Safety and Efficacy of Transvenous Embolization of Ruptured Brain Arteriovenous Malformations as a Last Resort: A Prospective Single-Arm Study

Twenty-one consecutive patients with ruptured brain AVMs who underwent transvenous embolization were prospectively followed between November 2016 and November 2018. Complete AVM nidus obliteration was shown in 16 (84%) of 19 patients. One (5%) patient with a small residual nidus after treatment showed complete obliteration at 13-month follow-up. There were 5 hemorrhages and 1 infarction; 4 patients' symptoms improved gradually. Transvenous embolization can be performed only in highly selected hemorrhagic brain AVMs with high complete obliteration rates, but it should not be considered as a first-line treatment. BACKGROUND AND PURPOSE: The efficacy and safety of transvenous embolization for brain arteriovenous malformations remains unclear, given the very limited number of cases reported. This prospective study was performed to assess this technique in ruptured AVMs. MATERIALS AND METHODS: Twenty-one consecutive patients with ruptured brain AVMs who underwent transvenous embolization were prospectively followed between November 2016 and November 2018. The Spetzler-Martin grade was I in 3 AVMs (14.3%), II in four (19.0%), III in eleven (52.4%), and IV in three (14.3%). The complete AVM occlusion rate was calculated from 6-month follow-up DSA images. Occurrence of hemorrhage and infarction after embolization was evaluated using CT and MR imaging within 1 month after the operation. The mRS was used to assess the functional outcomes. RESULTS: Complete AVM nidus obliteration was shown in 16 (84%) of 19 patients with technically feasible AVMs immediately after embolization. One (5%) patient with a small residual nidus after treatment showed complete obliteration at 13-month follow-up. There were 5 hemorrhages and 1 infarction; 4 patients' symptoms improved gradually. The percentage of cases with mRS ≤ 2 rose from 57.1% (12/21) before embolization to 66.7% (14/21) at 1-month follow-up. Both the morbidity and mortality rates were 4.8% (1/21). CONCLUSIONS: Transvenous embolization can be performed only in highly selected hemorrhagic brain AVMs with high complete obliteration rates, improved functional outcomes, and acceptable morbidity and mortality rates, but it should not be considered as a first-line treatment.

B rain arteriovenous malformations are characterized by an intervening plexus of abnormal vessels (nidus) between feeding arteries and draining veins. The most common presentations of AVMs include hemorrhages, seizures, headaches, and progressive neurologic deficits due to chronic vascular steal. 1 Hemorrhage can occur because of an inherent lack of normal smooth-muscle properties in the vascular architecture of AVMs. A first hemorrhagic event is associated with an increased risk of a new bleeding. Morbidity of AVM hemorrhage is estimated to be between 13% and 50%, and the mortality rate after intracranial hemorrhage from AVM rupture ranges from 12% to 67%. 2 Treatment strategies (microsurgery, radiosurgery, and endovascular embolization) are chosen on the basis of angioarchitecture, location, and presentation of AVMs. 3,4 Elimination of hemorrhage risk by extirpation or endoluminal closure of the nidus remains the primary goal of AVM treatment. Surgical resection of an AVM can be challenging in deep, inaccessible locations and eloquent areas. Radiosurgery may not be ideal in hemorrhagic AVMs as the first choice because of the long latency between treatment and AVM involution. Endovascular embolization through arterial routes may not be curative for AVMs in many situations, especially in cases with indirect feeders and extreme vessel tortuosity. 5 Transvenous embolization can overcome these disadvantages. [6][7][8][9][10][11][12][13][14] Current indications for transvenous embolization of AVMs include deep location, unfavorable arterial access, a small nidus, and a single draining vein. [6][7][8][9][10][11][12][13][14] Because there are only a limited number of cases reported using this approach, the effectiveness and safety remain unclear. In this study, we tried to further validate this method.

