulla oblongata (28%) and in the cervical spinal cord (35%), but rarely in the thoracic spinal cord (6.2%).

TABLE 1.

Comparison of baseline characteristics

CharacteristicNeurosurgery Group (n = 78)Endovascular Group (n = 19)p Value
Age, yrs67 ± 1168 ± 130.87
Male sex62 (79)10 (53)0.037
Symptom at onset0.27
 SAH38 (49)10 (53)
 Intramedullary hemorrhage4 (5.1)3 (16)
 Venous congestion24 (31)3 (16)
 Asymptomatic12 (15)3 (16)
Symptom duration, mos5.8 ± 9.35.1 ± 170.78
Pretreatment mRS score ≥331 (40)7 (37)>0.99
Misdiagnosis22 (28)3 (16)0.38
Steroid pulse7 (9.0)2 (11)>0.99
Fistula level0.13
 C1 lesion55 (71)11 (58)
 C2 lesion13 (17)7 (37)
 Multiple lesions10 (13)1 (5.3)
Rt side39 (50)11 (58)0.62
Feeding artery
 ASA &/or PSA involved24 (31)10 (53)0.11
 ASA involved21 (27)8 (42)0.26
Aneurysm/varix31 (40)9 (47)0.61
Draining vein
 Intradural drainage75 (96)16 (84)0.087
 Ascending drainage39 (50)10 (53)>0.99
 Epidural drainage9 (12)6 (32)0.069
Differential diagnosis0.31
 Dural AVF46 (59)9 (47)
 Intradural AVF13 (17)4 (21)
  Radicular AVF10 (13)2 (11)
  Perimedullary AVF3 (3.8)2 (11)
 Epidural AVF9 (12)5 (26)
  w/ intradural drainage6 (7.7)2 (11)
  w/o intradural drainage3 (3.8)3 (16)
 Multiple AVF10 (13)1 (5.3)
MR imaging
 Brainstem edema22 (28)5 (26)>0.99
 Cervical edema27 (35)7 (37)>0.99
 Thoracic edema5 (6.4)1 (5.3)>0.99

ASA = anterior spinal artery; PSA = posterior spinal artery.

Values are shown as the number (%) of patients or as the mean ± SD unless indicated otherwise. Boldface type indicates statistical significance.

Primary Endpoint

While the years of operators’ experience in the management of neurosurgical disease did not significantly differ between the 2 groups, the neurosurgery group had a lower retreatment rate (neurosurgery, 2.6% vs endovascular, 63%; p < 0.001; Table 2). The retreatment rate was also lower in the neurosurgery group than in the endovascular group among patients with dural (0% vs 56%, p < 0.001), intradural (15% vs 75%, p = 0.053), and epidural (0% vs 80%, p = 0.005) AVFs.

TABLE 2.

Comparison of retreatments, complications, and outcomes

VariableNeurosurgery Group (n = 78)Endovascular Group (n = 19)p Value
Operators’ experience, yrs21 ± 9.018 ± 9.00.18
Retreatment2 (2.6)12 (63)<0.001
 Dural AVF0/465/9<0.001
 Intradural AVF2/133/40.053
  Radicular AVF1/101/2
  Perimedullary AVF1/32/2
 Epidural AVF0/94/50.005
 Multiple AVFs0/100/1NA
Overall complications17 (22)8 (42)0.084
Ischemic complications6 (7.7)5 (26)0.037
Posttreatment mRS score0.043
 Improvement 47 (60)7 (37)
 No change25 (32)7 (37)
 Decline6 (7.7)5 (26)

NA = not available.

Values are shown as the number (%) of patients or as the mean ± SD unless indicated otherwise. Boldface type indicates statistical significance.

Secondary Endpoints

After the initial treatment, the overall complication rate and mortality rate did not significantly differ (22% vs 42%, p = 0.084; and 2.6% vs 0%, p > 0.99; respectively) (Table 3). However, the neurosurgery group had a lower ischemic complication rate (7.7% vs 26%, p = 0.037; Table 3) and a higher percentage of patients with an improved mRS score with a median follow-up of 23 months (60% vs 37%, p = 0.043; Table 2) than the endovascular group.

