Article Text
PDF
Abstract
Background/objective Spinal dural arteriovenous fistulas (SDAVFs) require pretreatment angiography; embolization can be performed in the same session. To validate this approach, obliteration and morbidity rates of ‘endovascular-first’ (embolization and microsurgery in the case of embolization failures) must be compared with rates for ‘microsurgery-first’ (microsurgical ligation without attempted embolization) approaches.
Methods We reviewed our institutional database (January 1998–October 2015) for SDAVFs, performing an intention-to-treat analysis comparing endovascular-first and microsurgery-first approaches.
Results A total of 71 patients underwent surgical and/or endovascular treatment for SDAVFs. All SDAVFs were ultimately occluded. Of 35 patients under consideration for an endovascular-first approach, radicular artery anatomy or anterior spinal artery embolization risk precluded attempting embolization in seven cases (20%). Among 28 patients undergoing embolization, angiographic non-opacification of the fistula was noted in 18 (64%). Fourteen patients had obliteration with excellent casting of the draining vein (50%) and did not undergo surgery. There were no significant differences in total complications (9% vs 11%; p=1.0) or permanent complications (3% vs 4%; p=1.0) after attempted endovascular and surgical treatment. Based on an intention-to-treat analysis, there were no significant differences in total complications (11% vs 14%; p=1.0), permanent complications (6% vs 3%; p=0.61), or the symptomatic resolution/improvement rate (80% vs 78%; p=1.0) between endovascular-first and microsurgery-first groups.
Conclusions Our results support attempted embolization of SDAVFs prior to consideration of microsurgery, allowing for a less invasive treatment option in the same session as diagnostic angiography.
- Fistula
- Liquid Embolic Material
- Spine
- Vascular Malformation
Statistics from Altmetric.com
Introduction
Spinal dural arteriovenous fistulas (SDAVFs) result in myelopathy due to venous hypertension.1–4 Since the importance of mere fistula site disconnection was recognized,5 surgical approaches have been highly successful in obliterating these lesions.2–4 As pretreatment angiography remains a requisite diagnostic modality prior to fistula occlusion, embolization approaches have also evolved.6–8 The utility of this approach is limited to selected cases where embolization does not place the anterior spinal artery at risk and casting of the draining vein is feasible from the arterial feeding pedicle.7 Although publications on endovascular treatment have generally attempted to illustrate the robustness of this approach,7 ,8 few have presented evaluations of the frequency of embolization feasibility and the rate of angiographic occlusion with vein casting. Furthermore, given the high success rate of surgical obliteration, the question arises whether endovascular embolization should be considered.4 To evaluate attempted embolization as a primary approach, an intention-to-treat (ITT) analysis is necessary to compare rates of obliteration, complications, and symptomatic resolution between an ‘endovascular-first’ and ‘microsurgery-first’ treatment approach. In this paper we analyze our endovascular experience to first evaluate the overall rate of feasibility of embolization. We then compare a cohort of patients with SDAVF who were considered first for embolization with a cohort who were managed with an initial intent of surgical obliteration to determine whether an endovascular-first approach is an effective and safe treatment option.
Methods
We performed a retrospective analysis of our institutional vascular database from January 1998 to October 2015 to identify patients with SDAVFs. We excluded patients with cranial, epidural, pial, and conus arteriovenous fistulas or malformations. We extracted demographic information, symptoms at the time of clinical presentation, treatment approach, angiographic results, complications, and follow-up clinical and radiographic outcomes. The decision to treat via surgical disconnection or embolization was based on referral pattern; patients initially referred to the endovascular service were explored for potential embolization while those initially referred to the surgical service were often managed with a microsurgery-first approach. Both embolization and microsurgical ligation were performed under general endotracheal anesthesia with neurophysiologic monitoring (motor and somatosensory evoked potentials). Surgical disconnection was performed intradurally after laminectomy and a standard midline durotomy. Embolization was performed with liquid embolic agents (n-butyl cyanoacrylate (NBCA) or Onyx; Covidien, Dublin, Ireland). After embolization, unsubtracted images were scrutinized for effective vein casting with embolysate (figure 1). Patients without effective vein casting were then referred by our endovascular service for surgery, except for one patient who underwent successful repeat embolization. Our results were tallied and compared between patients considered first for embolization and those referred directly for surgery in an ITT analysis. Categorical variable comparison was performed using the Fisher exact test; p values <0.05 were considered statistically significant.
Vein casting. Anteroposterior view of a T8 segmental artery injection demonstrating a spinal dural arteriovenous fistula (SDAVF) (A). Oblique view of superselective angiography of the T8 radicular artery branch supplying the SDAVF (B). After n-butyl cyanoacrylate (NBCA) embolization, a plain film demonstrates vein casting by the glue (C); no residual SDAVF was seen on control angiography (D). Used with permission from Barrow Neurological Institute, Phoenix, Arizona.
