Quality of life (QoL) after the treatments has been emphasized in recent years. The incidence of neurological deficits is regarded as a critical index in assessing the safety and efficiency of treatment. Thus, more attention had been paid to developing interventional techniques for neurological protection, and predictive tools for therapeutic risks evaluation (32, 40, 41, 44, 46, 49–55). Treatments of eloquent bAVMs have to achieve two goals: (1) the complete obliteration of nidus and arteriovenous shunt; and (2) the protection of neurological functions, which might severely impact postoperative QoL.
Treatments of eloquent bAVMs are carried by three elementary surgical methods, including microsurgical resection, endovascular embolization, and stereotactic radiosurgery (SRS). Microsurgical resection can be performed initially or subsequently to other treatments. Complete obliteration may be achieved in most cases receiving resection. Endovascular embolization is usually performed as a precursor to microsurgery and radiosurgery. Complete endovascular obliteration is seldom reported to be achieved in a single-staged or multi-staged therapy (56). Stereotactic radiosurgery is applied to the bAVMs in deep locations or small sizes as a primary or complementary treatment. In the treatment of complex eloquent bAVMs, different therapeutic elements are usually cooperatively utilized (AKA multimodality treatments). Reviewing current literature, indications of each therapeutic modality were concluded.
Microsurgery
Microsurgical resection is the most common approach to achieve the complete obliteration of bAVMs. A systematic review by van Beijnum et al. (56) reported that microsurgery achieved the highest complete obliteration rate (≈96%), comparing with that of endovascular embolization (≈13%), and stereotactic radiosurgery (≈38%). The complete obliteration of bAVMs can eliminate the morbidities and mortalities induced by its hemorrhage in the future.
Threshold of Microsurgery
Microsurgical resection is not recommended for all eloquent bAVMs. The neurological risk of microsurgical resection should be evaluated. As mentioned above, the SM grading system and its supplemented scale are useful in the evaluation of operative risks and the prediction of neurological outcomes (41, 43, 51, 57, 58). Five grades of the SM grading system are divided into three levels, low grade (grade I–II), medium grade (grade III), and high grade (grade IV–V) (3). Studies on low-grade bAVMs reported morbidities of neurological deficits to range from 0 to 6.6% after microsurgical resections (57, 59–66). The morbidity rate of bAVMs in SM grade II increased to 0–69.2% (65, 67, 68). However, the worst neurological outcome was not induced by the involvement of eloquence. The microsurgery on eloquent bAVMs in SM grade II resulted in a morbidity rate ranging from 0 to 9.5%. There were two subtypes of eloquent bAVMs in grade III, the subtype of S1E1V1 and S2E1V0. The morbidity of neurological deficits was reported to be 4.8–16.7% and 15–25%, respectively (66, 69–71). They were in similar rates with the subtype of S2E0V1, but much lower than that of S3E0V0 (70). In the eloquent bAVMs in SM grade IV and higher, neurological risks increased rapidly to as high as 38% in the mono-therapy of microsurgery (40, 51, 57, 72, 73). However, preoperative annual hemorrhagic rates ranged from 1.5 to 10.4% in lesions of SM grade IV and V (72, 74, 75), which suggested the need for microsurgery. Several case reports and series had reported the successful utilization of individualized multimodality treatment to cure high-grade bAVMs with satisfactory morbidity and mortality (76–85). Multimodality treatments refer to the combined therapies of more than one elementary surgical treatment. Microsurgery is involved in most combinations. The performing of multimodality treatment is primarily for tentative or salvage purposes for high-grade bAVMs. Further discussed referred to section Multimodality Treatment.
Lawton et al. (41) proposed the supplementary scale of the SM grading system to refine the prediction of neurological outcomes. The full scale (SM grading system + supplementary scale) had been validated in bAVMs with deep and superficial locations. According to the studies with the full scale applied, monotherapy of microsurgery could result in satisfactory neurological outcomes in eloquent bAVMs in grade V and the lower (58, 86, 87). For the lesions in higher grades, not enough data support the exclusive utilization of microsurgery.
