Carotid cavernous fifistula (CCF) is a pathological shunt that originates between a high-pressure internal carotid artery and a low-pressure cavernous venous system. The clinical presentation depends on hyperemia in the veins around the cavernous sinus (1). The most common clinical manifestations are pulsating exophthalmos, conjunctival edema, and cranial nerve palsy. Consequently, intracranial hemorrhage (ICH) is relatively rare, but it is a critical event and can result in a worse prognosis (2). Since January 2010, fifive cases of CCF presenting as intracranial hemorrhage have been treated in our department. All of these patients were treated endovascularly with an emergent procedure and had a good prognosis.

2.1. Clinical data 

From January 2010 to April 2017, fifive adult patients who harbored CCF presented with ICH and were admitted to our hospital; all of them had a history of head trauma. ICH was found in one patient after surgery for traumatic brain contusion, in two patients after embolization by detachable balloons (Goldbal2, Balt, France), and in another two patients who did not have any medical intervention. The duration from head trauma to ICH varied from 1 month to 20 years. All patients suffffered from headaches on admission, where three of them had additional pulsatile exophthalmos, and two of them had conjunctival edema as well. No cranial nerve palsy was noted (Table 1).

2.2. Imaging data 

All patients were diagnosed with ICH by emergency head computed tomography (CT), and there were two patients with intraparenchymal hemorrhage (IPH), one patient with subarachnoid hemorrhage (SAH), and two patients with IPH with SAH. All of the patients underwent emergency digital subtraction angiography (DSA) for the diagnosis of CCF. In the study, fifive mono-lateral direct CCFs with high flflow (Barrow type A) were noted, and all fifistulas were located in the cavernous segment (C4) of the internal carotid artery; two of them showed signs of total flflow steal. Reflflux drainage of cortical vein reflflux was detected in all patients, including three patients that had reflflux drainage to the straight sinus and four patients who had it to the superior sagittal sinus. Three of them had bulbous dilation of the drainage vein (s). 

2.3. Endovascular therapy 

Endovascular treatment by detachable balloon: The patients were placed in the supine position, and local anesthesia was given. The femoral artery approach was established by an 8F (French) sheath followed by the introduction of an 8F guiding catheter into the lesional ICA. Subsequently, a Goldbal2 balloon equipped at the tip of a MABDTE microcatheter (Balt, France) was navigated into the cavernous sinus through the fifistula by the arterial flflow. Then, the balloon was gradually dilated by fifilling it with diluting contrast agent (normal saline: ousu [iohyanol 300 mL, Yangzijiang Pharmaceutical Group Co., LTD., China] =1:1) until the fifistula had disappeared angiographically. Finally, the balloon was detached in situ after control angiography, by which the abnormal shunt disappeared. If the cavernous sinus cavity was too large, an additional balloon was used. Endovascular treatment by a detachable coil. The patients were placed in the supine position with general anesthesia. The femoral artery approach was established by a 6F sheath, followed by an introduction of a 6F guiding catheter into the lesional ICA. Under flfluoroscopic monitoring in the roadmap mode, two Echelon 10 microcatheters (eV3, USA) were navigated by

a Traxcess 14 microguidewire (MicroVention, USA) into the cavernous sinus through the fifistula successively. Thereafter, a series of detachable coils (Axium, eV3, USA) was introduced to occlude the fifistulae. After confifirmation of the obliteration of the abnormal shunt, the coils were detached, followed by retreatment with a microcatheter. If necessary, a HyperGlide 4 × 20 balloon (eV3, USA) was navigated to cover the fifistula in the ICA by an X-pedion 10 microguidewire (eV3, USA). Based on the complete occlusion of the ICA by dilation of the undetachable balloon, Onyx 18 glue (eV3, USA) was injected slowly under flfluoroscopy monitoring. Serial control angiography was carried out after the deflflation of the balloon until the abnormal shunt had disappeared. Thereafter, the balloon and microcatheters were removed. 

2.4. Follow-up 

Scheduled head CT was performed immediately after the endovascular procedure, on the 2nd day after treatment, and on the day of discharge. The modifified Rankin Scale (mRS) score for each of the patients was evaluated when they were discharged. All patients were followed up in the outpatient setting to detect any novel neurological defificits. The patients underwent follow-up CTs at 1 month, 3 months, 6 months, and 1 year after the endovascular procedure, and followup DSAs were performed at 6 months and 1 year after the procedure.

