Direct carotid cavernous fistula (DCCF) is a direct communication between the internal carotid artery (ICA) and cavernous sinus (CS). DCCF is caused by head and facial trauma or cavernous ICA aneurysm rupture; traumatic CCF (TCCF) accounts for approximately 75% of all direct CCF cases [30]. The arteriovenous shunt between ICA and CS leads to various signs and symptoms including exophthalmos, chemosis, decreased visual acuity, cranial nerve impairment, or life-threatening hemorrhage and epistaxis. However, these characteristics are similar to those of multiple head and facial trauma. Hence, the diagnosis of TCCF may be delayed.

At present, endovascular therapy is the mainstay treatment for TCCF. TCCF is primarily treated with parent artery occlusion, which is a deconstructive method and can be performed in patients with good collateral flow. Reconstructive methods and fistula occlusion with detachable balloons or coils are therapeutic options for failed balloon test occlusion. The use of covered stents, which is another reconstructive treatment option for direct CCF, with encouraging short-to-midterm clinical outcomes, has recently emerged. Hence, the present study aimed to present the diagnosis of TCCF and the use of covered stents as the first-line treatment for TCCF in a level 1 regional trauma center.

The procedures were technically successful in all 15 cases. After covered stent deployment with maximum multiple balloon inflation, endoleak still occurred in six patients (40%). Of these patients, five underwent additional endovascular treatment and one who had minimal endoleak underwent follow-up imaging. Additional fistula point occlusion was achieved using detachable coils (n=2), proximal bare stenting (n=1), stent-assisted coiling (n=1), and second covered stenting (n=1). Immediate angiographic outcomes after the procedure were total occlusion (n=12, 80%), near-total occlusion (n=2, 13%), and subtotal occlusion (n=1, 7%). All patients, except two who died due to multiple severe traumas, underwent follow-up imaging (range, 2 days to 4 years; mean, 15 months). Three patients presented with recurrence, which was completely occluded during the initial procedure. All patients with recurrence received additional treatments (transvenous CS occlusion, stent-assisted coiling via a distal endoleak point, and balloon angioplasty) and achieved complete occlusion. The patients underwent clinical follow-up (range, 2-108 months; mean, 32 months). During follow-up, the outcomes were mild, moderate, and severe in five, five, and three patients (modified Rankin Scale scores of 1-2, 3-4, and 5, respectively), respectively (Table 2).

The actual incidence of TCCF is not fully elucidated. Several studies have reported that the incidence ranges from 0.17% to 1.01% in patients with traumatic head and facial trauma [37,44] and is up to 4% in those with basal skull fractures [27]. TCCF could result in permanent neurologic impairment or even death if untreated. Thus, treatment is required to relieve symptoms and control massive bleeding. Nevertheless, the symptoms of TCCF may be masked in patients with head and facial trauma. Hence, its diagnosis could be delayed. TCCF should be considered, and patients with severe head and facial trauma must undergo imaging during the initial evaluation. In our trauma center, computed tomography angiography is performed based on the Denver criteria [11,17].

Although parent artery occlusion was performed in the early stages, the outcomes were not satisfactory. In other words, even if patients had good tolerance to balloon test occlusion, the ischemic complication rates ranged from 5% to 22% [3,21]. Then, a detachable balloon was introduced to prevent fistula formation without parental artery occlusion, and the success rate was 88-98% [7,10,15,20,23,24,29,31,35]. Hence, this method was considered the primary treatment. With the advent of endovascular techniques, the use of detachable coils combined with bare stents has been proven to be an excellent therapeutic strategy for the treatment of DCCF [1,4,19,28,29,33,39]. Although they are excellent therapeutic devices, they cannot be used in some cases, which include an extremely large orifice to cover with detachable balloons or coils and sharp osseous fragments within the sinus that cause premature balloon deflation. Moreover, these devices have disadvantages. That is, there is a risk of cranial nerve palsy caused by mass effect. Furthermore, parental artery occlusion or ischemic events caused by coil migration could occur [8,13,18,20,23,39]. Currently, the use of detachable balloons may also be limited by their availability [12].