MATERIALS AND METHODS
The Medical Ethics Committee of Henan Provincial People's Hospital approved this study. Key inclusion criteria were as follows: 1) patients with a ruptured brain AVM; 2) patients not suitable for intra-arterial embolization due to the absence of arterial access, narrow arterial feeders, extremely tortuous course, too many feeders, and so forth; and 3) patients in whom lesions were not amenable to surgery or radiosurgery or patients who refused to undergo surgery or radiosurgery (AVMs not amendable to surgery or radiosurgery are defined as cases with modified Spetzler-Martin grades of IIIþ, IV, and V based on modification of the Spetzler-Martin scale 15 and those with scores of >1.5 based on Pollock-Flickinger grading scale, 16 which are proved have high rates of iatrogenic complications); and 4) patients with favorable venous angioarchitecture and a single main draining vein.
Key exclusion criteria were as follows: 1) multiple AVMs, 2) patients with ≥2 main draining veins, 3) a history of severe allergies to contrast or nonadhesive embolic agents, and 4) uncontrolled active bleeding.
Twenty-one consecutive patients with ruptured brain AVMs underwent transvenous embolization between November 2016 and November 2018. Thirteen patients experienced brain AVM ruptures with intracranial hematoma and intraventricular hemorrhage, 6 patients had intracranial hematoma without intraventricular involvement, 1 patient had subarachnoid hemorrhage, and another one had intraventricular hemorrhage. The mean AVM size was 2.76 6 1.24 cm, ranging from 1.2 to 5.5 cm. A summary of patient characteristics is shown in Table 1.
All the embolization procedures were performed with the patient under general anesthesia. A 6F sheath was placed in the internal jugular vein followed by a 6F guiding catheter, which was advanced to the main draining vein of the brain AVMs. One or 2 microcatheters (Marathon, Covidien, Irvine, California; Apollo, Covidien; Echelon, Covidien; or Headway DUO, MicroVention, Tustin, California) were placed as close as possible to the nidus of the AVMs. A vascular sheath was placed in the right femoral artery followed by guide catheter placement though which a microcatheter was advanced into the feeding artery of the AVMs. Arterial inflow of the feeding  artery was reduced by transarterial coil or liquid embolization or balloon inflation. Transvenous embolization was initiated by injecting ethylene-vinyl alcohol copolymer (Onyx; Covidien) into the nidus through the venous access route. Transvenous partial coiling in the draining vein through 1 microcatheter (known as the transvenous pressure cooker technique) was used to prevent reflux of Onyx. [17][18][19] At the completion of the procedure, the microcatheter injecting Onyx was cut at the level of the jugular sheath. 14 Preoperative baseline functional status was determined using the mRS score. The same evaluation was performed on postoperative days 2, 7, and 30 and at 3, 6, and 12 months. Occlusion of the AVM nidus was categorized as complete (no residual nidus) and near-complete obliteration (residual nidus of <3 mm in diameter). Comparison of the degree of occlusion was classified as progressive, stable, or recanalized.
Procedural safety was evaluated by assessing the periprocedural complications occurring within 1 month after embolization. 20 Any deficit that resolved within the first 30 days was characterized as transient. Any deficit that persisted beyond 30 days was considered permanent. An mRS score of 2 indicated a nondisabling deficit. An mRS score of ≥3 indicated a disabling deficit. Periprocedural-related death was defined as any death occurring within 30 days after the procedure.

Statistical Analysis
Categoric variables are presented as numbers and percentages, and continuous variables are presented as mean and SD. A 2sided P value < .05 was considered significant.

Procedural/Technical Specifications
The median time between hemorrhage and transvenous treatment was 47 days (range, 9-164 days). The procedure was technically feasible in 19 (90.5%) cases (Table 2). There was failure of microcatheter placement into the draining vein via the nidus in 2 cases.
For the 19 embolization procedures, access included the straight sinus in 9 cases, the cortical veins via the superior sagittal sinus in 6 cases, and the cortical veins via the transverse sinus in 4 cases.
The transvenous pressure cooker technique and Onyx injection were accomplished using 1 microcatheter in 3 patients and 2 microcatheters in 16 patients. The microcatheters for Onyx injection were retained in all patients except one.
The mean procedural time of all 21 patients from puncture onset to puncture closing was 251.6 6 73.1 minutes. The mean volume of Onyx used for embolization was 3.79 6 3.30 mL (range, 0 -13 mL).

AVM Nidus Obliteration
Immediately after the procedure, complete obliteration was achieved in 16 cases (Table 2), with an obliteration rate of 84.2% in the 19 technically feasible cases and 76.2% in all 21 cases. A postoperative small residual nidus was present in 3 patients after transvenous embolization. Follow-up angiography was performed from 1 to 15 months after embolization. Stable obliteration was confirmed in 18 patients. No recurrence was noted (Fig 1). One AVM located in the parietal lobe and basal ganglia showed progressive occlusion (Fig 2).

Procedural Safety
Procedure-related complications occurred in 6 patients, including 4 intraventricular hemorrhages, 1 intraparenchymal hemorrhage with combined intraventricular hemorrhage (Fig 3), and 1 cerebral infarction (Fig 4). Three of these patients underwent ventriculostomy, 1 of whom also underwent lumbar drainage; the other patients just had conservative management. Good outcomes (mRS 2) at the 1-month evaluation were achieved in 4 patients following the above therapies. The clinical outcome was poor (mRS = 5) in 1 patient (Fig 4) at the 1-month evaluation, and another patient died.
Among the above complications, 5 occurred in small-sized AVMs, 5 in eloquent areas, 5 with deep venous drainage, 9 in deep locations, 4 with Spetzler-Martin III, 4 with aneurysms in the feeding artery or AVM nidus, and 1 with venous outflow stenosis. The Fisher exact test indicated no significant difference among these factors ( Table 3).