A total of fifive patients (harboring 5 CCFs) underwent six endovascular procedures; two of them were cured by detachable balloons (one patient by two balloons and another by three balloons), two patients by detachable coils, and one patient by Onyx glue and detachable coils. Among them, only one patient was cured in a second session by detachable balloons, and the other four patients were cured in one session; two patients who were treated with detachable balloons experienced transient headaches after the procedure, which were spontaneously alleviated by medication. There were no novel neurological defificits immediately after the endovascular procedures. The mRS score at discharge was 0 in four patients and 1 in one patient. All patients were followed up for 3–10 years (average 75.0 months), and no recurrence of intracranial hemorrhage or CCFrelated symptoms was noted. One patient who was treated with detachable balloons was found to have delayed spontaneous occlusion of the lesional ICA after 6 months, but the good compensation of the Willis circle resulted in an asymptomatic course (Figure 1). The lesional ICAs of the other four patients were intact. A pseudoaneurysm was noted by head CT angiography (CTA) in one patient who was treated with detachable balloons 3 months after the endovascular procedure, and the patient did not receive any further therapy. At the 3-year follow-up, a slight shrinkage of the pseudoaneurysm was noted in the patient’s CTA image (Figure 2)

4.1. Causes of ICH originating from CCF 

CCF was fifirst reported by Travers et al. in 1809. It can be divided into traumatic CCF and spontaneous CCF, according to the etiology. The former type is more common, accounting for ∼75% of the cases (3). Barrow et al. classifified CCFs into direct (Barrow A) and indirect types (Barrow B–D). The former has a direct shunt between the ICA and cavernous sinus, usually with high blood flflow (4). All fifive patients in this study had a clear history of head trauma attributed to Barrow type A. Thesymptoms of CCF depend on the direction of venous drainage, and the most common symptom is caused by drainage into the superior ophthalmic vein (SOV), named the Dandy triple sign, which presents as pulsating exophthalmos, a murmur, and bulbar conjunctival edema (5). Drainage to the posterior superior and inferior petrosal sinuses is relatively rare and may cause cranial nerve palsy and hemiplegia. Directed upward drainage via cortical veins into the sagittal sinus and deep venous system is more dangerous (6). In ∼9% of cases, increased venous pressure can generate reflflux drainage of the cortical veins, which are prone to various forms of ICH and even fatal brain stem hemorrhage. If bulbous dilation of the draining veins is noted, the risk of ICH will be higher (7). In this study, all fifive patients had reflflux drainage of the cortical veins, including the superior sagittal sinus in one patient, the straight sinus in one patient, and the superior sagittal and straight sinus in three patients. Three of them were complicated with a venous bulbous dilation. All of this evidence supports that cortical venous hypertension is a risk factor for CCF resulting in ICH. It has been reported that if anterior drainage through the SOV or posterior drainage through the superior and inferior petrosal sinuses is poor or absent, arterial hypertension will drain to the cerebral venous system through the sphenoid sinus or other channels, eventually causing ICH (8). Interestingly, one patient in this study had this extremely rare circumstance. This patient was diagnosed with CCF due to ocular symptoms, and half a month after the fifirst endovascular procedure with detachable balloons, she was hospitalized again due to ICH. Comparing the DSA images of the two procedures, the anterior venous drainage disappeared due to SOV obstruction by a premature deflflated balloon, but the cortical venous drainage caused by venous hypertension was much more obvious than before, which resulted in ICH of the right temporal lobe. 

4.2. Selection of endovascular strategy 

Generally, life-threatening CCF requires emergent management, and some of the high-risk factors for ICH include severe epistaxis, cortical venous reflflux, and angiographic venous bulbous dilation. With the development of endovascular technology, the open surgical strategy for CCF has been almost eliminated. The choice of endovascular strategy should depend on the feeding artery, the draining vein, the blood flflow velocity of the fifistula, and the integrity of the Willis circle (9). Malan et al. classifified traumatic CCFs into small, medium, and large CCFs based on the vascular structure, and this classifification is helpful for the selection of difffferent endovascular strategies (10).

