Introduction

Intracranial stent angioplasty is relatively easy and safe for well-selected cases and medically unstable or elderly patients can be treated under local anesthesia; however, it harbors many possible complications [5, 7, 13, 19, 20]. In this report, the authors present a case of direct carotid-cavernous fistula (DCCF) by arterial injury during stent angioplasty and its endovascular treatment. Here, the probable causes of DCCF and its treatment methods are discussed.

Case report

An 81-year-old man was admitted complaining of grade IV right hemiparesis with motor dysphasia. These symptoms had fluctuated over the past 2 weeks. In addition, the patient had a past medical history significant for hypertension, hyperlipidemia, diabetes mellitus, and chronic obstructive pulmonary disease. Brain magnetic resonance imaging (MRI) showed acute hemodynamic infarction scattered in the left cerebral hemisphere and perfusion studies showed delayed time to peak and decreased perfusion reservoir (Fig. 1a–c). Standard cerebral digital subtraction angiogram (DSA) revealed complete occlusion of left internal carotid artery (ICA) with collateral circulation from the right side through the anterior communicating (a-com) artery and moderate stenosis in the posterior genu of the right cavernous ICA (Fig. 1d, e). All of these symptoms and results of imaging studies indicated that this patient needed revascularization for the left cerebral hemisphere. As opposed to a bypass surgical procedure under general anesthesia, the authors attempted angioplastic dilatation of the stenotic segment of the right ICA in order to augment left cerebral hemisphere collateral flow through the a-com artery for this medically unstable patient.

Fig. 1
figure 1

a The perfusion computed tomography (CT) shows time-to-peak flow delay on left cerebral hemisphere, which is 0.69:1.46 compared with the right middle cerebral artery (MCA) territory. b Basal brain SPECT images. c Diamox-challenging SPECT images. These pictures indicate a decreased perfusion reservoir in the left cerebral hemisphere. Anteroposterior (d) and lateral (e) projection of right ICA angiography. Blood supply to the left cerebral hemisphere is totally dependent on the right ICA system (arrowheads in d). The cavernous segment of the right ICA is moderately stenotic (arrow in e)

Preoperative routine CT angiography showed multiple calcification of the right cavernous ICA in its two-dimensional multi-sliced images (Fig. 2). The authors obtained a measured diameter and length of the ICA to be covered by stent from CT angiogram and cerebral DSA as 4.5 mm and 13 mm, respectively. The procedure was performed according to the standard technique of stent angioplasty with caution not to injure the ICA of this old-age, tortuous and calcified lesion. Because of the tortuousness of the ICA, it was very difficult to advance the stent delivery catheter. Deployment of a balloon-mounted bare-metal DRIVER stent 4.5-mm wide and 15-mm long (Medtronic, Minn., USA) was followed by balloon dilatation. It was dilated to subnominal pressure as a 7-atm dilating stent to 4.34 mm. It was ballooned in a graded manner lasting about 15 s. During balloon dilation, the patient experienced a generalized tonic clonic seizure that lasted for approximately 10 s. Angiography confirmed proper positioning of the implant and sufficient widening of the stenotic lesion. However, in the early arterial phase of the angiogram, DCCF was found without cortical reflux (Fig. 3a). A follow-up brain perfusion CT the next morning showed aggravated time-to-peak flow delay in the left cerebral hemisphere, indicative of flow-steal by CCF (Fig. 3b). The patient’s neurological status was also deteriorating, and the authors decided to occlude the fistula.

Fig. 2
figure 2

Axial raw images of brain CT angiogram. Note the multiple calcifications through the whole length of the cavernous ICA (arrows). These calcifications can be the possible cause of dissection and rupture of the ICA during balloon inflation

Fig. 3
figure 3

a A DCCF occurred immediately after stent deployment. The fistula drains mainly into the inferior petrosal sinus and pterygoid plexus. There is no reflux to the superior ophthalmic vein or cortical veins. b More aggravated time-to-peak flow delay on left cerebral hemisphere and this indicates steal of flow by CCF

The exact fistulous site could not be located because of its fast shunt. Thus, embolization with Tornado coils (Cook, Bloomington, Ind., USA) via the inferior petrosal sinus for reduction of cavernous sinus reflux was followed by transarterial GDC coil (Boston Scientific Target, Calif., USA) embolization through the fistulous point. The DCCF was completely occluded without stent luminal narrowing or coil migration and clinical status improved with recovery of perfusion delay (Fig. 4). Ten months of clinical follow-up showed a modified Rankin scale of grade 2 without any fluctuation of ischemic symptoms.