Results of functional neuro-images help to predict the individualized neurological outcome of each eloquent bAVMs. Lin et al. (88) proposed the minimum lesion-to corticospinal tract distance (LCD) to be 5 ml to secure the motor functions. Afterward, Jiao et al. (32) proposed the HDVL scale, enrolling LED, and achieved higher predictive accuracy than the full scale. The worsening of neurological outcome occurred in 0% of lesions in HDVL grade I and II, 11.8% in grade III, 31.5% and more in grade IV–VI. Monotherapy of microsurgery was recommended in lesions < HDVL grade IV.
The brainstem is a critical location with dense fiber tracts and nervous nuclei. Different procedures and techniques had been piroposed to cure brainstem bAMVs (89–91). However, current studies only demonstrated the therapeutic outcomes of highly selected patients. Thus, microsurgery of bAVMs in critical locations is not recommended, unless the relevant progressive neurological deterioration or hematoma occupation could not be postponed by endovascular surgery or radiosurgery (92).
Recommendations
The Spetzler-Martin Grading system and its supplementary grading system are recommended to be utilized primarily to evaluate the risk of microsurgical resection (Class I; Level of Evidence B). The HDVL grading system is recommended for patients who have received DTI and fMRI assessments (Class I; Level of Evidence B). Microsurgical resection is reasonable to perform on lesions under Spetzler-Martin grade IV, or the combined grade VI (Spetzler-Martin grade plus the supplementary grade), or HDVL grade IV (Class IIa; Level of Evidence B). Individualized multimodality treatments can be useful to the bAVMs with Spetzler-Martin grade ≥IV, or combined grade ≥VI, or HDVL grade ≥IV (Class IIa; Level of Evidence C).
The microsurgical resection may be considered in bAVMs located in critical locations (brainstem, pons, medulla, mesencephalon, etc.) when bAVM-related neurological deficits or mass effect of hematoma are progressive, and cannot be postponed by endovascular or radiosurgical treatments (Class IIb; Level of Evidence C).
Elective, Semi-elective, and Emergency Microsurgery
For eloquent bAVMs that require microsurgery, timing of treatment differs across situations. The emergency operation is performed in urgent situations, such as the occurrence of life-threatening hematoma, without any delay to prevent death or serious disabilities. The semi-elective operation is performed on the patients with prior hemorrhagic presentation, progressive neurological deficits, or AEDs-resistant epilepsy to prevent deterioration or death. The semi-elective operation should be done as early as possible, but can be postponed for the thorough preoperative preparation and evaluation. The elective operation is performed to the patients without life-threatening risk, and carried out at the request of the patient, and availability of the surgeon and facility. Both elective operations and semi-elective operations are aiming at preventing the onset of (re-)hemorrhage or symptoms in the future to promote QoLs. To ensure optimal neurological outcomes, the criteria in the last section should be obeyed.
Intracranial hemorrhage was the major threat of bAVMs. It occurs in an annual risk ranging from 1.3 to 4.1% (93–97). For the bAVMs with rupture history, the annual hemorrhagic risk increases to 4.5–4.8% (94, 95), and as high as 6–15.8% in the first year (98–102). Angio-architectural Features had been reported to be associate with ruptures, including (1) aneurysms located in nidus, arterial feeders, or irrelevant arteries; (2) venous drainage anomaly, such as the stenosis, occlusion, ectasia, kinking, or reflex of draining veins, and occlusion of the sinus; (3) single arterial feeder with high blood flow (94, 103–106). Microsurgical resection could achieve complete obliterates in most cases (96% approximately), and significantly reduce the re-bleeding rate (107). Considering the high risk of rebleeding in the first year, microsurgery is recommended to perform, but can be postponed for a short time for preoperative preparations and assessments.
