Surgical Treatment of Spinal Dural Arteriovenous Fistulae: A Consecutive Series of 154 Patients

Abstract

BACKGROUND:

Embolization of spinal dural arteriovenous fistulae (SDVAFs) has emerged as an alternative to surgery. However, surgical disconnection is a simple and effective procedure.

OBJECTIVE:

To review results and complications of surgical treatment of 154 consecutive SDAVFs.

METHODS:

The records of 154 consecutive patients with SDAVFs were retrospectively reviewed.

RESULTS:

There were 120 males and 34 females (male/female ratio 3.5:1, mean age 63.6 years). The SDAVFs were located at the thoracic level in 92 patients and at the lumbar and sacral spine levels in 45 and 15 patients, respectively. The most common presenting symptoms were motor dysfunction (65 patients), sensory loss (31 patients), and paresthesias without sensory loss (13 patients). The mean interval from symptom onset to definitive diagnosis was 24.7 months (median 12 months). Surgery resulted in complete exclusion of the fistula at first attempt in 146 patients (95%). There were no deaths or major neurological complications related to the surgery. Six percent of patients experienced subjective or objective worsening of preoperative symptoms and signs by the time of discharge that persisted at follow-up. Other surgical complications consisted of wound infection in 2 patients and deep venous thrombosis in 3. Eight patients were lost to follow-up; 141 patients (96.6%) experienced improvement (120 patients, 82.2%) or stability (21 patients, 14.4%) of motor function at last follow-up compared with their preoperative status. Other symptoms such as numbness, sphincter dysfunction, and dysesthesias/neuropathic pain improved in 51.5%, 45%, and 32.6%, respectively.

CONCLUSION:

Surgical obliteration of SDAVFs is safe and very effective. Prognosis of motor function is favorable after surgical treatment.

Complications, Outcome, Progressive myelopathy, Spinal angiography, Spinal dural arteriovenous fistula, Surgery

Spinal dural arteriovenous fistulae (SDAVFs) are a rare but curable cause of progressive myelopathy. Surgical treatment, consisting of interruption of the anomalous arteriovenous connection, has long been considered the mainstay of treatment. With the advances in endovascular techniques and the availability of new embolic materials, endovascular embolization has been advocated as a valid alternative or even as a primary treatment of SDAVFs. However, surgical disconnection of SDAVFs is a straightforward procedure with a high success rate and virtually no risk of recurrence or incomplete treatment. In this report, we analyze outcome and risk of surgery in a large series of 154 consecutive patients treated with surgery as the sole or primary treatment modality over a 23-year period at our institution.

MATERIALS AND METHODS

The medical records and radiological reports of patients with a diagnosis of SDAVFs who underwent surgical treatment at Saint Mary's Hospital, Mayo Clinic (Rochester, Minnesota) from June 1985 until March 2008 were retrospectively reviewed. Patients with intracranial fistulae and retrograde venous drainage causing myelopathy were excluded. Similarly, patients with fistulae located in the region of the foramen magnum and the craniovertebral junction were excluded from the analysis because these lesions have clinical and angiographic characteristics different from those of classic SDAVFs. The clinical characteristics of patients treated between 1986 and 1999 were reported in a previous publication.¹ To investigate trends in clinical presentation and diagnosis, comparisons were made when appropriate between the early (1986-1999) and recent (2000-2008) experience.

All patients underwent a detailed history and neurological examination by a neurologist at the time of evaluation in our institution. Follow-up data based on examination at discharge and on postoperative visits were gathered. Motor impairment was retrospectively graded using the modified Aminoff-Logue myelopathy scale² (Table 1). Radiological information from computed tomography (CT) myelography, magnetic resonance imaging (MRI), magnetic resonance angiography (MRA), and conventional catheter angiography was collected.

TABLE 1.

