Endovascular treatment in bilateral cavernous sinus dural arteriovenous fistulas: a systematic review and meta-analysis

Few studies have discussed the disease nature and treatment outcomes for bilateral cavernous sinus dural arteriovenous fistula (CSDAVF). This study aimed to investigate the clinical features and treatment outcomes of bilateral CSDAVF. Embase, Medline, and Cochrane library were searched for studies that specified the outcomes of bilateral CSDAVF from inception to April 2022. The classification, clinical presentation, angiographic feature, surgical approach, and treatment outcomes were collected. Meta-analysis was performed using the random effects model. Eight studies reporting 97 patients were included. The clinical presentation was mainly orbital (n = 80), cavernous (n = 52) and cerebral (n = 5) symptoms. The most approached surgical route was inferior petrosal sinus (n = 80), followed by superior orbital vein (n = 10), and alternative approach (n = 7). Clinical symptoms of 88% of the patients (95% CI 80–93%, I2 = 0%) were cured, and 82% (95% CI 70–90%, I2 = 7%) had angiographic complete obliteration of fistulas during follow up. The overall complication rate was 18% (95% CI 11–27%, I2 = 0%). Therefore, endovascular treatment is an effective treatment for bilateral CSDAVF regarding clinical or angiographic outcomes. However, detailed evaluation of preoperative images and comprehensive surgical planning of the approach route are mandatory owing to complexity of the lesions.