Another common presentation of bAVMs is seizure, occurring in ~17–30% of patients (108). The onset of seizure was suggested to be associated with hemosiderin deposition, mass effect with cortical irritation, hemodynamic modifications, and/or vascular remodeling leading to stealing, ischemia, and neuronal damage (109). The seizures could be progressive and impact QoLs. Josephson et al. (110) reported that 76% of seizures would develop into epilepsy. Microsurgical resection was reported to be effective in the control of seizures in bAVM patients with refractory epilepsy (111, 112). The semi-elective microsurgery is recommended when the onset of seizure had progressed. Other presentations of bAVMs consist of neurological deficits due to the steal phenomenon, and headache. In these hemodynamic related symptoms, microsurgical resection had been reported to be effective for their relief (113–117). For asymptomatic bAVMs, elective microsurgical resection is considered to prevent the occurrence of symptoms and suboptimal events mentioned above.
In patients aged ≤ 40 years, 33% of intracerebral hemorrhages were caused by the rupture of bAVMs (118). In the acute phase of hemorrhage, clinical outcomes were associated with both the grade of bAVM and the degree of subarachnoidal hemorrhage (119). Kuhmonen et al. (120) reported that an early extirpation of bAVMs and evacuation of massive hematoma resulted in optimal outcomes in over 55% of patients. However, before evacuating hematoma or resecting bAVM, two points should be clear. Firstly, the etiology of a hematoma should be identified, such as bAVM rupture, hypertension, or amyloid (4). Secondly, if the hematoma is induced by bAVM rupture, the morphology and angioarchitecture of the lesion should be ascertained comprehensively. CTA or DSA had been proven effective on the etiological diagnosis of hematoma (121–124). However, the routine workflow usually takes a long door-to-operation (DTO) time, which may be intolerant to patients with life-threatening hematoma or progressive neurological deterioration. The hybrid-modality treatment integrates endovascular intervention (including catheter angiography) with microsurgical operations in one operating room, which could effectively shorten the DTO time of emergency patients. Hybrid-modality treatment had been proven to be feasible in the emergency disposal of severe trauma (brain trauma included), complex thoracoabdominal aortic pathology (125–127). Under a definitive diagnosis, emergency evacuations of hematoma and control of acute bleeding is warranted in the event of life-threatening mass effect, regardless of whether it is associated with the bAVM. Superficial bAVMs in small size (≤ 3 cm) can be resected simultaneously in an emergency operation. Meanwhile, bAVMs in deep locations or with large sizes are recommended to be resected via semi-elective microsurgical operations.
Recommendations
The semi-elective operation is reasonable for patients with any of the prior hemorrhagic presentations, progressive neurological deficits, or AEDs-resistant epilepsy (Class IIa, Level of Evidence B). A semi-elective operation is reasonable for bAVMs with angioarchitectural features, which imply high rupture risks (Class IIa, Level of Evidence B). Elective operation is probably recommended for patients without any features above (Class IIa, Level of Evidence B).
In an emergency operation, the evacuation of hematoma and control of acute bleeding can be beneficial to the event of life-threatening mass effect, regardless of whether it is associated with the bAVM (Class IIa, Level of Evidence B). Simultaneous resection of superficial bAVMs in small size (≤ 3 cm) in the emergency operation is reasonable (Class IIa, Level of Evidence C). The semi-elective microsurgical operation is probably recommended for the deep located or large-sized bAVMs (Class IIa; Level of Evidence C).
Hybrid-modality treatment is probably recommended for the diagnosis and subsequent treatment of every emergency intracranial hemorrhage (Class IIa; Level of Evidence C).
Strategy of Microsurgery
Single Microsurgery
The single microsurgical resection of eloquent bAVMs shares the same procedural process with the non-eloquent: (1) performing the craniotomy to expose bAVM and relevant vessels; (2) isolating and controlling the arterial feeders; (3) dissecting the nidus along edges from adjacent parenchymal and vascular structures; (4) coagulating and dissecting the draining veins; and (5) closing and suturing up the incision. In the microsurgery of eloquent bAVMs, steps from 2 to 4 are nuanced with the general ones, because of the adjacent or overlapping relation between nidus and eloquent structures. Approaches to eloquent bAVMs in deep locations, such as the insula, basal ganglia, thalamus, and callosum, should be planned precisely to protect the surrounding eloquent cortex and subcortical parenchyma (86, 128, 129). The circumferential dissection of the nidus should be limited as well, to keep the eloquence-related structures intact. Steiger et al. (90) reported their procedure to limit the damage to parenchyma around in a small group of cases, which was proposed to coagulate from draining veins to the nidus. Other microsurgical or endovascular techniques should be considered for the protection of eloquent structures.