Modified Aminoff-Logue Disability Scale

All patients included in this study underwent surgical disconnection of the fistula, consisting of target laminectomy, dura opening, and either coagulation and section of the draining vein or disconnection of the vein with an aneurysm clip. After surgery, all patients had a detailed neurological examination. Postoperative neuroradiological investigations varied and were left to the discretion of the treating neurosurgeon. However, all patients who showed any deterioration after surgery or failed to improve underwent postoperative imaging studies to document occlusion of the fistula.

Results are presented using descriptive statistics. The study was approved by the Mayo Foundation Institutional Review Board, and all patients included gave their informed consent to the use of their clinical data for research purposes.

RESULTS

This study comprises 154 patients, 120 men and 34 women (male/female ratio 3.5:1) with a mean age of 63.6 years (range 27-94 years). Eighty-six patients were treated in the early phase of the study (1986-1999) and 68 in the recent phase (2000-2008). The mean age was similar in male and female patients (64 and 62.2 years, respectively). The mean age and male/female ratio were 63.8 years and 3.8 in the early (1986-1999) experience and 64.5 years and 3.3 in the recent (2000-2008) experience.

Clinical Presentation

The mean interval from symptom onset to diagnosis (available in 151 patients) was 24.7 months. This interval ranged from 3 days (in a patient with history of cancer who experienced acute onset of leg weakness and underwent spinal MRI to rule out metastatic disease) to 276 months (in a patient in whom feet and calves paresthesias first developed that worsened after general anesthesia done for testicular cystectomy). The median time to diagnosis was 12 months. Surprisingly, the mean and median intervals from symptoms onset to diagnosis were not different between the early (mean, 24.1 months; median, 12 months) and the recent (mean, 25.7 months; median, 12 months) experience.

Before referral to our institution, 31 patients had undergone some form of invasive treatment for symptoms that, in retrospect, could be attributed to the SDAVF. Surgery for possible degenerative spine disease was the most common invasive treatment. Four patients had undergone attempted fistula treatment at other institutions by epidural surgical approaches in 2, acrylic embolization in 1, and preoperative acrylic embolization followed by epidurally directed fistula obliteration in the remaining patient.

Uniformly, patients reported worsening of symptoms after every surgical intervention not directed at obliteration of the fistula. Overall, 8.1% of patients in the early and 19.1% in the late experience underwent invasive surgical procedures other than attempted fistula obliteration for symptoms that, in retrospect, could be attributed to the SDAVF.

Neurological Symptoms and Signs

Presenting symptoms included isolated weakness³ in 66 patients (42.9%, 3 patients experienced a sudden presentation with paraplegia), decreased sensation in the lower extremities in 31 patients (20.1%), pain in 15 patients (9.7%), and paresthesias (tingling, tightening sensation) or dysesthesias (burning or cold sensation) in 13 patients (8.5%). Seventeen patients (11%) reported the association of pain with weakness (10 patients), numbness (5 patients), or both (2 patients). Five patients (3.2%) presented with an association of weakness and numbness in the lower extremities, whereas 7 patients (4.6%) had only sphincter dysfunction.

Subjective and objective motor weakness was the most common finding and was present during the presurgical course of the disease in all patients (154/154, 100%). Progressive lower extremity weakness was present in 115 patients. In 7 patients, lower extremity weakness was associated with upper extremity weakness. In 8 patients, the weakness had progressed to complete paraplegia by the time the diagnosis of SDAVF was made. A rather abrupt and acute onset of motor symptoms was reported by 20 subjects. Weakness present only after standing and/or walking was the main symptom at the time of diagnosis in 18 patients. The severity of motor impairment at diagnosis according to the modified Aminoff-Logue scale is summarized in Table 2.

TABLE 2.

Distribution of Motor Function at Diagnosis Based on the Aminoff-Logue Scale

Sensory symptoms consisted of numbness involving the lower extremities in 138 patients, associated in most cases with symmetrical diffuse loss of sensation in the lower extremities on examination. These sensory symptoms in the legs were usually symmetrical. However, asymmetry was noted in 7 patients and exclusively unilateral lower extremity involvement was present in 6 patients.