TABLE 3.

Specific complications

CharacteristicNeurosurgery Group (n = 78)Endovascular Group (n = 19)p Value
Overall complications17 (22)8 (42)0.084
Ischemic complications6 (7.7)5 (26)0.037
 Spinal infarction4 (5.1)4 (21)
 Brainstem infarction1 (1.3)1 (5.3)
 Cerebellar infarction1 (1.3)0
Hemorrhagic complications4 (5.1)3 (16)0.13
 Postop hematoma3 (3.8)1 (5.3)
 Intraopvascular injury1 (1.3)2 (11)
CSF leak2 (2.6)0>0.99
Hydrocephalus2 (2.6)0>0.99
Transient cerebellar edema1 (1.3)0>0.99
Symptomatic vasospasm1 (1.3)0>0.99
Medullary hemorrhage01 (5.3)0.20
Upper extremity paresis1 (1.3)0>0.99
Transient respiratory arrest01 (5.3)0.20
Renal failure1 (1.3)0>0.99
Death2 (2.6)0>0.99

Values are shown as the number (%) of patients. Boldface type indicates statistical significance.

Risk Factors

After multivariate analyses, the independent risk factor associated with retreatment was endovascular treatment (OR 54, 95% CI 9.9–300; Table 4). Independent risk factors associated with poor outcomes (a postoperative mRS score of 3 or greater) were a pretreatment mRS score of 3 or greater (OR 13, 95% CI 2.7–62) and complications (OR 5.8, 95% CI 1.3–26; Table 5).

TABLE 4.

Risk factors associated with retreatment

Risk FactorUnivariateMultivariate
OR (95% CI)p ValueOR (95% CI)p Value
Endovascular treatment65 (12–351)<0.00154 (9.9–300)<0.001
Male sex0.28 (0.086–0.89)0.0320.57 (0.11–2.9)0.50
Spinal feeding artery2.9 (0.92–9.3)0.0691.9 (0.39–9.2)0.43
Epidural drainage2.6 (0.70–9.8)0.15
Misdiagnosis0.44 (0.090–2.1)0.30
Brainstem edema1.5 (0.47–5.1)0.48
Ascending drainage1.4 (0.43–4.3)0.59
Subarachnoid hemorrhage0.73 (0.23–2.3)0.59
Age at diagnosis1.0 (0.94–1.1)0.64
Symptom duration1.0 (0.96–1.1)0.74
Pretreatment mRS score ≥ 31.2 (0.38–3.8)0.76
Steroid pulse0.84 (0.094–7.4)0.87
Operator’s experience1.0 (0.94–1.1)0.88
Aneurysm/varix1.1 (0.34–3.4)0.89

Boldface type indicates statistical significance.

TABLE 5.

Risk factors associated with posttreatment mRS score ≥3

Risk FactorUnivariateMultivariate
OR (95% CI)p ValueOR (95% CI)p Value
Pretreatment mRS score ≥321 (5.5–78)<0.00113 (2.7–62)0.001
Complications4.3 (1.6–12)0.0055.8 (1.3–26)0.020
Age at diagnosis 1.1 (1.0–1.1)0.0221.1 (1.0–1.2)0.054
Descending drainage5.5 (1.9–15)0.0013.5 (0.73–17)0.12
Brainstem edema 5.6 (2.0–15)0.0013.7 (0.65–22)0.14
SAH0.35 (0.13–0.96)0.0411.9 (0.28–13)0.51
Misdiagnosis 3.0 (1.1–8.2)0.0301.5 (0.34–6.7)0.58
Operator’s experience 0.96 (0.91–1.0)0.14
Endovascular treatment2.3 (0.77–6.7)0.14
Steroid pulse2.8 (0.68–12)0.15
Symptom duration1.0 (0.99–1.1)0.21
Aneurysm/varix0.70 (0.26–1.9)0.47
Male sex1.3 (0.44–4.1)0.61
Spinal feeding artery1.3 (0.48–3.3)0.64
Retreatment1.3 (0.38–4.8)0.65
Epidural drainage0.78 (0.20–3.0)0.71

Boldface type indicates statistical significance.

