Introduction

Tentorial dural arteriovenous fistulae (DAVF) are complex and extremely dangerous lesions. [1,2,3]. According to Lawton et al., the tentorial middle line region DAVF include galenic, straight sinus, and torcular DAVF [1]. They have summarized the pathological anatomy as well as the best surgical strategies for tentorial DAVF. However, endovascular treatment (EVT) is now a better option for DAVF [2, 4, 5] and tentorial middle line region DAVF has a deeper location, more complex angioarchitecture and more cognitive disorders than other region’s DAVF. The authors think it is necessary to summarize the treatment strategies for EVT in this region with a large sample size.

Methods

Data collection and definition

A total 78 tentorial middle line region DAVF were included in this study between 2000 and 2020 at Xuanwu Hospital in Beijing. The patient’s clinical data and follow-up outcomes were collected. All patients were evaluated using the modified Ranking score (mRS) at the preoperative, postoperative and follow-up periods. Poor outcomes were defined as follow-up mRS greater than or equal to 2. Cognitive disorders were defined as dementia, psychiatric symptoms, and hypomnesia. Ocular complaints included diplopia, hypopsia, and conjunctival congestion. A small percentage of DAVF patients were discovered by accident as a result of other diseases or routine checkup. Other diseases included intracranial aneurysms, intracranial tumors, pulsing mass in the scalp and spinal vascular malformations. Admission, outpatient and phone follow-up were all included. We investigated the outcomes and complications of each treatment modality. The local ethics committee of our organization gave its approval to this study.

Statistical analysis

SPSS 26.0 was used for statistical analysis. Means and standard deviations were used to describe continuous variables, while counts and percentages were used to present categorical variables. Fisher exact test and Chi-square test was performed to test significance of associations between categorical variables. Poor outcomes and cognitive disorders were predicted using univariate and multivariate logistic regression analyses. Significant results were defined as 2-sided P values < 0.05.

Results

Presentation

There were 78 patients with adult DAVF in the tentorial middle line region, including galenic (39/78, 50%), straight sinus (12/78, 15.4%) and torcular DAVF (27/78, 34.6%) (Table 1). Except for hydrocephalus (P = 0.007), there were no statistically significant differences between the galenic group and the straight sinus-torcular group. The mean patient age was 50 ± 12 years and 63 were male (63/78, 80.8%). The median duration of symptoms was 3 months. 67 patients (67/78, 85.9%) presented with related symptoms and 11 patients (11/78, 14.1%) were without symptoms or found incidentally. Three patients with galenic DAVF were discovered during a routine checkup, and three others were discovered by chance due to a pulsing mass in the scalp. In straight sinus DAVF, one patient was found accidentally during a DSA examination for spinal vascular malformation, and the other was identified during a routine checkup. Torcular DAVF was found by chance in two patients as a result of intracranial aneurysms and one as a result of an intracranial tumor. The most common symptoms of the tentorial middle line region DAVF were cognitive disorders (29/78, 37.2%) (16 in galenic, 16/39, 41%) (7 in straight sinus, 7/12, 58.3%) (6 in torcular, 6/27, 22.2%) followed by headache (26/78, 33.3%) and intracranial hemorrhage (19/78, 24.4%) (7 in galenic, 7/39, 17.9%) (4 in straight sinus, 4/12, 33.3%) (8 in torcular, 8/27, 29.6%). When admitted, 41 patients’ mRS ≥ 2 (41/78, 52.6%) (22 in galenic, 22/39, 56.4%) (8 in straight sinus, 8/12, 66.7%) (11 in torcular, 11/27, 40.7%).

Table 1 Baseline characteristics of 78 tentorial midline region DAVF patients
Full size table

Straight sinus occlusion or bilateral transverse sinus occlusion (thrombus) were found in 20.5% of the patients (16/78) (7 in galenic, 7/39, 17.9%) (5 in straight sinus, 5/12, 41.7%) (4 in torcular, 14.8%). Tentorial middle line region DAVF with internal cerebral vein drainage was found to be a risk factor for the onset of cognitive disorders in a multivariate analysis (Table 2) (OR 7.189, 95% Cl 1.787–28.923).

Table 2 Univariate and multivariate logistic regression to predict cognitive disorders in 78 tentorial midline DAVF patients before treatment
Full size table

Angioarchitecture

The majority of the DAVF had numerous arterial feeders as well as venous drainage (Table 3). The most common arterial feeds were branches from the occipital artery (OA) (71.8%), followed by branches from the middle meningeal artery (MMA) (67.8%), the posterior meningeal artery (PMA) (52.5%), tentorial branches from the superior cerebellar artery (SCA) and posterior cerebral artery (PCA) (48.7%), and the tentorial artery originating from the meningohypophyseal trunk (MHT) (47.4%). Other arterial feeders included branches from the anteroinferior cerebellar artery (AICA) and posteroinferior cerebellar artery (PICA) (11.1%), the ascending pharyngeal artery (AphA) (12.8%), the posterior auricular artery (PAA) (2.6%), the superficial temporal artery (STA) (12.8%) and branches of the anterior cerebral artery (ACA) (3.8%).