Sphincter dysfunction was present in 112 patients. Manifestations included urinary retention or incontinence, either alone or associated with constipation or fecal incontinence.

Imaging Studies

Early in our series, CT myelography was commonly performed as a first diagnostic examination (29 patients). Since 1997, CT myelography was limited to patients who could not undergo MRI, and only 5 patients underwent CT myelography as the first diagnostic procedure. A total of 147 patients underwent preoperative spinal MRI evaluation. MRA was performed before definitive spinal catheter angiography in 42 patients. All but 1 patient underwent spinal angiography in 1 or more sessions with the intent to include the total spine and posterior extracranial and intracranial vasculature when necessary. The 1 patient who did not undergo spinal angiography had MRI findings interpreted as a possible inflammatory process and underwent laminectomy for possible spinal cord biopsy. In this patient, the fistula was recognized and ligated on direct surgical exploration.

Myelography with subsequent CT performed at our institution revealed enlarged vessels on the dorsal surface of the spinal cord in all patients in this study and occasionally (10%) on the ventral side as well. No patient had only ventral spinal cord abnormal vasculature.

MRI showed findings consistent with an SDAVF in 146 of 147 patients (99.3%) (Figure 1A). In the early experience, 1 patient underwent CT myelography after MRI to confirm abnormal vascularity. In 126 patients (85.7%), MRI showed T2-signal hyperintensity involving the lower spinal cord and conus medullaris, as well as a variable length of perimedullary vessel abnormalities that were usually abnormally prominent on the dorsal surface of the spinal cord and conus. The abnormal spinal cord signal affected the conus medullaris alone in 2 of these patients, whereas abnormally prominent perimedullary vessels extending from the conus up to the foramen magnum were the only MRI findings in 1 patient. In 1 other patient, a syrinx was associated with the T2-signal changes. Six (4.1%) had segmental intramedullary T2-signal changes that did not involve the conus medullaris as well as a variable length of perimedullary vessel prominence. Seven patients (4.8%) did not have an abnormal intramedullary T2 signal, but abnormally increased flow voids along the surface of the spinal cord were visible on noncontrast images and were confirmed on contrast-enhanced images. In 1 patient, vessel abnormalities were confined to the craniocervical junction region, and the fistula was located at C4. In 7 additional patients (4.8%), there were signal changes on T2-weighted images in the spinal cord but no appreciable vessel abnormalities.

FIGURE 1.

Most common findings on preoperative imaging in patients with a spinal dural arteriovenous fistula (SDAVF). A, T2-weighted sagittal magnetic resonance imaging showing increased intramedullary signal involving the lower spinal cord and conus medullaris. Abnormally increased flow voids are present along the dorsal surface of the spinal cord, representing arterialized perimedullary veins. B, magnetic resonance angiography image showing a prominent, tortuous perimedullary vein as part of an abnormally dilated coronal venous plexus. C, selective injection of the left T7 radicular artery demonstrates early filling of an enlarged and tortuous draining medullary vein, consistent with a diagnosis of an SDAVF.

Spinal MRA was performed in 42 patients as a second-level imaging examination before angiography. MRA demonstrated perimedullary vessel abnormalities consistent with a diagnosis of DAVF in 40 patients (95.2%) (Figure 1B). Findings correctly predicted the level of the fistula in 18 patients (42.9%). In 6 more patients (14.3%), MRA findings predicted the side of the fistula and the level to within 1 (5 patients) or 2 (1 patient) vertebral levels.

Catheter angiography detected the fistula on the first attempt in 150 patients (Figure 1C). A repeat angiogram was obtained at 1 year for progressive clinical worsening in 1 patient, and a SDAVF was identified on this second angiogram. One angiogram was negative for the presence of a direct dural arteriovenous shunt but demonstrated prominent vessel abnormalities from the T6 to the L2 levels, and the patient then underwent explorative surgery. One angiogram was negative, but MRA performed previously had suggested a SDAVF; the patient was operated on at the level indicated on the magnetic resonance angiogram, with definitive intraoperative diagnosis and exclusion of the fistula.