Retreatment

Fourteen patients required retreatment (Table 6). In the neurosurgery group, 2 patients with intradural AVFs required additional neurosurgery, which was curative (Fig. 1). In the endovascular group, 12 patients with dural (n = 5), intradural (n = 3), and epidural (n = 4) AVFs required multiple additional treatments: 9 underwent neurosurgery, which was curative; however, the other 3 underwent endovascular procedures, which again resulted in treatment failure (Fig. 2). These 3 patients required a total of 10 additional treatments due to endovascular failure.

TABLE 6.

Characteristics of 14 patients who required retreatment

Pt No.Age (yrs), SexPresentationDiagnosisInitial Tx2nd Tx3rd Tx4th Tx5th Tx
175, MMyelopathypAVFSurgerySurgery
269, MMyelopathyrAVFSurgerySurgery
380, MAsymptomaticdAVFEndoEndoEndo
458, MHematomyeliadAVFEndoSurgery
577, MMyelopathydAVFEndoEndoEndoEndoSurgery
661, MMyelopathydAVFEndoEndoEndoGKRSSurgery
779, FSAHdAVFEndoSurgery
857, FHematomyeliaedAVFEndoSurgery
982, FSAHedAVFEndoSurgery
1133, FSAHedAVFEndoSurgery
1266, FSAHedAVFEndoSurgery
1365, MSAHpAVFEndoSurgery
1446, FSAHpAVFEndoSurgery
1573, FAsymptomaticrAVFEndoSurgery

dAVF = dural AVF; edAVF = epidural AVF; GKRS = Gamma Knife radiosurgery; pAVF = perimedullary AVF; pt = patient; rAVF = radicular AVF; Tx = treatment.

FIG. 1.
FIG. 1.

Patient 1. This 75-year-old male patient had a perimedullary AVF that was treated with neurosurgery. The patient presented with venous congestive myelopathy and underwent neurosurgery, in which the feeding artery at the C2 level was divided. However, an angiogram revealed a residual fistula fed by the C1 radiculopial artery. The patient underwent retreatment with neurosurgery, in which the feeding artery at C1 level was divided. Complete obliteration of the fistula was achieved. A: Anteroposterior image of a right vertebral artery angiogram showing a perimedullary arteriovenous fistula (arrow) fed by the C2 radiculopial artery (arrowheads). B: Angiogram obtained after the first surgery, showing a residual fistula fed by the C1 radiculopial artery (arrowheads). C: Angiogram obtained after the second surgery, showing the complete obliteration of the fistula. D: Intraoperative photograph from the first surgery, showing the C2 radiculopial artery (asterisk). E: Intraoperative photograph from the second surgery, showing the C1 radiculopial artery (asterisk). Figure is available in color online only.

FIG. 2.
FIG. 2.

Patient 5. This 77-year-old male patient underwent treatment of a dural AVF treated using an endovascular procedure. The patient presented with venous congestive myelopathy and underwent endovascular treatment, in which the feeding artery at C1 level was embolized. However, angiography showed a residual fistula. Endovascular embolization was performed four times but failed to obliterate the fistula. The patient underwent retreatment with neurosurgery, in which the intradural draining vein was clipped. Complete obliteration of the fistula was achieved. A and B: Anteroposterior images of left vertebral artery angiograms showing a dural AVF (arrows). C: Posteroanterior image of a right vertebral artery angiogram showing the intradural draining vein (arrow). D: CT scan obtained after the endovascular treatments, showing n-butyl cyanoacrylate and embolization coils (arrowhead) and the residual intradural draining vein (arrow). E: Intraoperative photograph of the last neurosurgical treatment, showing clipping of the intradural vein. Figure is available in color online only.