Results
A total of 71 patients underwent surgical disconnection and/or endovascular embolization of their SDAVFs during the evaluated period (table 1). Mean±SD patient age was 63±12 years and 77% of the patients were men. All but two patients had symptomatic SDAVFs (97%). Fistula location was thoracic in 63% of cases, lumbar in 24%, sacral in 7%, and cervical in 6%. Preoperative magnetic resonance angiography was performed in 26 cases. Fistula location was correctly identified in 73% of cases, in 19% of cases it was non-diagnostic, and in 8% of cases it identified an incorrect level for the fistula.
Demographics, lesion location, and complications for all patients, those undergoing surgery and those under consideration for embolization*
All SDAVFs were ultimately obliterated following either endovascular or surgical treatment. The overall complication rate was 13%; permanent complications occurred in 4% of cases. After a mean clinical follow-up of 16 months, the condition of 79% of patients was improved, 15% were the same, and 6% were worse. Follow-up MRI was performed for 28 patients over a mean period of 29 months post-obliteration without evidence of fistula recurrence. Another 12 patients underwent delayed follow-up digital subtraction angiography over a mean period of 12 months post-obliteration without evidence of fistula recurrence.
Surgical disconnection
Surgical disconnection was performed in 56 patients. Of these patients, 36 were referred directly for initial surgery. An additional 20 patients who underwent surgery were under consideration for initial embolization; 7 were precluded from embolization attempts and 13 underwent initial embolization with either NBCA or Onyx but with inadequate penetration of the fistula point. There was no significant difference in patient age, sex, presentation modality, and fistula location between patients undergoing surgery, those under consideration for embolization or the entire patient cohort (table 1). The surgical obliteration rate was 100%. After a mean clinical follow-up period of 17 months, 80% of the 56 patients were improved, 14% were the same, and 5% were worse.
Overall there were six operative complications (11%). One patient was taken back to the operating room for evacuation of an epidural hematoma. The remaining five patients had operative wound complications (two for infection that required a return to the operating room and three for cerebrospinal fluid leak). One case of cerebrospinal fluid leak was managed successfully with lumboperitoneal shunting, one patient suffered permanent sequelae from meningitis and arachnoiditis and one patient suffered permanent sequelae from meningitis and an epidural abscess (4% (2/56) permanent complication rate).
Embolization
Embolization was considered as the initial treatment for 35 patients. There was no significant difference in patient age, sex, presentation modality, and fistula location between patients under consideration for embolization, those treated surgically or the entire patient cohort (table 1). In seven of the 35 cases the arterial anatomy precluded attempted embolization despite attempted microcatheterization (20%). Specifically, in five patients adequate purchase into the supplying radicular branch could not be achieved. In two others, embolization was not attempted due to a risk of anterior spinal artery embolization from the final microcatheter position; this is not a global insurmountable risk in well-selected cases, however (figure 2). Figure 3 summarizes treatment approaches with crossovers from the endovascular cohort.
Cervical spinal dural arteriovenous fistula (SDAVF). Superselective anteroposterior angiography after catheterization of the radicular branch supplying this cervical SDAVF demonstrates opacification of the fistula as well as the anterior spinal artery and bilateral vertebral arteries (A). After adjustment of the microcatheter into another branch of the radicular branch, superselective anteroposterior angiography demonstrates opacification only of the fistula and no pial branches (B). The fistula was uneventfully embolized with n-butyl cyanoacrylate (NBCA) with penetration of the draining vein as seen on this anteroposterior unsubtracted film (C). This anteroposterior right subclavian injection demonstrates no residual arteriovenous shunting (D). Used with permission from Barrow Neurological Institute, Phoenix, Arizona.
Flowchart of patient treatment. Used with permission from Barrow Neurological Institute, Phoenix, Arizona.
In the remaining 28 cases, embolization was carried out with either Onyx (16/28, 57%) or NBCA (12/28, 43%). After embolization, 18 patients had no further opacification of the SDAVF on control angiography (64%). In 14 cases there was no residual opacification of the SDAVF with good casting of the draining vein with embolysate (50%). This corresponded to 67% of cases where NBCA was used and 38% of cases where Onyx was used (p=0.25). Three patients without adequate draining vein casting were referred for subsequent surgery along with the other 10 who continued to have angiographic opacification after embolization. One patient with inadequate vein casting was successfully retreated via Onyx embolization with successful casting of the draining vein.
There were three complications among the 35 patients considered for embolization (9%); this corresponds to 11% (3/28) of patients who underwent embolization with Onyx or NBCA. There were two cases of radicular artery rupture requiring immediate treatment with NBCA. One of these two cases had no sequelae after treatment; the other patient had permanent worsening of her lower extremity strength. The third complication was in a patient who experienced transient worsened numbness after treatment. The overall (9%) and permanent (3%) complication rates for endovascular treatment did not differ significantly from the respective complication rates after all surgeries (p=1.0). The rate of symptomatic improvement at clinical follow-up also did not differ between actual endovascular treatment and all surgery (both 80%).