The artery of Adamkiewicz was identified as arising from the same level as the fistula in only 2 patients, although lesser radiculomedullary arteries were identified occasionally at the level of the fistula. Four patients had an associated varix of the draining vein, compatible with the high-flow, high-pressure conduit visualized on angiography. The spinal levels of the fistulae in this series are summarized in Table 3. There were no major complications related to angiography. Three patients who were experiencing subacute deterioration immediately before angiography experienced slight worsening of their symptoms after angiography.

TABLE 3.

Fistula Level

Surgical Treatment

Operative treatment consisted of 2-level laminectomy and exposure of the fistula site at the intervertebral foramen. The feeding vessel was often identified epidurally. The dura was opened, and the draining vein identified as it pierced the dura and either coagulated and divided or was excluded with an aneurysm clip. Epidural feeding arteries were also coagulated as adjunctive, although not indispensable treatment. All procedures were performed with the aid of the surgical microscope. Early in this series, partial vein removal from the dorsal surface of the spinal cord in conjunction with fistula disconnection was performed in 1 patient. Of the 154 SDAVFs, 146 (94.8%) were successfully obliterated at the initial surgery. Second and third procedures were necessary to close the fistula in 7 patients and 1 patient, respectively. Intraoperative electrophysiological monitoring was not routinely used.

Complications

By the time of discharge, 130 patients (84.4%) experienced improvement (24 patients, 15.6%) or stability (106 patients, 68.8%) of their preoperative symptoms. Worsening of preoperative symptoms or onset of new symptoms immediately after surgery occurred in 24 patients (15.6%). Eight patients experienced worsening of their motor status. Five of these 8 patients returned to their preoperative status or better in 1 month (2 patients), 3 months (1 patient), and 6 months (2 patients). Two patients who experienced worsening of their preoperative status required repeat surgery to successfully obliterate the fistula. One of these patients regained his preoperative (before the first surgery) functional status 3 months after the second surgery. The other patient experienced some improvement but did not return to the functional status he had before the first surgery. Only 1 patient (with a history of multiple previous spinal surgeries for a developmental spine anomaly) did not experience any improvement after his postoperative worsening. Ten patients experienced worsening of their sphincter function immediately after surgery. Six of these 10 patients had urinary retention before surgery. Five of these 10 patients regained their preoperative function or were better after 1 month (1 patient), 3 months (1 patient), and 6 months (2 patients) or at last follow-up (1 patient). Two patients experienced some degree of improvement without regaining their preoperative function. Three patients had permanent postoperative worsening of sphincter control.

Five patients experienced both worsening of motor function and sphincter control after surgery. Three patients experienced recovery to their preoperative function or better (1 of them after a second surgery); improvement was only partial in the remaining 2 patients. Finally, 1 patient had untreatable neuropathic pain with functional loss for 2 weeks after surgery. At 1-month follow-up, her function had improved beyond her preoperative status.

Other surgical complications included a superficial wound infection in 2 patients and deep venous thrombosis in 3 patients with paraplegia or severe paraparesis.

Follow-up

Follow-up information was available for 146 patients (94.2%) at a mean of 17.7 months (range 3-98 months). The mean and median follow-up periods were 7.5 months and 6 months in the first group and 31.1 months and 18 months in the second group, respectively. Eight patients were lost to follow-up, and no follow-up information after discharge was available. At the last follow-up, 141 patients (96.6%) had experienced either improvement (120 patients, 82.2%) or stability (21 patients, 14.4%) of their motor function based on the Aminoff-Logue scale compared with their preoperative status (Table 4). Five patients (3.4%) reported worse motor symptoms.

TABLE 4.