Angiographic Occlusion

After the initial treatment, angiography was performed in 73 of 78 patients in the neurosurgery group (digital subtraction angiography in 49 patients, MR angiography in 13, and CT angiography in 11), and complete obliteration was achieved in 96%. On the other hand, digital subtraction angiography was performed in all 19 patients in the endovascular group, and complete obliteration was achieved in 26% (p < 0.001).

After retreatment(s), angiography was performed in 73 of the 78 patients in the neurosurgery group, and complete obliteration was finally achieved in 99%. Angiography was performed in 17 of the 19 patients in the endovascular group, and complete obliteration was achieved in 88%. All of the 37 aneurysms/varices evaluated (100%) were successfully obliterated. Of the 83 intradural draining veins evaluated, 96% were successfully obliterated. Of the 12 epidural draining veins evaluated, 83% were successfully obliterated. After retreatments, the obliteration rates of aneurysms, varices, intradural veins, and epidural veins did not significantly differ between the two groups.

Complications

Among the overall complications in the 97 patients included in the present study (Table 3), ischemic complications were the most common (11%), followed by hemorrhagic complications (7.2%), CSF leakage (2.1%), and hydrocephalus (2.1%). Ischemic complications included 8 spinal infarctions, 2 brainstem infarctions, and 1 cerebellar infarction, which were associated with treatment procedures, including extensive coagulation or deep embolization of multiple radiculopial arteries. Ischemic complications resulted in permanent neurological deficits, such as motor paresis, sensory disturbance, and brainstem dysfunction.

Hemorrhagic complications included 4 postoperative hematomas and 3 intraoperative vascular injuries. Postoperative hematomas were removed in additional surgery without permanent neurological deficits. Intraoperative vascular injuries were repaired surgically or with endovascular treatment, with spinal infarction occurring in 1 patient. Most of the other complications, such as CSF leakage and hydrocephalus, were curable with additional treatment. Two patients died of systemic complications: renal failure in one patient and the impact of severe subarachnoid hemorrhage (World Federation of Neurosurgical Societies grade IV) in the other.

Discussion

In the present study, treatment results for CCJ AVFs were directly compared between neurosurgical and endovascular procedures. The results obtained revealed that neurosurgery was superior to endovascular treatment because of the lower rates of retreatment and ischemic complications as well as better outcomes. Complications associated with neurosurgical and endovascular treatments for CCJ AVFs were also described in detail. Ischemic complications resulted in permanent neurological deficits, while most of the other complications were curable with additional treatment. Endovascular treatment may have an advantage because of its minimal surgical invasiveness; however, the lower treatment success rate and attendant risk for significant morbidity outweighed its therapeutic benefits.

Since the late 1990s, the number of studies published on neurosurgical and endovascular treatments for CCJ AVFs has been increasing.1016 Two recent systematic reviews showed that between neurosurgical and endovascular treatments, neurosurgery is the preferred method for good outcomes.4,5 However, no information was provided on the comparability of the groups at baseline, retreatments and complications, or long-term outcomes. Furthermore, these studies included an older data set of patients (19704 and 19835). Therefore, evidence for the most effective treatment options for CCJ AVFs is limited in the literature. In contrast to previous studies, we provided information on comparability between the neurosurgery and endovascular groups at baseline, retreatments and complications, and long-term outcomes based on the most recent data set of patients. To the best of our knowledge, for the first time, we add to the growing body of literature demonstrating scientific evidence for the treatment choice for CCJ AVFs.

Clinical Implications

The angioarchitecture of CCJ AVFs may have been the reason for the higher retreatment and ischemic complication rates in the endovascular group. The diagnosis and treatment of CCJ AVFs are more difficult than those of thoracolumbar AVFs because of the small and complex neurovascular lesions.17 Recent angiographic studies reported that CCJ AVFs are frequently fed by spinal arteries as well as radiculomeningeal arteries;1,17 their arteriovenous shunts develop along the C1 or C2 nerve roots and may be located on the spinal cord, on the spinal nerves, and on the inner or outer surface of the dura mater,1 and they may have multiple AVFs.17 In the present study, it was difficult to embolize an intradural draining vein via the transarterial approach because of multiple feeding arteries and their dangerous anastomosis with spinal arteries in 63% of patients in the endovascular group. On the other hand, direct clipping of an intradural draining vein was simple and straightforward in 97% of patients in the neurosurgery group. Based on these results, direct surgery needs to be considered as the first-line treatment for CCJ AVFs.