A recent study used a covered stent in DCCF and reported the outcomes [2,5,12,16,25,26,38,40,41]. Covered stents can be used in some conditions in which detachable balloons and coils cannot be utilized. The reported success rate to date of detachable balloon, flow diverter and covered stent for occlusion of TCCF as primary treatment modality was 55-98% [8,14,20], 71% [42], and 75-100% [12,26,41], respectively and our results were similar with previous reports. Moreover, it has advantages in the concept of one-step and time-saving treatment in emergency situations. The Graftmaster coronary stent graft system described in this study was originally designed for the management of coronary perforations in relatively straight coronary vessels [2,34,41]. Therefore, there is a limit in the application of covered stents with a stiff delivery system to the tortuous cavernous ICA system, which causes endoleak. This condition was correlated with poor stent vessel apposition or size mismatch. Previously, endoleak was expected to disappear and was not actively treated. If endoleak persisted in follow-up imaging, additional second bare stent placement or CS occlusion was performed [2,5,12,16,25,26,38,40,41]. We completed the first procedure with total occlusion in all patients, except in three with tolerable remnant endoleak. On follow-up imaging, three patients who had complete occlusion during the initial procedures presented with CCF recurrence. According to a root cause analysis, one patient who had low volume status was admitted due to severe epistaxis after surgery for pituitary adenoma. Due to excessive blood loss, the ICA collapsed and the covered stent was positioned in the temporarily constricted vessel. After volume resuscitation, vasospasm was treated and size mismatch between the covered stent and the dilated vessel caused endoleak. In another case, endoleak was believed to be attributed to incomplete apposition between the covered stent and the vessel wall at the proximal end. According to a previous study, the incidence of endoleak varies from 9% to 83% [16,25,26,34,40].

To reduce the incidence rate of endoleak without destructive procedure such as CS occlusion, some techniques including balloon angioplasty and use of longer and oversized covered stents were applied [16,25,26,40]. If the endoleak persisted in such attempts, we sealed the endoleak with a microcatheter via the leakage point and occluded it with detachable coils. Herein, we reported four cases of fistula point microselection and coil occlusion for endoleak, three of which were performed at the time of primary covered stent placement and one during the subsequent angiography after the placement of a covered stent. There might be limitations in vessel wall apposition due to shape and size mismatch, particularly at the proximal and distal ends, which usually cause type 1 endoleak [16,43].

Endovascular aortic aneurysm repair for anatomically suitable abdominal aortic aneurysms has gained wide acceptance in the past decade [9], and some studies have performed prophylactic proximal bare stent placement in hostile necks because patients with such anatomy are at an increased risk of proximal endoleak (type 1a) and migration, resulting in adverse events in up to 70% of cases [6,22,32]. We believe that prophylactic bare stent placement might also be applied to an intracranial covered stent. One patient with recurrent symptoms experienced complete occlusion during the initial procedure. However, the final angiographic imaging showed proximal end mismatch. Thus, endoleak was likely to occur, which could have been prevented with a prophylactic bare stent (patient number 13). The placement of prophylactic bare stent could maximize apposition of the covered stent to the vessel wall, and the use of a prophylactic bare stent with or without endoleak during the initial procedure is believed to reduce the rate of postprocedural endoleak. The limitations of present study include its retrospective nature, the recruitment of a limited number of patients, and relatively short follow-up duration. However, we presented indications of covered stent as a first-line treatment for TCCF and stent-based intracranial artery reconstruction could become the first-line treatment modality for TCCF.

TCCF is fatal and life-threatening; therefore, it must be considered and diagnosed in patients with severe head and facial trauma. Covered stent is the most reconstructive method, with a high total occlusion rate. Although endoleak has been the primary concern in the use of a covered stent, it could be managed with detachable coils at the leakage point or prophylactic bare stent in specific cases. Covered stents could be considered as the first-line treatment for TCCF. However, long-term follow-up and larger-scale studies are still warranted to further develop the specifications and indications of such device.