Surgical Outcome Based on Preoperative Aminoff-Logue Scale Grade^a

With regard to sensory symptoms, numbness was present at diagnosis in 138 patients. Four of these patients were lost to follow-up. Of the remaining 134, 69 (51.5%) reported improvement, 63 (47%) stability, and 2 (1.5%) worsening of their preoperative symptoms (both of these patients started experiencing worsening numbness 4 years after surgery in the absence of fistula recurrence on repeat neuroimaging studies). Dysesthesias and neuropathic pain were present in 84 patients at diagnosis; 1 of these patients was lost to follow-up. Dysesthesias and/or neuropathic pain were better in 27 (32.6%), the same in 47 (56.6%), and worse at follow-up compared with the time of diagnosis in 9 (10.8%). Three additional patients had neither dysesthesia nor neuropathic pain at the time of diagnosis but reported such symptoms at follow-up.

Sphincter dysfunction was present in 112 patients at diagnosis, 3 of whom were lost to follow-up. Of the remaining patients, 49 (45%) were better, 51 (47%) were stable, and 9 (8%) were worse than before surgery. Sphincter dysfunction had developed at last follow-up in 4 patients who had no urinary symptoms at admission.

Eleven patients, after experiencing improvement of their symptoms, noted a delayed functional deterioration. All but 1 of these patients underwent follow-up imaging that demonstrated persistent obliteration of the fistula.

DISCUSSION

Diagnosis

SDAVFs continue to be an unrecognized cause of progressive myelopathy despite significant advances in imaging studies.^1,4–9 In our study, no difference was noted in the interval from presentation to final diagnosis between the early (mean 12 months, median 24.1 months) and late (mean 12 months, median 25.7 months) phases of this series. Symptoms at onset are usually nonlocalizing and subtle and often attributed to aging, stress, or exercise.^5,9–19 Furthermore, early symptoms of SDAVFs are often erroneously attributed to coexisting common abnormalities on neuroradiological imaging such as lumbar stenosis, vertebral antero/retrolisthesis, and lumbar disc herniation. In fact, nearly 10% of patients in our series underwent surgery other than correction of the fistula for symptoms that, in retrospect, were related to the SDAVF. The increase in the percentage of patients undergoing incorrect surgery in the later period of the series (14.7% vs 5.8% in the earlier series) may be related to greater tendency to perform spine surgery over the past decade. Alternatively, this observation may be related to the fact that patients with more complex and less typical symptoms were referred to our institution in the more recent years.

Imaging

The diagnosis of an SDAVF is based on classic, although not pathognomonic, radiological changes. On MRI, intramedullary hyperintensity on T2-weighted images ascending from the conus medullaris and abnormal vascular flow voids or enlarged enhancing vessels along the surface of the spinal cord are classic findings,^4,20 but they are not the rule. Only 1 of these described changes may be present, and/or the hyperintensity on T2-weighted images may be localized in a portion of the spinal cord other than the conus, engendering suspicion of different etiologies of the signal abnormality such as neoplasm, infection, and inflammatory disease.²¹ Thus, an SDAVF cannot be excluded if classic MRI findings are not seen. Spinal angiography remains the reference standard.

Spinal angiography in such cases can be challenging because many of these patients are elderly with tortuous and atherosclerotic vasculature, making selective catheterization of the intercostal and lumbar arteries difficult. Because of the large number of vessels that need to be studied for a complete study, the contrast material load can be significant, which can be a problem in patients with underlying borderline renal function. In recent years, we routinely performed MRA before spinal angiography. Although MRA does not always allow precise fistula localization, it has allowed us to decrease the contrast load in many cases by restricting the area on which to focus the catheter angiographic study.^22–28 Since 2003, MRA has been routinely used as a second-level study in all patients with a suspected SDAVF at our institution.

Recently, we also often perform 3-dimensional C-arm cone beam CT during selective intercostal or lumbar arterial injection after an SDAVF has been first identified with biplane catheter spinal angiography.²⁹ This provides exquisite multiplanar anatomic detail of the site of fistulization, the feeding arteries, the draining veins, and the spatial relationship of these vascular structures to adjacent osseous structures, which can be helpful in preoperative planning (Figure 2).

FIGURE 2.