Table 3 Angiographic features and treatment outcomes of 78 tentorial midline DAVF
Full size table

Most of the tentorial middle line region DAVF were high grade (Borden type II or III). A few torcular DAVF had direct venous drainage to the straight sinus, vein of Galen and internal cerebral vein. The drainage veins included the vein of Galen (66.7%), the internal cerebral vein (39.7%), cortical and cerebellar veins (48.7%), the transverse sinus (21.8%) and the cavernous sinus (5.1%).

Treatment and outcomes

74 patients (74/78) underwent EVT (Table 3) (36 in galenic, 12 in straight sinus, 26 in torcular). Four patients were not treated because they either refused treatment or the interventional approach was difficult. Four intracranial hemorrhages occurred during the embolization and 1 patient died due to the hemorrhage in hospital. One cranial nerve palsy (IX) and two brain infarctions occurred after the embolization. At the follow-up, they were completely symptom-free. Transarterial embolization (TAE) was the most common access route (66/74, 89.2%). Only one torcular DAVF were treated transvenously (1/26), and six torcular DAVF (6/26) were treated with mixed approach. The most common embolization materials were ONYX (61/74, 82.4%), Glubran (24/74, 32.4%), balloon (14/74, 18.9%), and coil (1/74, 1.4%). Complete obliteration of the fistulas was achieved in 87.5% (64/74) (31 in galenic, 31/36, 86.1%) (11 in straight sinus, 11/12, 91.7%) (22 in torcular, 22/26, 84.6%) of patients. Two or more embolizations were performed in 31.3% (20/64) (12 in galenic, 12/31, 38.7%) (8 in torcular, 8/11, 72.7%) of the patients who achieved complete obliteration.

Follow-up

71 patients (91%) (mean, 56 months) (34 in galenic, 34/39, 87.2%) (12 in straight sinus, 12/12, 100%) (25 in torcular, 25/27, 92.6%) had phone, outpatient, or admission follow-up. The DSA follow-up period (25/78, 32.1%) was 13.8 (6–21) months. Two of them (2/25, 8%; both in Galen; 1 month and 3 years after embolization) had fistula recurrences after complete embolization and were embolized again. The phone follow-up period (70/78, 89.7%) was 76.6 (40–92.3) months. One patient died while being treated in the hospital, and six patients were lost to follow-up. 75% (3/4) of the untreated patients remained stable but experienced cognitive disorders. 84.5% (60/71) (28 in galenic, 10 in straight sinus, 22 in torcular) of patients had related symptoms at presentation: when patients were followed up, 43.3% (26/60) (9 in galenic, 9/28, 32.1%) (4 in straight sinus, 4/10, 40%) (13 in torcular, 13/22, 59.1%) had complete symptom resolution, 8.3% (5/60) (1 in galenic, 4 in torcular) reported worsening symptoms. One of the patients’ symptoms (in torcular) deteriorated as a result of hemorrhaging during TAE. The rest of the patients (29/60) remained stable. 89.7% (26/29) (18 in galenic, 18/28, 64.3%) (6 in straight sinus, 6/10, 60%) (5 in torcular, 5/22, 22.7%) of the stable patients still had cognitive disorders. When followed up, 15 patients’ mRS was greater than or equal to 2 (15/71, 21.1%) (6 in galenic, 6/34, 17.6%) (3 in straight sinus, 3/12, 25%) (6 in torcular, 6/25, 24%). Tentorial middle line region DAVF with internal cerebral vein drainage (OR 6.514, 95% Cl 1.201–35.317) and intracranial hemorrhage (OR 17.034, 95% Cl 1.122–258.612) during embolization were the risk factors for predicting poor outcomes (Table 4).

Table 4 Univariate and multivariate logistic regression to predict poor outcomes (follow-up mRS ≥ 2) in 71 tentorial midline DAVF patients when admission or phone follow-up
Full size table

Discussion

Treatment

In our facility, TAE is the first-line treatment for the tentorial middle line region DAVF. TAE pathways include the MMA, OA, PCA, AphA, and PMA. Transvenous embolization (TVE) is as an adjuvant treatment. The typical interventional procedures of tentorial middle line region DAVF were illustrated (Figs. 1, 2, 3, 4).