Although there are increasing reports of the transvenous approach for the treatment of CCF, the transarterial approach is safer and simpler and is still suitable for the majority of such lesions. Many materials, including detachable balloons, detachable coils, Onyx glue, and covered stents, are available (11). From the perspective of economy and convenience, the detachable balloon is the preferred method, which can be completed under local anesthesia, and the operation is simple. Compared with detachable coil and Onyx glue embolization, detachable balloon embolization can be performed under local anesthesia, and the operation process is simpler (12). However, as in two of the patients in this study, after the fifirst balloon embolization of case 2, the patient’s right eye visual acuity decreased on the 10th day after the operation, and the CT examination showed bleeding. Further improvement of DSA showed that the ophthalmic vein pathway was blocked by the deflflated balloon, the direction of venous drainage was changed, and a large number of cortical veins were countercurrent, resulting in bleeding. Therefore, detachable balloon embolization of the right internal carotid cavernous fifistula was performed again at the same time. Postoperative cerebral angiography showed fifistula occlusion. In case 4 after the fifirst balloon embolization, on the 3rd day after the operation, the examination of the anterior and lateral cranial radiographs showed that only one of the two balloons remained. Combined with clinical practice, considering the possibility of balloon leakage, the leakage is broken again, and secondary interventional surgery is needed to re-block the leakage. DSA examination showed that the original embolization balloon disappeared, the fifistula reopened, and the cortical vein drainage was obvious, so the detachable balloon embolization of the right internal carotid cavernous fifistula was performed at the same time. Postoperative cerebral angiography showed fifistula occlusion. Herein, we think recurrence caused by premature deflflation is a major problem that needs to be considered. In addition, detachable balloons may eventually cause ICA occlusion and/or a cavernous sinus pseudoaneurysm, and these balloons have been less commonly used in the treatment of CCF (13). The detachable coil has good control ability, but there have been reports that mass stuffiffiffing in the cavernous sinus can cause compression of the cranial nerves and the possibility of aseptic inflflammation, in addition to the high medical cost (14). In this study, two patients treated with a detachable coil had a small, cavernous sinus cavity that was adjacent to the fifistula and were regarded as having cavernous aneurysms during the procedure. Some neurosurgeons advocated using Onyx glue, not only for its low cost but also to improve the safety of patients who had a higher risk of vascular injury due to their connective tissue disease (15). Another advantage of using Onyx glue is that collateral feeders, which are not visible on angiography, can be revealed during the progressive injection, and the use of Onyx glue results in improved cure rates. Therefore, the combination of Onyx glue with a detachable coil can not only retard shunt flflow but also limit the diffffusion of the Onyx glue, which ensures the therapeutic effffect and reduces the fifilling degree of the Onyx glue in the cavernous sinus, and this is the preferred treatment at present (16). Covered stents and flflow diverters (FDs) are also options for the treatment of CCF, but their use in patients with ICH remains controversial due to the need for subsequent dual antiplatelet therapy (17, 18). 

The overall disability and mortality rate of CCF is low, and consequent ICH is relatively rare. However, it is a serious complication that may cause irreversible neurological defificits or a life-threatening prognosis. The recovery rate of endovascular procedures is 90–100%, the complication rate is low, and the mortality rate is <1% (19–21). Therefore, individualized endovascular strategies should be carried out actively and in a timely manner for such patients. 

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. 

Written informed consent was obtained from the individual(s) and/or minor(s)’ legal guardian/next of kin for the publication of any potentially identififiable images or data included in this article. 

Conceptualization, supervision, and project administration: QZ. Methodology: QZ, A-LC, and C-GD. Software and visualization: Z-LL, A-LC, C-GD, D-HY, Y-HW, and YW. Validation and writing: Z-LL, A-LC, and YC. Investigation: A-LC and QZ. Resources and data curation: Z-LL and A-LC. All authors have read and agreed to the published version of the manuscript.

This article was funded by the Suzhou Science and Technology Development Program (SKJYD2021088), the Suzhou Science and Technology Development Program (SYS2019067), and the Second Affiffiffiliated Hospital of Soochow University Research Foundation Project (SDFEYBS2217). 

The authors declare that the research was conducted in the absence of any commercial or fifinancial relationships that could be construed as a potential conflflict of interest. 

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiffiffiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.