Discussion
Bilateral CSDAVF is less studied than unilateral CSDAVF. To the best of our knowledge, this review is the first to emphasize bilateral lesions and analyze the outcomes. This systematic review included eight studies (97 patients).
Classification. Historically, researchers have classified CSDAVF according to the Barrow 21 ,Cognard 22 , or Satomi 23 classifications, with Cognard incorporating venous drainage as a risk factor and Satomi focusing on the outcomes. Although these two classifications were generated from general DAVF, they were specific to CSDAVF. More recently, Su et al. 24 and Thomas et al. 25 announced newer classifications in an attempt to summarize the complexity of the angiographic nature and clinical presentation of CSDAVF, which has been verified in recent studies 26,27 . However, none of these classifications were designed exclusively for bilateral CSDAVF. Wenderoth then reported the modified classification based on Cognard classification, adding a specific "c" classification for the bilateral group 28 . In addition, he specified the patency of each IPS for treatment planning. In this review, multiple classification methods were used, with five studies using Barrow classification, four studies using Cognard classification, and only one study reporting patients with the Satomi classification system. The heterogeneity was www.nature.com/scientificreports/ www.nature.com/scientificreports/ high between the studies; therefore, large-sample studies are warranted in the future to substantiate the associations of the classification with the nature and outcomes of bilateral CSDAVF.
Clinical presentation. In the current study, orbital and cavernous symptoms were significantly more common than cerebral symptoms, with low heterogeneity (I 2 = 0) in orbital symptoms and high heterogeneity (I 2 = 68) in cavernous symptoms ( Table 2). The largest cohort in our review 2 had a lower cavernous symptom rate (23%) than others. Previous studies have shown the relationship between fistula drainage and clinical symptoms and concluded that anterior drainage may cause more orbital symptoms, while posterior drainage may cause more neurological symptoms 8 . A higher orbital symptom rate seemed to indicate a more indolent disease course. However, in a recent study, cortical venous reflux, which is strongly associated with intra-cerebral hemorrhage before treatment, mostly presented with chemosis or orbital pain 29 . Therefore, a comprehensive study including magnetic resonance imaging, computed tomography angiography, and DSA is warranted for patients with orbital symptoms to determine if pial venous reflux exists. In addition, interestingly, bilateral presentation was only observed in 46% of patients with bilateral CSDAVF in our review. Fay et al. attributed this to the direction of fistula flow 3 . Taken together, patients with suspicious symptoms and signs should be transferred to an experienced physician for full evaluation and sophisticated treatment plans. www.nature.com/scientificreports/ Surgical approach and nuance in bilateral lesions. IPS is usually the first choice for transvenous endovascular surgery owing to its simplicity, effectiveness, and the shortest connection with the cavernous sinus from the jugular bulb. In our review, the most common route was the IPS, with low heterogeneity. If the IPS route is chosen, a unilateral or bilateral approach can be applied to bilateral CSDAVF. However, in our review, unilateral or bilateral IPS occlusion rates were 22% (95% CI 6-57%, I 2 = 80%) and 15% (95% CI 7-28%, I 2 = 31%), respectively, similar to previously published data regarding CSDAVF 29,30 . The high variability could be caused by the limited number of studies and patient numbers. The IPS route becomes more important for bilateral CSDAVF because it would be difficult to completely obliterate lesions on each side via a single alternative route. Therefore, embolization via an occluded IPS has become challenging, but somehow an inevitable procedure. Multiple methods have been used to deal with occluded IPS, including "Pocket-Flash method, " 31 "Frontier-Wire Probing technique, " 32 and "microguidewire looping technique" 33 . However, some have opposed the breaching of the occluded IPS technique and considered it a dangerous maneuver 34 . In the current review, the IPS injury rate was 4% for bilateral CSDAVF. Several other approach routes for CSDAVF have been reported and are summarized in Fig. 3. The SOV route has been previously reported to have a satisfactory embolization rate 38 . Direct puncture or surgical cutdown has been utilized to approach SOV and avoid the difficulty in navigating the catheters and the possibility of vessel wall injury during the procedure 34,41,44 . Possible complications include periorbital structural damage and hematoma 55,56 . However, in bilateral CSDAVF, if a unilateral approach is chosen, it would be more difficult to pass the cavernous sinus connection due to poor catheter support. On the other hand, bilateral SOV routes increase post-operative suffering and worsen cosmetic results 57 . Therefore, for bilateral CSDAVF, a unilateral or bilateral approach through the traditional IPS, facial vein, or SOV approach through a direct puncture or surgical cutdown are all reasonable choices, and detailed treatment plans should be made before the surgery and adjusted during the surgery.
Another issue for bilateral CSDAVF is whether a single-session or staged operation should be performed. Some previous studies have advocated staged operation for unilateral and bilateral CSDAVF 58,59 . The reasons were to reduce the coil amount, which has been proven to be associated with postoperative cranial nerve VI palsy 60,61 owing to the anatomical features of this nerve [62][63][64] . In addition, the hemodynamic change between surgeries may also have the possibility of reducing the coils needed for second-stage surgery 3 . However, staged surgery still has some obstacles. Firstly, navigating the microcatheter into the venous pouch or through the connection of the cavernous sinus with the resistance of previous coils and onyx can be challenging, since "Turn-Back Embolization Technique" is usually applied 65 . Second, the timing of surgery can be ambiguous. Clinical embolization outcomes or paradoxical cranial nerve VI palsy are difficult to evaluate in this situation [66][67][68] . Single-stage surgeries have the advantage of avoiding these difficulties. Although no studies have compared the efficacy of single or multi-stage surgery for bilateral CSDAVF, a careful assessment of preoperative images to ensure that all the venous pouches and fistulas were targeted is of paramount importance as the opacity of the mass of coils that may hide a residual flow could be especially challenging in bilateral CSDAVF than in unilateral lesions during the surgery [69][70][71] . www.nature.com/scientificreports/ Complication rate. One of the included studies 15 with only three patients had a higher complication rate (66.6%), reporting two transient cranial nerve palsy patients. The remaining studies in current review reported low complication rates. The most common complication was cranial nerve VI palsy, with 13% of the patients recovering spontaneously and 4% of the patients developing permanent nerve deficits. A previous meta-analysis of general CSDAVF group had a complication rate of 7.75% (95% CI 3.82-12.7%) with minimal permanent deficits (0.15%) 10 . In addition to the cranial nerve palsy and IPS injury mentioned above, leakage of embolization agents was also a possible consequence. Onyx or nBCA, which refluxes back into the feeding arteries, can result in non-target embolization and have catastrophic complications. This is especially important for bilateral lesions, since adjuvant onyx or nBCA is frequently used for complete embolization. In this study, none of the previously reported serious complications, such as brainstem infarction, brainstem hemorrhage, and intra-cerebral hemorrhage, were speculated to be related to the advancement of the techniques and were well aware of the anatomy of the related structures 67,72,73 . However, Wakhloo et al. reported a case of nBCA leakage without severe stroke or hemorrhagic episode 14 . Finally, in complications related to uncontrolled bleeding or strategies for endovascular bailout, surgery could always be considered to obliterate the fistula and achieve hemostasis 74,75 . Limitations. One of the main limitations of this systematic review was the retrospective design of the majority of the included studies, which was a potential source of bias due to confounding factors. In addition, with intension to reveal the whole picture of the disease, we included several studies with small case numbers. Therefore, the results had to be interpreted carefully. Second, the classification for CSDAVF was not uniform between the studies. Third, the definition of bilateral CSDAVF has not been clarified in previous studies. Fourth, a majority of study was excluded as they failed to report the outcomes specifically for bilateral lesions. This can lead to bias during data analysis. Fifth, none of the studies reported the intraocular pressure measurement before or after the treatment.

Conclusion
Management of bilateral CSDAVF remains challenging. The patient can present with unilateral symptoms, which pose difficulties in disease diagnosis. The endovascular treatment strategies for bilateral CSDAVF should be tailored according to the patency of the IPS, accessibility of the SOV or other routes, and if staging operation is needed. According to meta-analysis of modest quality of data, weak suggestions can be made that a transvenous embolization is a feasible treatment method for bilateral CSDAVF.

Data availability
The data that support the findings of this study are available on request from the corresponding author.