Fig. 1
figure 1

Galenic DAVF. A Sagittal high-resolution magnetic resonance imaging (HRMRI) showing the dilated vein of Galen and straight sinus. Note enlargement of the lateral ventricle. B Anteroposterior view of the right external carotid artery (ECA) angiogram demonstrating tentorial midline region DAVF with multiple artery supplies. C, D Lateral view of the left internal carotid artery (ICA) and the left ECA angiogram showing arterial supply to the fistula from the left pericallosal artery (C, yellow arrow), MMA (D, yellow arrow) and OA (D, blue arrow) converging into the mural channel with the dilated vein of Galen and straight sinus drainage. Note the mural channels (D, white arrows). E, F Super-selective injection through pericallosal artery (E). TAE via pericallosal artery using Glubran was performed. Note the glue cast (F, white arrow). G–I Super-selective injection through left falcine artery from the MMA (G). TAE via falcine artery using ONYX was performed. Note the ONYX cast in the mural channels around the vein of Galen (H, white arrows) and the ONYX not completely occluded the vein. J There is no residual shunting. 4 years after the embolization, the patient complained of mild hypomnesia, but the severe headache she had before the operation was gone. She was satisfied with the results of the treatment

Full size image
Fig. 2
figure 2

Straight sinus DAVF. A Axial T2 MRI showing increased signal in the bilateral basal ganglia. B Anteroposterior view of the vertebral artery (VA) angiogram demonstrating tentorial midline region DAVF. C–E Lateral view of the right VA (C) and left ECA (D, E) angiogram demonstrating arterial supply to the fistula (C, yellow arrow) from the PCA, the SCA, the meningeal branches of OA (D, yellow arrow) and posterior branches of the MMA (E, yellow arrow and white arrows) with the reflux to the straight sinus, inferior sagittal sinus and internal cerebral vein (D, white arrows). Note the occlusion of the straight sinus. F Super-selective injection through posterior branches of the MMA. G TAE via MMA using ONYX was performed. Be careful not to allow ONYX reflux to vein of Galen. H Posterior angiogram showing no residual fistula. I, J The angiograms 1 year after the embolization showing no recurrence. The patient still complains of hypomnesia and becomes agitated, but his condition has significantly improved since the operation

Full size image
Fig. 3
figure 3

Torcular DAVF. A Axial computer tomography (CT) showing the left temporal lobe hemorrhage. B Anteroposterior view of the right ECA angiogram demonstrating tentorial midline region DAVF. Note the bilateral transverse sinuses occlusion (late venous phase). C, D Lateral view of the left ECA angiogram showing arterial supply to the fistula from the MMA (C, yellow arrows), the AphA (D, yellow arrow) with the straight sinus, vein of Galen (C, white arrow), the superior sagittal sinus (C, black arrow) and vein of Trolard (C, blue arrow) drainage. The left MHT, PMA, SCA also supplied the fistula (not shown). E Balloon-dilated occluded right transverse sinus was performed. F–H Transvenous balloon (F, location of the torcular herophili, white arrow)-assisted ONYX embolization was performed via MMA (G) and AphA (H). I After complete embolization of the fistula, DSA on late venous phase showing patency of right transverse-sigmoid sinus drainage. J–L The angiograms 2 years after the treatment showing no residual shunting. Related headache and tinnitus completely disappeared

Full size image
Fig. 4
figure 4

Torcular DAVF. A Anteroposterior view of the right ECA angiogram demonstrating tentorial midline region DAVF. Note the right transverse sinuses occlusion. B, C Lateral view of the right ECA and VA angiogram showing arterial supply to the fistula from the MMA, the OA, the PAA and the PCA. D Super-selective injection through pial branches from PCA. TAE via PCA using Glubran was performed. E–H Transvenous balloon (E, lateral view)-assisted ONYX embolization was performed via MMA (F, G) and no residual shunt was left (H). Related ocular complaints almost disappeared 3 years after the embolization

Full size image

Tentorial DAVF are often with pial arterial supply [1, 2]. Before fistula embolization through dural feeders, Wu et al. recommended obliterating the pial feeding arteries to reduce the risk of hemorrhage [6]. However, four intracranial hemorrhages happened when obliterating the pial feeding arteries and the patients had bad outcome (OR 17.034, 95% Cl 1.122–258.612). When pial feeders’ obliteration is difficult to achieve, it should not be forced. The second choice is to reduce the blood flow reflux to the deep venous system and cortical veins.

In general, we were able to achieve Cognard degradation in the partially embolized patients (Cognard III or IV to Cognard I or IIa). Following that, we recommend that patients be readmitted to the hospital for DSA after 1 to 3 months to see if they can be re-intervened to complete embolization. If the intervention still fails, the patient is advised to have gamma knife therapy [7].