Study Strengths and Limitations

The present study has a number of limitations. Since it was a retrospective analysis, there may have been a selection bias for the initial treatment of choice. A prospective randomized study is ideal for establishing the most effective treatment. However, a prospective study is difficult to perform because of the rarity of CCJ AVFs. Since our study included the largest number of patients, we consider its results to be of importance for vascular neurosurgeons and endovascular neurointerventionalists.

Conclusions

Neurosurgery resulted in better treatment results than endovascular treatment, specifically regarding retreatments, ischemic complications, and clinical outcomes; therefore, neurosurgery is recommended as the first-line treatment for patients with CCJ AVFs.

Acknowledgments

This study was financially supported by the Neurospinal Society of Japan and the Tokyo Metropolitan Government (Grant number, R030603004).

We thank the late Makoto Taniguchi for his contribution to this study. We also thank the investigators of the Neurospinal Society of Japan CCJ AVF study for data registration (listed in the Appendix).

Disclosures

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Appendix

Neurospinal Society of Japan CCJ AVF Investigators

Keisuke Ito: Department of Neurosurgery, Toho University Ohashi Hospital, Tokyo, Japan. Motoyuki Iwasaki: Department of Neurosurgery, Otaru General Hospital, Hokkaido, Japan. Hisaaki Uchikado: Department of Neurosurgery, Kurume University Hospital, Fukuoka, Japan. Daisuke Umebayashi: Department of Neurosurgery, Kyoto Prefectural University of Medicine, Kyoto, Japan. Munehiro Otsuka: Department of Neurosurgery, Saitama Medical University International Medical Center, Saitama, Japan. Tatsuya Ohtonari: Department of Spinal Surgery, Brain Attack Center, Ota Memorial Hospital, Hiroshima, Japan. Junpei Oda: Department of Comprehensive Strokotology, Fujita Health University School of Medicine, Aichi, Japan. Hiroto Kageyama: Department of Neurosurgery, Hyogo College of Medicine, Hyogo, Japan. Ryu Kurokawa: Department of Neurosurgery, Dokkyo Medical University Hospital, Tochigi, Japan. Satoshi Koizumi: Department of Neurosurgery, The University of Tokyo Hospital, Tokyo, Japan. Taku Sugawara: Department of Spinal Surgery, Akita Cerebrospinal and Cardiovascular Center, Akita, Japan. Yasuhiro Takeshima: Department of Neurosurgery, Nara Medical University School of Medicine, Kashihara, Japan. Yoshitaka Nagashima: Department of Neurosurgery, Nagoya University Hospital, Nagoya, Japan. Misao Nishikawa: Department of Neurosurgery, Moriguchi-Ikuno Memorial Hospital, Osaka, Japan. Masashi Fujimoto: Department of Neurosurgery, Mie University Graduate School of Medicine, Mie, Japan. Fumiaki Honda: Department of Neurosurgery, Gunma University Hospital, Gunma, Japan. Seishi Matsui: Department of Neurosurgery, Ehime University Graduate School of Medicine, Ehime, Japan. Yoshihisa Matsumoto: Department of Neurosurgery, St. Mary’s Hospital, Fukuoka, Japan. Yasuyuki Miyoshi: Department of Neurosurgery, Kawasaki Medical School General Medical Center, Okayama, Japan. Hidetoshi Murata: Department of Neurosurgery, Yokohama City University Graduate School of Medical Sciences and School of Medicine, Kanagawa, Japan. Takao Yasuhara: Department of Neurological Surgery, Okayama University Graduate School of Medicine, Dentistry, and Pharmaceutical Sciences, Okayama, Japan. Hitoshi Yamahata: Department of Neurosurgery, Kagoshima University Graduate School of Medical and Dental Sciences, Kagoshima, Japan. Shinji Yamamoto: Department of Neurosurgery, Spine Center, Ohnishi Neurological Center, Hyogo, Japan. Yu Yamamoto: Spine and Peripheral Nerve Center, Department of Neurosurgery, Inazawa MunicipaI Hospital, Aichi, Japan.

