Article Text
Abstract
Objective Carotid cavernous fistula (CCF) development after Pipeline Embolization Device (PED) treatment of cavernous carotid aneurysms (CCA) can be a challenging pathology to treat for the neurointerventionalist.
Methods A database of all patients whose aneurysms were treated with the PED since its approval by the Food and Drug Administration in 2011 was retrospectively reviewed. Demographic information, aneurysm characteristics, treatment technique, antiplatelet regimen, and follow-up data were collected. A literature review of all papers that describe PED treatment of CCA was then completed.
Results A total of 44 patients with 45 CCAs were identified (38 women, 6 men). The mean age was 59.9±9.0 years. The mean maximal aneurysm diameter was 15.9±6.9 mm (mean neck 7.1±3.6 mm). A single PED was deployed in 32 patients, with two PEDs deployed in 10 patients and three PEDs in 3 patients. Adjunctive coiling was performed in 3 patients. Mean follow-up duration based on final imaging (MR angiography or digital subtraction angiography) was 14.1±12.2 months. Five patients (11.4%) developed CCFs in the post-procedural period after PED treatment, all within 2 weeks of device placement. These CCFs were treated with a balloon test occlusion followed by parent artery sacrifice. Our literature review yielded only three reports of CCFs after PED placement, with the largest series having a CCF rate of 2.3%.
Conclusions CCF formation is a known risk of PED treatment of CCA. Although transvenous embolization can be used for treating CCFs, parent artery sacrifice remains a viable option on the basis of these data. Studies support the view that adjunctive coiling may have a protective effect against post-PED CCF formation. None of the coiled aneurysms in our database or in the literature have ruptured. Follow-up data will lead to a better understanding of the safety profile of the PED for CCA.
- Aneurysm
Statistics from Altmetric.com
Introduction
Flow diversion with the Pipeline Embolization Device (PED; Covidien, Irvine, California, USA) is currently approved by the Food and Drug Administration for the treatment of large and giant wide-necked internal carotid artery (ICA) aneurysms, although the PED has become increasingly used for off-label treatment, both smaller anterior circulation aneurysms and posterior circulation aneurysms.1–4 Within the group of large ICA aneurysms, the subclass of cavernous carotid aneurysms (CCAs) confers a more benign natural history with a lower risk of morbidity and mortality associated with rupture.5 ,6 Specifically, they have a lower risk of subarachnoid hemorrhage but are more likely to create a direct carotid cavernous fistula (CCF) in the event of rupture, potentially resulting in ‘cavernous sinus syndrome’, causing cranial nerve (CN) palsy, increased intraocular pressure, proptosis, and chemosis. The long-term risks associated with PED treatment of intradural and extradural ICA aneurysms remain an area of debate and research.7 ,8 While delayed hemorrhagic and embolic complications following flow diversion have been well studied,9–12 CCFs following CCA treatment with the PED have not been commonly published.13 ,14 The management of CCFs after PED treatment of CCAs at a high-volume cerebrovascular tertiary care center is reported and the relevant literature reviewed.
Methods
A prospectively maintained single-center database of patients whose aneurysms were treated with the PED at a high-volume cerebrovascular referral center was retrospectively reviewed. All patients treated for CCAs were then collected. Demographic information, aneurysm characteristics, treatment details, antiplatelet regimen, and follow-up data were extracted from the medical record. Peri- and post-procedural complications were recorded. A literature review of all papers that describe PED treatment of CCAs was then completed.
Results
A total of 44 patients (38 women, 6 men) with 45 treated CCAs were identified from the database of patients who underwent PED placement. One patient had bilateral aneurysms treated at separate procedures, which are included as individual aneurysm treatment entries. Follow-up data listed below are per treated aneurysm. The mean age of the group was 59.9±9.0 years. The mean maximal diameter of treated aneurysms was 15.9±6.9 mm and the mean neck size was 7.1±3.6 mm. A single PED was deployed to treat 32 aneurysms, with two PEDs deployed for 10 aneurysms and three PEDs for three aneurysms. Adjunctive coiling was performed in three cases. Mean follow-up duration was 14.1±12.2 months based on the date of most recent post-procedural imaging (MR angiography (MRA) or digital subtraction angiography (DSA)). Follow-up data were available for 43 aneurysms (95.5%). Complete occlusion was achieved for 38 aneurysms (88.4%) while minimal filling of the neck was present in five aneurysms (11.6%).