Transvenous Obliteration
Total occlusion of AVMs with detachable or nondetachable microcatheters in transarterial procedures is difficult to achieve, mostly due to a variety of challenges, including microcatheter navigation, multiplicity of arterial feeders, and achieving deep and complete Onyx penetration of the AVM nidus. 5 In these challenging situations, transvenous embolization may be a viable alternative with total occlusion rates as high as 92.6%, 21 which is comparable with ours in this study. Transvenous microcatheter navigation, which occurred in 2 of our cases, might be difficult. However, this issue might be resolved by balloon-assisted microcatheter navigation techniques 22 or hybrid surgical techniques. 23,24 The high occlusion rates of AVMs may be due to the small, compact architecture, which is more easily penetrated. Additionally, the single, embolized draining vein may, in turn, induce the occlusion of many shunts in the nidus. Because arteriovenous fistulas compose most brain AVMs with plexiform nidi and upstream shunts, arterial occlusion without obliteration of the venous drainage of brain AVMs can lead to recurrence. 25 In the report of Viana et al, 26 1 patient in whom immediate angiographic occlusion was not achieved showed spontaneous occlusion at the 6-month follow-up, 26 which was also observed in this study.
Brain AVM-related events can be controlled once the AVM nidus has been obliterated. A case series demonstrated no events occurring in the 6-month follow-up in 11 patients. 26 Another report indicated that only 1 (2.5%) patient had significant disability at the 6-month follow-up, but unfortunately, no detailed information was provided. 21 In this study, 1 seizure occurred 3 months after the operation.

Procedural Safety
Previous reports on the transvenous embolization of AVMs showed low rates of periprocedural complications, disability, and morbility. 21,27 In contrast, the complication rate was relatively high in our study. There are several possible reasons: First, the present study enrolled only patients with ruptured AVMs (100%), which had a higher risk for rebleeding. 1 In the report of Mendes et al, 21 the rate of ruptured AVMs was only 67.5%. Second, we enrolled more patients with higher Spetzler-Martin grades, and 66.7% of our patients were classified as Spetzler-Martin grade III or even higher. The percentage of higher Spetzler-Martin grades was 58.5% in the report of Mendes et al, 21 and only 25% in the report of Trivelato et al, 11 which are architecturally more complex.
Infarctions caused during transvenous embolization are supposed to be related to either penetration of Onyx into the feeding artery of the normal brain tissue or hyperemia and edema caused by draining vein occlusion. The outcome may depend on the local collateral circulation. 2,28 In this study, 1 cerebral infarct that caused disability was thought to be due to the occlusion of the feeding artery with Onyx. Trivelato et al 11 and Mendes et al 21 reported 2 cases each of postprocedural parenchymal edema thought to be related to venous outflow obstruction, one of which caused a disability.
Hemorrhagic complications might occur in different locations. Among the 5 hemorrhagic complications, 4 occurred in the eloquent area and 4 with deep venous drainage, which are also the risk factors for transarterial embolization. 29 All 5 AVMs with hemorrhagic complications were located in deep brain structures and around ventricles and bled into the ventricles, which is not a usual complication of transarterial embolization. Periventricular location has been cited as a risk factor for hemorrhage. 30 These hemorrhages may be related to the following reasons: 18 high arterial input pressure and venous outflow restriction in the deep AVMs; not enough brain tissue around the ventricle; and a large pressure gradient between the nidus and ventricle. Premature occlusion of draining veins during transvenous embolization before complete occlusion of the nidus can increase the pressure in the AVM nidus, 31 which has been reported as one of the main causes of hemorrhagic complications. 32 Therefore, transvenous embolization is considered an "all-or-nothing" technique. In this study, 3 AVMs had residual nidi postembolization, with hemorrhages occurring in 2. The 1 AVM with a residual that did not bleed had continued drainage through a cortical vein after the deep drainage vein had been occluded, which may explain the lack of hemorrhage postprocedure.
Most of the patients in this study with complications (4/6) had good outcomes after receiving appropriate treatment. For the ischemic complications, the outcome depends on the location and size of the infarction, especially whether it is in an eloquent area. In this study, the patient with infarction involving the midbrain and thalamus did not recover well. For the other 5 hemorrhagic complications, 4  Note:-þ indicates yes; À, no.
patients had good outcomes and 1 died due to severe intracranial infection after hemorrhage. This might be related to several factors: First, intraparenchymal hemorrhage often occurs in the same area as prior hemorrhage and is less likely to immediately damage healthy brain; second, the subarachnoid hemorrhage can be treated by drugs or ventriculostomy; and finally, the patients were relatively young in this study.

Limitations
Although the number of cases in this study was the second largest in all the publications currently available, 21 the sample size was still small. The patients were carefully selected and were selected for salvage procedures due to lack of good microsurgical, transarterial, and radiosurgical options, which may imply greater propensity for procedural risk than in other studies. Treatment strategy, patient factors, and doctors' experience are highly individualized and variable worldwide; thus, the findings of our study may not be consistent with other centers.

CONCLUSIONS
Transvenous embolization with high complete obliteration rates, improved functional outcomes, and acceptable morbidity and mortality rates can be performed in only carefully selected brain AVMs. However, more experience is necessary to discern the role of this technique in the management of ruptured AVMs, and it should not be considered a first-line treatment.