Microsurgery in Multimodality Treatments
Multimodality treatments refer to the performing of different elementary surgical therapies in a single stage or multiple ones. In staged-multimodality treatments, the microsurgical resection is commonly performed as a conclusive treatment, or as an initial treatment in a few cases as well (see section Detection and Treatment of Residue). As a conclusive treatment, the microsurgical resection is utilized as a salvage treatment to residual bAVMs, that has not been completely obliterated by other treatments (78, 130). Meanwhile, the microsurgical resection can be performed systematically in some bAVMs, which have been down-graded by endovascular embolization or/and SRS from higher grades (76, 131–133). Studies on staged-multimodality treatments suggested that the prior endovascular or/and radiosurgical treatments could improve the microsurgical condition with lower risks and difficulties. However, it should be noted that the risks of prior treatments needed to be acknowledged. The most commonly utilized paradigms are microsurgical resection combining with prior endovascular embolization. The prior embolization is supposed to decrease the operative risks of subsequent microsurgery in the following aspects: (1) aiding the elimination of feeders from deep sites of the operative field (for example, perforating arteries and branches of posterior cerebral arteries), and (2) making the nidus dissection easier with clearer borders, which is important to diffusive nidus adjacent to eloquence (79, 133, 134). Besides, it has been proposed for neurological protection. Han et al. (91) and Wang et al. (84) reported their experience of applying the staged or hybrid paradigms to protect the neurological function of bAVMs in brainstem and eloquence, respectively. In their procedures, the nidus adjacent to eloquence had been embolized through the prior endovascular manipulations. The subsequent microsurgery only removed the nidus distal to the eloquence, while leaving the embolized nidus in-situ. Another paradigm of microsurgical resection combining with prior SRS was reported in a few cases. The prior SRS decreases the risk of microsurgery as well, by a different pathological process from endovascular embolization (76, 135). It usually requires probation of 3–5 years to evaluate the effect of the prior SRS before microsurgery (136, 137). However, the risks of prior treatment would not be diminished by the cooperation of multimodality treatments. Prior embolization and SRS were reported to share a similar rate of adverse events, morbidity, and mortality with their independent implementations (76, 79, 82, 132, 138). Most of the suboptimal events occur in the process of endovascular procedures, or the latency period before microsurgery.
Hybrid-modality treatment is the up-to-date paradigm, which combines endovascular and microsurgical procedures in a single stage without any interval (81, 139). It has been proved to be feasible in treating bAVMs, especially the eloquent ones (80, 84). The hybrid-modality treatment possessed the advantages of endovascular and microsurgical manipulations, and expanded the operative techniques. Wang et al. (140) reported their intraoperative transvenous embolization technique to patients with difficult arterial and venous approaches. Most importantly, latent risks in the intervals of staged treatments were fundamentally diminished in the hybrid-modality treatment. Brown et al. (141) reported their experience in 19 patients who received hybrid-modality treatment, in which neurological outcomes were similar with staged multimodality treatment without the occurrence of intracranial hemorrhage after embolization or microsurgery. Hybrid-modality treatment may be a safer method of curing eloquent bAVMs radically and effectively, but need to be further validated.
Recommendations
Multimodality treatments with microsurgery involved are probably recommended for the treatment of eloquent bAVMs in high grades. Downgrading the lesion before microsurgery can be useful to decrease operative risks and protect neurological functions (Class IIa; Level of Evidence B).
The staged multimodality treatment of the microsurgery subsequent to endovascular embolization can be useful for treating bAVMs with diffusive nidus or feeders from deep sites of the operation field, which may reduce the risks of intraoperative bleeding and neurological deficits of the subsequent microsurgery (Class IIa; Level of Evidence B).
Hybrid-modality treatment is probably recommended with similar safety and efficiency to the staged multimodality treatment, but with fewer risks than the latter (Class IIa; Level of Evidence C).