A spinal dural arteriovenous fistula arises from the right T7 intercostal artery, as visualized on 3-dimensional C-arm cone beam computed tomography angiography, which was performed during catheter spinal angiography. The feeding artery entering under the T7 pedicle (white arrow) and an enlarged draining perimedullary vein (black arrow) are seen in this image section.

Outcome After Surgical Therapy

Surgical disconnection between the nidus within the dural leaves and the draining vein emerging intradurally at the level of the neural foramen along with its corresponding nerve is a technically simple procedure and a very effective treatment. Surgical correction of the fistula is a particularly effective treatment for the motor symptoms and signs related to the fistula.^11,30 In our series, approximately 97% of patients experienced some improvement (82.2%) or stabilization (14.4%) of their progressive motor symptoms at last follow-up. Because improvement of motor symptoms is common after successful obliteration, lack of improvement or worsening of motor symptoms should prompt further imaging studies to rule out persistence or recurrence of the fistula.

Symptoms other than motor are less likely to improve even after successful disconnection of the fistula. Numbness, which was present at the time of diagnosis in 90% of patients in our series, improved at last follow-up in 51.5%. Dysesthesias and neuropathic pain improved in only 32.6% of cases. Sphincter dysfunction improved after surgery in 45% of patients.

Although no major neurological morbidity and no deaths occurred in this series related to surgery, 15% of patients experienced subjective or objective worsening of preexisting symptoms or new symptoms (usually sphincter dysfunction) at discharge. For the majority, this worsening was transient. However, in 6% of patients, worsening of motor (2 patients) or sphincter (5 patients) function or both (2 patients) persisted at last follow-up despite imaging confirmation of complete and persistent fistula obliteration. A possible explanation for this worsening is that patients with an SDAVF are hemodynamically fragile and the balance between normal vascularization/drainage and ischemia/edema can be negatively affected by any increase in intra-abdominal pressure (such as during surgery in the prone position), which may further increase resistance to venous drainage.^31,32 This hypothesis may also explain why patients with an SDAVF always worsened after spine surgeries performed for indications other than treatment of the fistula. Correct positioning, avoiding increasing intra-abdominal pressure, and minimizing operative time may be critical to achieve a good functional outcome. Similar to other series, the incidence of other (non-neurological) complications was very low.

As occurs in other forms of myelopathy (eg, spinal cord injury), we and others³³ have observed a delayed deterioration in the absence of fistula recurrence on repeat neuroimaging studies. These cases may be explained by increased spasticity and dysesthetic pain caused by the chronic changes of spinal cord damage. In fact, delayed functional decline was often observed in the setting of intercurrent diseases or conditions that forced patients to have a period of decreased physical activity. This observation suggests that patients with a previously treated SDAVF have a lower functional reserve with decreased capacity to compensate for the functional loss that occurs with intercurrent diseases or even with normal aging.

Endovascular Treatment

Traditionally, surgery has been considered the treatment of choice for SDAVFs. However, with the advances in endovascular therapy, embolization has been proposed as an alternative to surgical treatment by a number of centers^17,34,35 (endovascular curative embolization of SDAVFs is an off-label application of liquid embolic agents in the United States). Earlier experiences with endovascular treatment were associated with almost universal recurrence because of limitations of the materials used.^36–38 With the availability of liquid embolic agents that can be pushed well into the draining vein, better and more permanent results have been reported. Still, recurrences are not uncommon.^16,18,19 Newer embolic agents may improve endovascular results, but more studies and longer term follow-up are necessary.³⁹ Embolization may potentially be a valid therapeutic approach, but the results of endovascular treatment must match the results obtained with surgery before embolization can be considered the therapeutic modality of choice. Given the risk of recurrence even after successful embolization, patients should be studied early (within 2-3 weeks) after “successful” embolization, and surgical exploration must be pursued immediately if postembolization studies suggest residual/recurrent fistula, especially in the absence of clinical improvement after embolization.