Five patients (11.4%) developed CCFs in the post-procedural period after PED treatment, all of which occurred within 2 weeks of device placement (table 1). With the exception of patient 1, all patients were treated with steroids after admission. Two patients (patients 3 and 4) did call us reporting worsening headaches a day prior to presentation. PED apposition was closely analyzed for these five patients and revealed good device placement. The mean fluoroscopy time in minutes for the patients with CCFs and non-ruptured aneurysms was 56.6±25.4 and 35.9±15.9, respectively. There was no statistical difference by the Mann-Whitney U test although a trend was seen (p=0.06). The mean maximal diameter for patients who developed CCFs was 20.1±6.3 mm and 15.5±6.9 mm for unruptured aneurysms. CCF formation was uniformly distributed across our years of experience with the PED. All CCFs were treated with a balloon test occlusion (BTO) followed by parent artery sacrifice. Two representative cases are discussed below.
Excluding CCFs, post-procedural neurological complications occurred in three patients (6.8%, table 2). In addition, device-related complications without any associated neurological morbidity were noted in three patients (6.8%). Clot was noted on the PED after deployment in two cases, which resolved after a bolus of abciximab. The other was a distal migration of the PED which required coil sacrifice of the parent vessel.
Illustrative cases
Patient 4 presented to a referring hospital with headaches and was found to have a right middle cerebral artery aneurysm and a 19×16×13 mm CCA (figure 1A). The patient initially underwent uneventful clipping of the right middle cerebral artery aneurysm and the CCA was observed. Approximately 6 weeks after clipping the patient developed double vision and a right CN VI palsy. CT angiography (CTA) did not reveal any changes in the cavernous segment aneurysm. Dual antiplatelet therapy with aspirin and clopidogrel was initiated and a PED was placed (figure 1B). On post-procedure day (PPD) 6 the patient presented with worsening headaches, nausea and diplopia, and cerebral angiography revealed a new CCF (figure 1C). Transvenous embolization was initially attempted via the right internal jugular vein. Access to the inferior petrosal sinus and the cavernous sinus was obtained but the fistula could not be catheterized (figure 1D). The patient subsequently underwent a BTO followed by parent artery sacrifice (figure 1E). The diplopia persisted 1 month after treatment but her proptosis and chemosis significantly improved.
Patient 5 presented to the clinic with a left partial CN III palsy and was diagnosed with an 18.8×10.2×12.6 mm left CCA (figure 2A). The patient was initiated on dual antiplatelet therapy with aspirin and clopidogrel and a single PED was placed to treat the aneurysm (figure 2B). On PPD 7 the patient presented with a worsening CN III palsy, headaches, and diplopia. Follow-up CTA and MRI revealed a left CCF (figure 2C). The patient underwent a successful BTO and subsequent parent artery sacrifice (figure 2D). By the 6-week follow-up visit the patient's left CN III palsy had significantly improved.
Discussion
This report details a single high-volume center experience wherein the PED was used to treat CCAs and the subsequent management of CCFs that resulted as a complication of the treatment. CCFs following PED placement are type A CCFs in the Barrow classification scheme (high-flow direct fistulas).15 ,16 These cases can be a challenging complication for a neurointerventionalist to manage. The transarterial route to treat these fistulas is not a viable option if the goal is to maintain ICA patency, given the presence of the PED in the cavernous carotid segment. The only arterial treatment option is vessel sacrifice of the ICA. Certainly, careful preoperative evaluation with a BTO is required prior to ICA sacrifice. In a large case series of 88 patients treated with ICA occlusion for aneurysms, Bechan et al17 reported three subclinical infarcts and three patients with a transient hemiparesis, all of which were in the watershed territory between the anterior and middle cerebral artery. The only permanent complication in their series was a hypoperfusion infarction due to a retroperitoneal hematoma. Despite adequate tolerance of a BTO, patients may still suffer significant morbidity following parent artery sacrifice. Contralateral flow-related aneurysm formation has been documented after carotid artery sacrifice.18 In our series, all patients underwent a BTO followed by a nuclear medicine study to assess for any asymmetry in radiotracer activity as a surrogate for subclinical hypoperfusion. All patients in this series have thus far tolerated parent artery sacrifice and long-term follow-up is ongoing.