Detection and Treatment of Residue
Although microsurgical resection has achieved the highest complete obliteration rate among treatments, residual nidus can still be detected in follow-up angiographies. The cumulative incidence of residuals and recurrences had been reported to be 3% for SM grade I-II and 8% for SM grade III or higher (142, 143). The cause of residue is induced by the absence of adjunctive tools for an angiogram or bAVM detection. Most residues can be quickly detected through intraoperative DSA (80, 139, 144). Meanwhile, omissions might occur in intraoperative DSA. Aboukais et al. (145) reported that intraoperative or early postoperative angiography did not ensure the cure of bAVMs in several pediatric cases. Residues may be caused by vasospasm or recanalization of abnormal vessels. Delayed angiographies, via DSA, CTA, or MRA, were suggested in follow-ups, especially to pediatric patients (see section Follow-Up). It should be noticed that residual dysplastic vessels after cerebral arteriovenous malformation resection might be confounded with residual nidus (146). Residual dysplastic feeding vessels without an early draining vein do not necessarily represent residue after resection.
Intraoperative Doppler ultrasound can be used to detect the residual nidus. Four parameters could be obtained, including peak systolic velocity (PSV), end-diastolic velocity (EDV), mean flow velocity (MV), and resistance index (RI). RI (defined as the ratio of PSV to EDV) is regarded as the key parameter to identify the residual nidus. It has been generally accepted that RI is between 0.55 and 0.75 for the normal internal carotid artery, and higher in branches downstream (147). Arterial vessels with an RI lower than 0.55 should be noticed. Griffith et al. (148) reported their RI range of arterial feeders to be 0.14–0.50. With the enhancement of contrast, ultrasonographic angiography is able to help identify arterial feeders both on the surface and deep in the tissue (149, 150). Doppler ultrasonography cannot replace the intra/post-operative DSA. Regions of low-velocity blood flow or areas with very small vessels (i.e., much smaller than 0.6 mm in diameter) might be missed by ultrasonography. Hence, abnormalities such as venous angiomas on cryptic AVMs with low-velocity blood flow or thrombosed vessels may not be discernible in sonography. So the intraoperative ultrasound findings should be confirmed by angiography (147).
Patients with residual and recurrent nidus after microsurgery are still exposed to the threats of hemorrhage throughout their lives (135, 151). Only a few case-reports demonstrated the spontaneous thrombosis occurring in residual nidus (152, 153). The expectant or conservative treatment of residual bAVMs has not been widely accepted by neurovascular surgeons (152). The residue of bAVMs is usually disposed of through microsurgery, endovascular embolization, and SRS. A salvage microsurgical resection could be performed in one session with the assistance of intraoperative DSA (144, 154, 155), or another operation (146). Neurological risks of salvage microsurgery should be considered. SRS and endovascular embolization were proposed to be utilized for the salvage treatment of residues adjacent to eloquence or in a deep location (128, 156–158). Indications of endovascular embolization and SRS should be obeyed, respectively.
Recommendations
Intraoperative or early postoperative DSA is recommended for detecting residual nidus (Class I; Level of Evidence B).
Intraoperative Doppler ultrasonography is useful to primarily detect the residual nidus (Class I; Level of Evidence B). The complete obliteration should still be confirmed by a subsequent DSA.
Repeat angiographies in follow-ups within 3–5 years after surgical resection is reasonable for detecting missed residues or recurrence (see section Follow-Up) (Class IIa; Level of Evidence B).
Residual nidus after microsurgery is still exposed to the risk of hemorrhage. Salvage surgical treatments are recommended for the disposition of residues, including microsurgery, endovascular embolization, and SRS (Class I; Level of Evidence B). Subsequent SRS and endovascular embolization can be effective in achieving complete obliterations in residues adjacent to eloquence, following their indications, respectively (Class IIa; Level of Evidence C).
Surgical Adjuncts
Surgical adjuncts are indispensable for intraoperative localization and mapping of eloquent (sub-)cortical structures, including neuro-navigation, electrical cortical stimulation with awake craniotomy, and transcranial magnetic stimulation.