Author Contributions

Conception and design: Takai. Acquisition of data: all authors. Analysis and interpretation of data: Takai, Endo, Seki, Inoue. Drafting the article: Takai. Critically revising the article: Endo, Seki, Inoue, Koyanagi, Mitsuhara. Reviewed submitted version of manuscript: Endo, Seki, Inoue, Koyanagi, Mitsuhara. Approved the final version of the manuscript on behalf of all authors: Takai. Statistical analysis: Takai. Study supervision: Takai, Endo.

Supplemental Information

Previous Presentations

Portions of this work were presented in abstract form at the 35th Annual Meeting of the Neurospinal Society of Japan, Yokohama, Japan, November 9, 2020.

References

  • 1

    Hiramatsu M, Sugiu K, Ishiguro T, Kiyosue H, Sato K, Takai K, et al. Angioarchitecture of arteriovenous fistulas at the craniocervical junction: a multicenter cohort study of 54 patients. J Neurosurg. 2018;128(6):18391849.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Fujimoto S, Takai K, Nakatomi H, Kin T, Saito N. Three-dimensional angioarchitecture and microsurgical treatment of arteriovenous fistulas at the craniocervical junction. J Clin Neurosci. 2018;53:140146.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Endo T, Shimizu H, Sato K, Niizuma K, Kondo R, Matsumoto Y, et al. Cervical perimedullary arteriovenous shunts: a study of 22 consecutive cases with a focus on angioarchitecture and surgical approaches. Neurosurgery. 2014;75(3):238249.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Wang JY, Molenda J, Bydon A, Colby GP, Coon AL, Tamargo RJ, Huang J. Natural history and treatment of craniocervical junction dural arteriovenous fistulas. J Clin Neurosci. 2015;22(11):17011707.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Zhao J, Xu F, Ren J, Manjila S, Bambakidis NC. Dural arteriovenous fistulas at the craniocervical junction: a systematic review. J Neurointerv Surg. 2016;8(6):648653.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Takai K. Update on the diagnosis and treatment of arteriovenous fistulas at the craniocervical junction: a systematic review of 92 cases. J neuroendovascular Ther. 2019;13(3):125135.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Takai K, Endo T, Yasuhara T, Seki T, Watanabe K, Tanaka Y, et al. Neurosurgical versus endovascular treatment of spinal dural arteriovenous fistulas: a multicenter study of 195 patients. J Neurosurg Spine. 2021;34(3):514521.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8

    Bakker NA, Uyttenboogaart M, Luijckx GJJ, Eshghi OS, Mazuri A, Metzemaekers JD, et al. Recurrence rates after surgical or endovascular treatment of spinal dural arteriovenous fistulas: a meta-analysis. Neurosurgery. 2015;77(1):137144.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9

    Goyal A, Cesare J, Lu VM, Alvi MA, Kerezoudis P, Brinjikji W, et al. Outcomes following surgical versus endovascular treatment of spinal dural arteriovenous fistula: a systematic review and meta-analysis. J Neurol Neurosurg Psychiatry. 2019;90(10):11391146.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Onda K, Yoshida Y, Watanabe K, Arai H, Okada H, Terada T. High cervical arteriovenous fistulas fed by dural and spinal arteries and draining into a single medullary vein: report of 3 cases. J Neurosurg Spine. 2014;20(3):256264.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Alshekhlee A, Edgell RC, Kale SP, Kitchener J, Vora N. Endovascular therapy of a craniocervical pial AVF fed by the anterior spinal artery. J Neuroimaging. 2013;23(1):102104.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Nakagawa I, Park HS, Hironaka Y, Wada T, Kichikawa K, Nakase H. Cervical spinal epidural arteriovenous fistula with coexisting spinal anterior spinal artery aneurysm presenting as subarachnoid hemorrhage—case report. J Stroke Cerebrovasc Dis. 2014;23(10):e461e465.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Chen G, Wang Q, Tian Y, Gu Y, Xu B, Leng B, Song D. Dural arteriovenous fistulae at the craniocervical junction: the relation between clinical symptom and pattern of venous drainage. Acta Neurochir Suppl. 2011;110(Pt 2):99-104.