Intention-to-treat analysis
Comparing our two cohorts in our ITT analysis of endovascular-first (n=35) and microsurgery-first (n=36) approaches, there was no significant difference in patient age, sex, fistula location, or presentation modality (table 1). All fistulas were ultimately obliterated and there was no significant difference in overall complications (11% vs 14%; p=1.0) or permanent complications (6% vs 3%, p=0.61) between the endovascular ITT and the surgery ITT cohorts, respectively. There was also no significant difference between rates of symptomatic improvement between the endovascular ITT and surgery ITT cohorts (80% vs 78%, respectively; p>0.99).
Discussion
SDAVFs, also known as type 12–4 or intradural dorsal1 arteriovenous shunts, are supplied by extradural radicular arteries with drainage into an intradural vein. As seen in our series, these lesions have a known proclivity to occur in the thoracic spine and are associated with male gender; the vast majority of patients present with sequelae of spinal venous hypertension.2–4 ,8 Prior to a clear understanding of the anatomy and pathophysiology of SDAVFs, decades ago most SDAVFs were treated with multilevel laminectomies and vein stripping.9 With pioneering work by Djindjian and colleagues in spinal angiography and embolization,10 our anatomic understanding of these lesions improved along with our broadened endovascular options. The introduction of cyanoacrylates as permanent embolysates by Kerber and colleagues was a particularly crucial advance.11 Kendall and Logue emphasized the key pathophysiologic insight that vein disconnection is a curative approach for these lesions,5 which catalyzed the ongoing cascade of series describing successful microsurgical disconnection.2–4
Much of the literature on SDAVFs has emphasized the relative robustness of surgical disconnection as a means for definitive obliteration.2–4 Nevertheless, systematic reviews and meta-analyses may fail to capture the nuances of a particular treatment modality, such as operative wound infections or factors that may not directly impact long-term neurological function. The luxury of a single institutional review allows us to evaluate potential advantageous and adverse factors related to either the surgical or endovascular treatment of these lesions. As our results emphasize, surgical obliteration is associated with exceptional obliteration rates. However, as spinal magnetic resonance angiography remains an imperfect modality for evaluation of the fistula site, we remain dependent on the performance of a diagnostic spinal angiogram.
Importantly, we demonstrate that, overall, embolization can be attempted in approximately 80% of cases. When performed, we achieved a 50% rate of obliteration with casting of the draining vein. We do not consider angiographic non-opacification without casting of the draining vein adequate occlusion and refer these patients on for surgery given the known high rate of recurrence.12 Interestingly, as has been noted in another recent series,12 we saw a trend toward higher rates of occlusion with casting of the draining vein with NBCA usage compared with Onyx (67% vs 38%, respectively, p=0.25). This result may be confounded in part by the proclivity to use NBCA with closer better microcatheter positioning relative to the fistula point.
Most series dwell on greater obliteration rates after surgical disconnection of SDAVFs.3 ,4 If patients are presented with the option of one approach with a high obliteration rate and another with a low obliteration rate and generally similar complication rates, most would opt for that with a high obliteration rate (surgery). However, this ignores the practical question of whether to attempt embolization during requisite preoperative spinal angiography; the patient already has a sheath in his or her femoral artery, a catheter in the parent segmental artery, and an immediate option for potential therapeutic cure. This study demonstrates that approximately half of patients who can undergo embolization need not go on for an additional procedure (surgical disconnection). Importantly, the ITT analysis illustrates that the collective group of embolization successes and embolization failures who go on to surgery had similar overall outcomes to the group treated empirically with surgery after diagnostic angiography. Specifically, overall complications and follow-up neurological condition were comparable between patients considered for endovascular treatment first and those managed initially with microsurgery without attempted embolization. Armed with this information, from the perspective of informed consent, we suspect most patients would opt for attempted endovascular embolization in the same setting as their diagnostic angiogram, knowing they have an approximately 50% chance of not requiring an additional more invasive surgical procedure. Although we do not typically perform routine follow-up angiography in patients with improving symptoms, future studies may more carefully scrutinize the rate of long-term angiographic and clinical recurrence.
Conclusion
In approximately 80% of cases, embolization of SDAVFs may be attempted with an associated 50% rate of angiographic occlusion with glue casting of the draining vein. Compared with direct referral for microsurgical ligation, attempted endovascular embolization with or without subsequent microsurgical disconnection as needed has comparable rates of angiographic obliteration, complications, and clinical improvement at follow-up.
References
Footnotes
Contributors Conception and design: BAG, FCA, KM and CGM. Drafting the article: BAG. Data acquisition, interpretation and analysis: BAG, FCA, KM and CGM. Critical revision of the article: BAG, FCA, KM and CGM. Statistical analysis: BAG. Study supervision: FCA, CGM. The authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.