Most neuro-navigations are utilized based on structural and functional MRI. Data reconstructions should demonstrate the 3-dimensional information of brain, lesion, and eloquent structures. Despite providing a wealth of information, neuro-navigation was suggested to be ineffective in improving the neurological outcomes of microsurgeries in a randomized controlled trial (159). This was attributed to the different resecting strategies of bAVMs and brain tumors. The resections of a tumor could be stopped when reaching the edge of eloquence. By contrast, the resection of bAVMs cannot be stopped until completely removing the nidus. The alert of neuro-navigation hardly affects the damage to eloquence. However, neuro-navigation is not useless in the microsurgery of bAVMs. Torne et al. (160) reviewed the bAVMs surgically treated by Michael Lawton and emphasized the importance of identifying the location and border of the nidus, which could reduce the rupture risk during operation and improve clinical outcomes. Although neuro-navigation was proved to be ineffective in improving neurological outcomes, it was helpful to reduce operative risks by clearly demonstrating the localization and borders (88, 161–163). Intraoperative three-dimensional (3D) ultrasound was usually used as an assistance to the neuro-navigation to correct the brain-shift induced by craniotomy (164–168).
Intraoperative direct electrical stimulation (DES) on the cortex is regarded as the gold standard of eloquent mapping (169). It allows a safe real-time identification and hence preservation of essential pathways for motricity, sensibility, language, and even memory in the treatment of brain tumor and cerebrovascular diseases under general or local anesthesia (awake craniotomy) (170–172). Currently, the mapping through DES was usually performed on basis of fMRI (173, 174). Besides the direct stimulations to the cortex, Gamble et al. (175) revealed the value of subcortical stimulation on identifying subcortical eloquent structures. Concluding the proposals of current studies, the electrical cortical/subcortical stimulation was recommended if the lesion (1) adjacent to eloquence on fMRI (not distancing nor overlaying); (2) in large sizes and high SM grades; (3) with diffusive nidus. Differences existed in the application of DES between general and local anesthesia. The DES under local anesthesia had become popular in recent years. It was capable of mapping not only essential cortico-subcortical areas of motricity, but also areas of sensitivity, language, and even memory (174, 175). It had been proved to be effective in helping identifying eloquence and preserving neural functions (171, 176). Although DES under local anesthesia could make awake patients without any pain or discomfort (177, 178), the psychological effect still needed concern. By contrast, the DES under general anesthesia was performed earlier, usually accompanying with electrocorticography (ECoG) and neurophysiological monitor. Its capability of eloquent mapping was limited to motor and somatosensory areas, in which optimal neurological outcomes could be achieved (170, 171). The ECoG, which used accompanying DES, was proved to be effective in identifying epileptogenic cortex for subsequent surgical management (179, 180). Despite its high accuracy and efficiency, DES faces a similar situation with fMRI-based neuro-navigation. Mapping of the eloquent area could take effect as an alarm to the neurosurgeon, but helpless to limit the extensive excision. Besides, DES might cause generalized seizures and result in disastrous consequences (171). The utilization of DES remains controversial and needs further investigation. The requirement of gross identifying of motricity and somatosensory area can be met by a neurophysiological monitor.
Navigated transcranial magnetic stimulation (nTMS) is applied to the preoperative mapping of cortical motor and language regions in recent years. The comparison between nTMS and DES resulted in similar accuracies (181–185). Besides, nTMS had been proved to be superior in mapping the eloquence to other non-invasive techniques, such as fMRI and magneto-encephalography (185–187). Ille et al. (188) reported their experience of utilizing nTMS to fix the mapping result of fMRI. Kronenburg et al. (189) reported the utilization of TMS to non-cooperative patients and achieved satisfactory results. Although nTMS is widely applied in brain tumors, it has seldom been used for cerebrovascular disease. The preparation and implementation of microsurgery are different between tumors and bAVMs. More studies are needed to validate its safe utilization to bAVMs, especially on seizure and hemorrhage-related complications.
Recommendations
The utilization of fMRI-guided neuro-navigation does not improve neurological outcomes, but can be useful for locating and delineating the lesions (Class IIa; Level of Evidence B). Intraoperative 3D ultrasound is effective in correcting the brain-shift induced by craniotomy (Class I; Level of Evidence B).