    • Search Google Scholar
    • Export Citation
  • 14

    Goto Y, Hino A, Shigeomi Y, Oka H. Surgical management for craniocervical junction arteriovenous fistula targeting the intradural feeder. World Neurosurg. 2020;144:e685e692.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Kinouchi H, Mizoi K, Takahashi A, Nagamine Y, Koshu K, Yoshimoto T. Dural arteriovenous shunts at the craniocervical junction. J Neurosurg. 1998;89(5):755761.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Kim DJ, Willinsky R, Geibprasert S, Krings T, Wallace C, Gentili F, Terbrugge K. Angiographic characteristics and treatment of cervical spinal dural arteriovenous shunts. AJNR Am J Neuroradiol. 2010;31(8):15121515.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Sato K, Endo T, Niizuma K, Fujimura M, Inoue T, Shimizu H, Tominaga T. Concurrent dural and perimedullary arteriovenous fistulas at the craniocervical junction: case series with special reference to angioarchitecture. J Neurosurg. 2013;118(2):451459.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Images from Minchev et al. (pp 479–488).

  • View in gallery

    Patient 1. This 75-year-old male patient had a perimedullary AVF that was treated with neurosurgery. The patient presented with venous congestive myelopathy and underwent neurosurgery, in which the feeding artery at the C2 level was divided. However, an angiogram revealed a residual fistula fed by the C1 radiculopial artery. The patient underwent retreatment with neurosurgery, in which the feeding artery at C1 level was divided. Complete obliteration of the fistula was achieved. A: Anteroposterior image of a right vertebral artery angiogram showing a perimedullary arteriovenous fistula (arrow) fed by the C2 radiculopial artery (arrowheads). B: Angiogram obtained after the first surgery, showing a residual fistula fed by the C1 radiculopial artery (arrowheads). C: Angiogram obtained after the second surgery, showing the complete obliteration of the fistula. D: Intraoperative photograph from the first surgery, showing the C2 radiculopial artery (asterisk). E: Intraoperative photograph from the second surgery, showing the C1 radiculopial artery (asterisk). Figure is available in color online only.

  • View in gallery

    Patient 5. This 77-year-old male patient underwent treatment of a dural AVF treated using an endovascular procedure. The patient presented with venous congestive myelopathy and underwent endovascular treatment, in which the feeding artery at C1 level was embolized. However, angiography showed a residual fistula. Endovascular embolization was performed four times but failed to obliterate the fistula. The patient underwent retreatment with neurosurgery, in which the intradural draining vein was clipped. Complete obliteration of the fistula was achieved. A and B: Anteroposterior images of left vertebral artery angiograms showing a dural AVF (arrows). C: Posteroanterior image of a right vertebral artery angiogram showing the intradural draining vein (arrow). D: CT scan obtained after the endovascular treatments, showing n-butyl cyanoacrylate and embolization coils (arrowhead) and the residual intradural draining vein (arrow). E: Intraoperative photograph of the last neurosurgical treatment, showing clipping of the intradural vein. Figure is available in color online only.