CCFs may also be treated by transvenous embolization. Transvenous embolization of CCFs requires access to the cavernous sinus usually via the inferior petrosal sinus. In select cases, access to the cavernous sinus via surgical access to the superior ophthalmic vein or direct transorbital puncture through the superior orbital fissure can also be attempted.19 Tanweer et al13 treated one CCF with transvenous access via the ipsilateral facial vein, superior ophthalmic vein to the cavernous sinus. Generally, transvenous embolization more effectively treats indirect CCFs rather than direct CCFs. However, successful transvenous treatment of direct CCFs following flow diversion for CCAs has been described. Lin et al14 recently reported two such cases, in both of whom coils were successfully deposited into the cavernous sinus, through the fistula and into the aneurysm. In one case the coil loops migrated into the aneurysm sac and in the second case they were able to directly access the CCA through the rupture site. Mustafa et al20 published a case report of transvenous embolization of a CCF following flow diversion with the SILK device (Balt Extrusion; Montmorency, France). In our series, transvenous embolization was attempted in two cases but could not be successfully accomplished. Given that direct CCFs are challenging to definitively treat transvenously, BTO followed by parent artery sacrifice has become the primary treatment modality in our center.
An exhaustive review of the literature yielded only three publications that discuss CCFs after PED placement, with the largest series having a CCF rate of 2.3% (table 3).13 Tanweer et al13 reviewed their experience of CCA treatment with the PED and reported one CCF among a group of 41 patients with 43 treated CCAs at 6-month follow-up. The fistula was incidentally discovered and treated transvenously, as described above. Clinically, the patient only had mild scleral injection. However, the patient remembered having pulsatile tinnitus on PPD 2 which he did not report. Lin et al14 reported two CCFs from CCAs treated with the PED that were successfully treated via the transvenous route as discussed earlier. The first patient was noted to have worsening diplopia at his 6-week follow-up, with angiography revealing a CCF with retrograde cortical venous drainage. The second patient was noted to have described a humming sound on PPD 3 but CTA did not reveal an underlying fistula. At 1 month the patient reported an increased intensity of pulsatile tinnitus with angiography revealing a CCF. To the best of our knowledge, these are the only three reports of CCFs specifically after PED treatment of a CCA.
Chan et al21 reported on a CCA that was treated with a Neuroform stent and Axium coils. The patient presented at 1 month postoperatively with proptosis, chemosis, and ophthalmoplegia and angiography revealed a CCF. The patient had cessation of her symptoms over the next few days after being started on steroids and a follow-up angiogram revealed resolution of the CCF. At 1 year the patient had recurrence of the aneurysm, which was treated by stent-assisted coiling. The patient mentioned earlier in the paper by Mustafa et al20 presented 2 weeks postoperatively with ptosis following minor trauma, with angiography revealing a CCF. Kulcsár et al22 reported a retrospective analysis from 12 centers with 13 cases of delayed post-procedural aneurysm rupture with the use of the SILK flow diverter. Two of the cases were CCAs leading to CCFs. Although a mean treatment to rupture time of 16 days is mentioned, specific numbers and treatments for the CCFs are not given.
Siddiqui et al23 suggested in one of the early reports on complications associated with PED deployment that coil placement into the dome of the aneurysm may protect against delayed rupture. Lin et al24 reviewed their experience of 29 patients treated with adjunctive coiling versus 75 patients with the PED alone and reported a higher aneurysm occlusion rate in the coiled group with similar rates of complications. In their early experience with the PED, Szikora et al2 performed adjunctive coiling in 10 of 19 aneurysms with only one reported rupture that was felt to be from a distal aneurysm. The recent IntrePED25 database results of aneurysms that underwent adjunctive coiling revealed no ruptures in the group that underwent coiling versus five ruptures in the PED alone group. While the difference is not statistically significant, it is an important one to note. The addition of coiling does significantly increase procedure time and often requires a larger groin sheath as reported in the IntrePED database, but we are now prospectively coiling the majority of our PED treatments for large cavernous aneurysms which will allow for some crucial future observations if the rate of CCF formation is reduced.