CONCLUSION

Despite modern imaging techniques, early diagnosis of SDAVFs continues to be challenging. Surgical obliteration of SDAVFs is an effective and safe procedure. After successful surgical obliteration, motor symptoms improve in more than 80% of patients. Other symptoms such as numbness, sphincter dysfunction, and dysesthesias/neuropathic pain improve in one third to one half of patients.

ABBREVIATIONS:

• MRA

  magnetic resonance angiography

• SDAVF

  spinal dural arteriovenous fistula

COMMENTS

The results of this largest series to date of surgically treated spinal dural arteriovenous fistulae (SDAVFs) are impressive. The authors ought to be congratulated on this thoughtful and honest retrospective study with complete and long follow-up. I am especially impressed by the reliability and accuracy of their magnetic resonance imaging (MRI) and magnetic resonance angiography (MRA) for the diagnosis, the sensitivity of catheter angiography in depicting the fistula at the first attempt in 150 of 154 patients, and the success rate of 95% of surgically obliterating the fistula at the first attempt, the more while many patients were treated in an era without sophisticated MRA and angiographic techniques. As in other studies, clinical improvement after successful obliteration of the fistula is variable, with most improvement in muscle strength.

An important and for me new finding was the relatively high rate of 15% of patients with postoperative worsening of symptoms or new symptoms, persisting during follow-up in 6%. I can hardly imagine that an increase in abdominal pressure as a result of the prone position during surgery can be held responsible for this postoperative deterioration. Finally, the authors mention the recent use of 3-dimensional (3D) computed tomography (CT) during selective angiography. Apparently, the patient is transported to the CT scanner with a selective catheter in place. This surprises me because over 10 years, excellent images can be obtained by 3D rotational angiography during catheter angiography.

In this series, surgery was the only treatment offered. Embolization with glue was not considered because in the United States, the use of acrylic glue for SDAVFs is off-label. Management of SDAVFs in our institution is quite different: when an SDAVF is diagnosed on MRI, we now perform high-resolution MRA at 3 T to assess the level of origin of the fistula. Then, spinal angiography is performed, and the level of the fistula is imaged in 3D rotational angiography. With 3D rotational angiography, the anatomy of the feeding vessels and the relationship with the surrounding bony structures can be studied from every direction, and the possibility of embolization can be assessed. When vessel geometry is considered favorable for embolization (no extremely sharp curves in the supplying arteries and an artery diameter suitable to accommodate a microcatheter), embolization with acrylic glue is immediately performed. In our practice, about 40% to 50% of SDAVFs are embolized. Embolization is considered successful when glue has penetrated and occluded the proximal draining vein.¹ Embolization is often straightforward and very quick; when MRA correctly identifies the level of the fistula, angiography and embolization are mainly accomplished within an hour without the need for general anesthesia. When vessel geometry does not favor embolization or when embolization is unsuccessful with no penetration of glue into the draining vein, we proceed to surgery. As a rule, in all cases for surgery, a short platinum coil is placed in the feeding pedicle close to the fistula to help the surgeon localize the correct level on fluoroscopy. After treatment, patients are followed clinically and with MRI/MRA. Since 1994, we have treated 64 patients with SDAVFs. Despite our optimal logistics, our results are not as good as those of Saladino et al.^2,3 In some patients, diagnosis was difficult. In 9 patients, repeat treatments were necessary. We had 3 patients with 2 SDAVFs; the second one was detected too long after treatment of the first one in 2 of 3 patients. Two patients with sacral SDAVFs underwent embolization and were operated on several times. In 2 patients, a new SDAVF developed at a different level after 7 and 9 years. Thus, in our hands, diagnosis and both endovascular and surgical treatment was not always straightforward.

Despite the excellent results that are obtained with surgery as reported in this article, in my opinion, endovascular embolization and surgery are complementary treatments in the management of patients with SDAVFs. Diagnosis is established with MRI. Modern 3-T MRA is mainly capable of assessing the level of the fistula,⁴ and catheter angiography can be restricted to several pedicles. With 3D rotational angiography, the anatomy of the SDAVF can be precisely evaluated, and a sensible treatment decision can be made. When embolization is possible and successful with occlusion of the draining vein, the patient is quickly and definitely cured. In all other cases, surgery should follow.