  • 1

    Hiramatsu M, Sugiu K, Ishiguro T, Kiyosue H, Sato K, Takai K, et al. Angioarchitecture of arteriovenous fistulas at the craniocervical junction: a multicenter cohort study of 54 patients. J Neurosurg. 2018;128(6):18391849.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Fujimoto S, Takai K, Nakatomi H, Kin T, Saito N. Three-dimensional angioarchitecture and microsurgical treatment of arteriovenous fistulas at the craniocervical junction. J Clin Neurosci. 2018;53:140146.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Endo T, Shimizu H, Sato K, Niizuma K, Kondo R, Matsumoto Y, et al. Cervical perimedullary arteriovenous shunts: a study of 22 consecutive cases with a focus on angioarchitecture and surgical approaches. Neurosurgery. 2014;75(3):238249.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Wang JY, Molenda J, Bydon A, Colby GP, Coon AL, Tamargo RJ, Huang J. Natural history and treatment of craniocervical junction dural arteriovenous fistulas. J Clin Neurosci. 2015;22(11):17011707.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Zhao J, Xu F, Ren J, Manjila S, Bambakidis NC. Dural arteriovenous fistulas at the craniocervical junction: a systematic review. J Neurointerv Surg. 2016;8(6):648653.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Takai K. Update on the diagnosis and treatment of arteriovenous fistulas at the craniocervical junction: a systematic review of 92 cases. J neuroendovascular Ther. 2019;13(3):125135.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Takai K, Endo T, Yasuhara T, Seki T, Watanabe K, Tanaka Y, et al. Neurosurgical versus endovascular treatment of spinal dural arteriovenous fistulas: a multicenter study of 195 patients. J Neurosurg Spine. 2021;34(3):514521.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8

    Bakker NA, Uyttenboogaart M, Luijckx GJJ, Eshghi OS, Mazuri A, Metzemaekers JD, et al. Recurrence rates after surgical or endovascular treatment of spinal dural arteriovenous fistulas: a meta-analysis. Neurosurgery. 2015;77(1):137144.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9

    Goyal A, Cesare J, Lu VM, Alvi MA, Kerezoudis P, Brinjikji W, et al. Outcomes following surgical versus endovascular treatment of spinal dural arteriovenous fistula: a systematic review and meta-analysis. J Neurol Neurosurg Psychiatry. 2019;90(10):11391146.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Onda K, Yoshida Y, Watanabe K, Arai H, Okada H, Terada T. High cervical arteriovenous fistulas fed by dural and spinal arteries and draining into a single medullary vein: report of 3 cases. J Neurosurg Spine. 2014;20(3):256264.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Alshekhlee A, Edgell RC, Kale SP, Kitchener J, Vora N. Endovascular therapy of a craniocervical pial AVF fed by the anterior spinal artery. J Neuroimaging. 2013;23(1):102104.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Nakagawa I, Park HS, Hironaka Y, Wada T, Kichikawa K, Nakase H. Cervical spinal epidural arteriovenous fistula with coexisting spinal anterior spinal artery aneurysm presenting as subarachnoid hemorrhage—case report. J Stroke Cerebrovasc Dis. 2014;23(10):e461e465.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Chen G, Wang Q, Tian Y, Gu Y, Xu B, Leng B, Song D. Dural arteriovenous fistulae at the craniocervical junction: the relation between clinical symptom and pattern of venous drainage. Acta Neurochir Suppl. 2011;110(Pt 2):99-104.

    • Search Google Scholar
    • Export Citation
  • 14

    Goto Y, Hino A, Shigeomi Y, Oka H. Surgical management for craniocervical junction arteriovenous fistula targeting the intradural feeder. World Neurosurg. 2020;144:e685e692.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15

    Kinouchi H, Mizoi K, Takahashi A, Nagamine Y, Koshu K, Yoshimoto T. Dural arteriovenous shunts at the craniocervical junction. J Neurosurg. 1998;89(5):755761.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Kim DJ, Willinsky R, Geibprasert S, Krings T, Wallace C, Gentili F, Terbrugge K. Angiographic characteristics and treatment of cervical spinal dural arteriovenous shunts. AJNR Am J Neuroradiol. 2010;31(8):15121515.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Sato K, Endo T, Niizuma K, Fujimura M, Inoue T, Shimizu H, Tominaga T. Concurrent dural and perimedullary arteriovenous fistulas at the craniocervical junction: case series with special reference to angioarchitecture. J Neurosurg. 2013;118(2):451459.

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 494 494 0
Full Text Views 789 786 243
PDF Downloads 739 735 219
EPUB Downloads 0 0 0