The timeline of development of CCFs after PED placement seems to favor early development. In our five cases and the other three reported in the literature, all complications arose within 6 months of device placement. In all patients in our series, CCFs were discovered within 2 weeks of device placement. The cause of this early formation is not completely understood, although several studies have now been completed on the causes of delayed aneurysm rupture after PED treatment. Early studies by Cebral et al26 reported that increases in intra-aneurysmal pressure following PED treatment could be the cause of delayed rupture. However, later studies have now postulated a more multifactorial etiology. Kulcsár et al22 have suggested that SILK flow diverter implantation may in some cases lead to an intra-aneurysmal flow pattern which could lead to rupture. It has also been discussed that, in some cases, thrombosis within the aneurysm leads to increased autolysis and wall rupture, although it is unclear why in most cases thrombosis does not lead to this state. Larrabide et al27 conducted computational flow dynamic analysis with supraclinoid ICA aneurysms and flow diverter implantation and did not find changes in mean intra-aneurysmal pressure but, instead, found reductions in the intra-aneurysmal pressure range. Kerl et al28 investigated hemodynamic parameters in silicone models of cerebral aneurysms into which the PED was deployed and also did not find any significant changes in static intra-aneurysmal pressure. They also suggest that thrombus-associated autolysis may be the cause of delayed rupture.
Our overall CCF rate (11.4%) is higher than that reported by Tanweer et al13 (2.4%). We investigated whether inadequate apposition of the PED could be contributing to the higher rate of CCFs, although a closer analysis of PED placement in these patients demonstrated good placement of the PED (figures 3⇓⇓⇓–7). We did not find any significant differences in mean fluoroscopy times between the two groups, although a trend was seen towards increased fluoroscopy time in patients with CCFs. While these could reflect increased complexity of the CCA that led to CCF formation, given the overall limited numbers we cannot reliably conclude this. A contributing factor to the difference in rates could be the number of PEDs used per aneurysm. On average, we deployed 1.36 PEDs per aneurysm versus 3.8 PEDs per aneurysm by Tanweer et al.13 A protective effect from flow pattern changes with more coverage of the aneurysm with increasing numbers of PEDs is postulated; however, the recent IntrePED database9 has also discussed intraparenchymal hemorrhages being associated with multiple PEDs. Our group ascribes to the principle that treating an aneurysm with as few and as short a device as possible reduces both thromboembolic and hemorrhagic complications associated with PEDs. Given such a treatment paradigm in association with CCFs associated with the PED for CCAs, we have changed our treatment approach to simultaneously treating CCAs with both PED and coils if the aneurysm is >15 mm. As a general principle, we attempt to obtain coil packing density of one-third to one-half the amount typically achieved for complete embolization of an aneurysm treated without flow diversion, in order to embolize the aneurysm without adding to the mass effect of the aneurysm, which is often causing neurologic symptoms.
Conclusion
As PEDs have now been in use for over 5 years, we are beginning to obtain a better understanding of their long term-effects and safety profile. Although CCF development is uncommon, it is certainly not a rare entity after PED treatment of CCA. We have hypothesized that addition of coils along with placement of additional PEDs may have a protective effect and lead to reduced formation of CCF.
References
Footnotes
AKR and JAG are co-first authors.
Contributors AKR conducted literature searches, drafted the initial version of the paper and revised the later versions. JAG conducted initial data collection, provided the initial idea for the study and revised the later versions of the paper. JWO assisted in selecting appropriate images and revision of the draft. SLS assisted in initial data collection and revision of the draft. BMH, FUA, and FT assisted in revision of the draft. JED assisted in literature searches and revision of the draft. CMC provided the initial idea of the study, monitored overall data collection and revision of the drafts.
Competing interests None declared.
Ethics approval Ethics approval was obtained from Emory University Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.