Galenic DAVF

Early on, combined endovascular and surgical treatment was the first-line treatment for DAVF involving the deep venous system [1, 8, 9]. At present, TAE has the most effective treatment modality for galenic DAVF [10,11,12]. Only one case of galenic DAVF was treated with mix approach in our center. In this case, TVE was performed via straight sinus, with a coil placed in the dilated vein of Galen to reduce blood flow. The embolization was then achieved with TAE via MMA using ONYX.

According to Kim et al., mural channels were described as tubular structures that were isolated from the sinus’s main lumen and into which feeders converged [13]. We discovered a number of DAVF near the vein of Galen with similar angioarchitecture (Fig. 1D, white arrows) [14]. Several arterial feeders converged into the channel. TAE administered through the arterial feeders to embolize the channels could result in complete embolization. There is no need to obstruct the entire non-functional vein of Galen (Fig. 1H, white arrows).

Straight sinus DAVF

Straight sinus DAVF is sometimes associated with straight sinus occlusion [15] and often with fragile pial vein drainage [1, 12] which prevents TVE. TAE is still the first-line treatment.

We believe there are two types of straight sinus DAVF, which resembles the DAVF in the sinus region: sinus type and sinus wall type. The feeders directly converging into the sinus, with or without reflux, are referred to as sinus type. Sinus wall type refers to feeders that are isolated from the sinus's main lumen and reflux to an adjacent cortical vein, such as the superior vermian vein in straight sinus region, due to the clot of the main sinus. [1]

In the illustrated case (sinus type), several arterial feeders directly converged into the straight sinus and refluxed into the deep venous with obstruction of the straight sinus (Fig. 2). In this particular type of straight sinus DAVF, complete occlusion of the fistula could be easily achieved by completely occluding the non-functional straight sinus.

Torcular DAVF

TAE, TVE, and mixed approach are all used to treat torcular herophili DAVF [12, 16, 17]. TAE alone cannot always achieve complete occlusion of the fistula. TVE may be performed as an adjuvant method. In the illustrated case (Fig. 3), a balloon was used to recanalize the occluded right transverse sinuse, and transvenous balloon-assisted ONYX embolization was performed via MMA and AphA. It is not recommended to occlude the entire straight sinus in order to achieve complete occlusion. Although in this case, the patient’s outcome was good. However, without the high blood flow, the recanalized transverse sinus is more likely to occlude again. In this case, we recommend partial embolization of the fistula rather than occluding the venous structure, which still functions normally.

For torcular DAVF in which the artery feeders confluence at one point of the sinuses or a venous lake next to the sinuses, ONYX embolization can be performed by positioning the balloon directly at the torcular herophili via the TVE alone.

Prognosis

Tentorial middle line region DAVF induces a higher rate of thalamic dementia because of thalamic venous congestion (Table 3) [1, 10, 18,19,20]. In our study, the most common symptoms of the tentorial middle line region DAVF were cognitive disorders (29/78, 37.2%) (16 in galenic, 16/39, 41%) (7 in straight sinus, 7/12, 58.3%) (6 in torcular, 6/27, 22.2%). Although it is reported that related symptoms are potentially reversible [10, 18], when patients were followed up, 89.7% (26/29) (18 in galenic, 18/28, 64.3%) (6 in straight sinus, 6/10, 60%%) (5 in torcular, 5/22, 22.7%) still had cognitive disorders. Since the follow-up period is long enough, we believe the neurological deficits caused by DAVF are permanent. When we followed up with phone calls, we discovered that 5 patients (5/29, 17.2%) died as a result of a car accident when driving. We believe the tragedy was caused by cognitive disorders. Family members should pay more attention to the patients.

Limitations

This is a non-randomized retrospective single-center study. Many patients or family members described the patient’s symptoms over the phone during the most recent follow-up. It may have an effect on the accuracy with which doctors evaluate patients. Regular follow-up procedures included both MRI and DSA. Peng Zhang, the chief physician, reviewed the MRIs that were typically done in the outpatient clinic. DSA reviews were only performed as necessary. Annual MRIs are advised for follow-up, and angiography is still the gold standard for a precise diagnosis of complete dAVF obliteration [7, 21]. However, only 32.1% of all patients have an DSA re-examination. In this study, a number of patients were followed up using MRI [22,23,24], which may reduce the detection of recurrence fistulas.

Conclusion

Most patients with tentorial middle region DAVF are symptomatic at presentation. Most of them have cortical venous drainage. Drainage to the internal cerebral vein is associated with cognitive disorders. TAE is the first-line treatment for this region. When pial feeders’ obliteration is difficult to achieve, it should not be forced due to the poor outcomes after intracranial hemorrhage.