Fig. 4
figure 4

a Complete obliteration of DCCF by coil embolization with preservation of in-stent luminal integrity. The stenotic lesion is fully dilated and there is no coil herniation. b Improved perfusion to the left cerebral hemisphere. Time-to-peak ratio compared with the right MCA territory improved from 0.60:1.67 of perfusion CT before coil embolization of DCCF to 0.72:1.38. This is also an improved result compared with initial perfusion CT in Fig. 1

Discussion

There are many possible complications of intracranial stent angioplasties, most commonly thromboembolization; however, arterial injury, hyperperfusion syndrome and stent occlusion by acute or subacute thrombosis could also occur [5, 13, 19, 20]. Arterial wall injury can cause particularly devastating clinical results, such as subarachnoid hemorrhage, thrombosis, and infarction. However, the reports of iatrogenic DCCF during endovascular or other intracranial surgical procedures are a few in number [8, 9, 12, 15, 21], and there are even fewer reports of stent-graft treatment of DCCF which occurred during neurointervention [8, 12]. To our knowledge, the development of DCCF after a stent angioplasty and its treatment with transarterial coil embolization has not been reported.

It is very important to predict the factors that would lead to this complication in order to avoid it. In this case, we could consider factors related to arterial wall injury into two categories. One category is patient factors, such as vascular tortuousness, calcification and intolerance to ischemia. The tortuousness weakens the tracking force of the stent delivery catheter by dispensing it to various vectors. Struggling to advance the catheter accumulates force on the tip of the stent delivery catheter, which could eventually injure the arterial wall [3, 6, 17]. Calcification of the vessel wall is another factor, which was observed in the ICA on the axial raw data of initial CT angiography (Fig. 3). Calcification is an outgrowth of atherosclerosis that reduces vascular elasticity and compliance [17]. During ballooning, the genu portion of the ICA could be stretched to some degree, and the concave portion of the stenotic calcified ICA might have been ruptured. A third factor is intolerance to ischemia because the patient’s brain is totally dependent on a unilateral blood supply [2]. Therefore, when the balloon was dilated, the ischemic time might exceed the threshold of ischemic tolerance, causing seizure. This could cause the operator to hurry through the procedure and eventually over-dilate the balloon. The seizure might occur not by ischemic intolerance but also by sudden occurrence of DCCF. The other category includes technical and stent factors. Thus, the stiffness of the coronary stent could harm the arterial wall. Excessive balloon dilatation by overestimation of vessel diameter could also be a cause [3, 8, 15]. Sufficient preoperative evaluation of the risk factors and indication, and experience of highly trained surgeons is very important for overcoming all of these obstacles. Moreover, even though its long-term follow-up result is not quite satisfactory nor established, future consideration of a newly developed self-expending stent is warranted. Predilatation provides a route of atraumatic passage for the soft-profiled self-expandable stent and reduces the resistance to expansion of the stent [3, 11, 22].

Since Serbinenko [18] reported his experience with a detachable balloon, endovascular treatment has become the first choice for the treatment of DCCF. Fistula occlusion using a detachable balloon is the preferred method for treating DCCF while preserving a patent parent ICA; however, this method may fail if the fistula is too small or large or cannot be reached due to its location [1]. Additionally, the balloon sometimes deflates within a few hours after deployment, which may result in recurrence of the fistula. Several reports about stent-graft for treatment of DCCF with various causes have been published recently [8, 9]. A covered stent graft is an ideal material theoretically because it isolates blood flow from the fistula. However, because of its thick profile, trackability is very low in a tortuous vascular environment. Deployment of the stent graft for this case could cause another arterial wall injury and additional luminal narrowing.

Transarterial coil embolization of the DCCF is feasible for small fistulas, and can be accomplished safely with the aid of a stent to prevent coil migration into the parent artery [1, 4, 10, 14, 16]. In this case, it was performed easily by selecting a fistula site with a microcatheter after reducing the shunt velocity by transvenous partial embolization. A previously placed coronary stent was helpful in detaching the coil safely without coil migration into the ICA.

Vascular tortuousness, calcification, intolerance to ischemia, overestimation of the size of the artery, overdilatation and other specific characters of stents should be considered as predictive factors of procedure-related arterial injury. Here, the authors report a rare case of transarterial coil embolization of DCCF, which was caused by stent angioplasty.