Willem Jan van Rooij

Tilburg, The Netherlands

This is a landmark article on the surgical results of the treatment of SDAVFs in 154 consecutive patients. The report includes a careful analysis of the complications of treatment, and clinical follow-up was available for 94% of patients. Worsening of preoperative symptoms or new symptoms developed immediately after surgery in 15.6% of patients, and 6% did not return to baseline at discharge. In the literature, there is no comparable series looking at the results of embolization, and the data on early postembolization worsening are not clear. Cure rates from embolization vary from 25% to 90% in highly selected series.^1–3 The authors appropriately state “endovascular treatment must match the results obtained with surgery before embolization can be considered the therapeutic modality of choice.” At this point, we cannot conclude that embolization is as efficacious as surgery. Early reports on embolization had high recanalization rates because of the use of particles or the failure to appreciate that closure of the draining vein is mandatory.

At our own institution, selective spinal angiography is done in all patients suspected of having an SDAVF. In most patients, we are able to predict the level of the fistula before the catheter angiogram using gadolinium-enhanced MRA.⁴ The catheter angiogram is critical to demonstrate the angioarchitecture of the fistula because spinal arteries can arise from the same pedicle as the fistula and a spinal cord arteriovenous fistula may mimic an SDAVF on noninvasive imaging. To obtain high-quality studies, we perform all spinal angiography with the patient under general anesthesia. Our current strategy is to attempt embolization at the time of the diagnostic catheter angiogram if the anatomy is suitable. If we do not achieve a definite occlusion of the draining vein, surgery is done during the same hospital admission. To date, we have not had objective or subjective worsening of symptoms post-embolization, but our cure rate is low (approximately 25% of patients).¹ To continue using our current approach, we need to see embolization outcomes comparable to the surgical results presented here by Saladino et al.

Robert A. Willinsky

Toronto, Ontario, Canada

1.

Jellema K, Sluzewski M, van Rooij WJ, Tijssen CC, Beute GN. Embolization of spinal dural arteriovenous fistulas: importance of occlusion of the draining vein. J NeurosurgSpine. 2005;2(5):580–583.

2.

Jellema K, Tijssen CC, van Rooij WJ, et al. Spinal dural arteriovenous fistulas: long-term follow-up of 44 treated patients. Neurology. 2004;62(10):1839–1841.

3.

Jellema K, Canta LR, Tijssen CC, van Rooij WJ, Koudstaal PJ, van Gijn J. Spinal dural arteriovenous fistulas: clinical features in 80 patients. J Neurol Neurosurg Psychiatry. 2003;74(10):1438–1440.

4.

Backes WH, Nijenhuis RJ. Advances in spinal cord MR angiography. AJNR Am JNeuroradiol. 2008;29(4):619–631.

1.

Van Dijk JMC, terBrugge KG, Willinsky RA, et al. Multidisciplinary management of spinal dural arteriovenous fistulas: clinical presentation and long-term follow-up in 49 patients. Stroke. 2002;33(6):1578–1583.

2.

Song JK, Gobin YP, Duckwiler GR, et al. N-butyl 2-cyanoacrylate embolization of spinal dural arteriovenous fistulae. AJNR Am J Neuroradiol. 2001;22(1):40–47.

3.

Sherif C, Gruber A, Bavinzski G, et al. Long-term outcome of a multidisciplinary concept of spinal dural arteriovenous fistulae treatment. Neuroradiology. 2008;50(1):67–74.

4.

Farb RI, Kim JK, Willinsky RA, et al. Spinal dural arteriovenous fistula localization with a technique of first-pass gadolinium-enhanced MR angiography: initial experience. Radiology. 2002;222(3):843–850.

FIGURE. Caption not available.