Intraoperative direct electrical stimulation (DES) is feasible on mapping motricity eloquence under general, and other sensibilities, language, and even memory eloquence under general anesthesia. DES might be considered for delineating the nidus adjunct to eloquence, or in large sizes, or with diffusive nidus. The risks and psychological impacts of DES are not well-established (Class IIb; Level of Evidence C).
Neurophysiological monitor can be useful for gross mapping of motricity and somatosensory areas (Class IIa; Level of Evidence C).
Electrocorticography (ECoG) is recommended for localizing epileptogenic cortex (Class I; Level of Evidence B).
Navigated transcranial magnetic stimulation (nTMS) has been utilized for mapping and fixing mapping results of fMRI in the preoperative preparation, mostly in brain tumors and seldom in bAVMs. The risks of nTMS are not well-established (Class IIb; Level of Evidence C).
Endovascular Neurosurgery
Endovascular embolization (EE) occludes blood flow by delivering occlusive agents into the feeding arteries and nidus through microcatheters (1, 190). The role of endovascular embolization in treating bAVMs includes (1) as a therapeutic strategy; (2) as palliative or targeted strategies; and (3) as adjuvant management before microsurgical resection or SRS, to minimize the risk of hemorrhage (191, 192).
Curative Embolization
Monotherapy of curative embolization was believed to be difficult to completely eliminate bAVMs, with rates of angiographic obliteration ranging from 13 to 96% (56, 193–195). Complete cure was attainable for small-sized (≤ 3 cm), superficially located bAVMs (194). Angioarchitectural characteristics with a single feeding artery also achieved favorable outcomes (196). Properly applied EE might decrease the size and grade of bAVMs without sudden changes of pressure and reduce the risk of adjacent arterial recruitment (106). The overall complication rate of endovascular therapy for bAVMs was 25.0%, with an incidence of 6.6% in permanent neurological deficits (56). The number and diameter of feeding arteries, nidal volume, deep venous drainage, and eloquent location were risk factors of embolization-related complications (47, 54, 55). When disposing of bAVMs in corpus callosum with complex angioarchitectures and eloquence involved, curative embolization achieved complete obliteration in only 40–60% of cases and hemorrhage complications occurred in 7% of cases (197, 198).
Strategies of embolization have been discussed. Sahlein et al. (199) proposed that the single-stage embolization reached a lower rate of mortality and morbidity than the multi-staged embolization. It had been demonstrated that staged embolization was an independent risk factor for unfavorable outcomes after embolization (200, 201). It was supposed to attribute to the recanalization and recruitment of arterial feeders during staged procedures (199). Even so, the staged embolization had been used as a strategy to reduce the risk of normal perfusion pressure breakthrough by progressively minimizing the blood flow, particularly for medium-large bAVMs (>3 cm) (202, 203). Previous studies had supported that the interval between each session should be 4–6 months when applying the staged embolotherapy (202, 203). Ma et al. (204) reported that staged embolization was effective in treating eloquent bAVMs with large sizes. Different paradigms of staged embolization were proposed without a consensus. Ma et al. (204) reported their paradigm, which achieved an obliteration rate within 60% in the initial session, and achieved complete occlusion in 2–3 months through the following 1–2 sessions. Katsaridis et al. (202) reported another paradigm, which proposes to embolize ≤ 30% of nidus in each session. The optimal paradigm of endovascular embolization remains to be researched.
Recommendation
Curative embolization of small-sized (≤ 3 cm), superficially located bAVMs with a single feeding artery can be useful (Class IIa; Level of Evidence B).
Staged embolization can be beneficial for curing medium-large bAVMs (>3 cm) by gradually reducing the risk of NPPB, but the probability of recanalization remains (Class IIa; Level of Evidence B).
The interval between each stage and extent of embolization has not been well-determined (Class IIb; Level of Evidence C).
Palliative Embolization
The selective embolization of high-flow feeding arteries might postpone the progression of frequent seizures or neurological deficits caused by venous hypertension and arterial steal syndrome (106). A retrospective study suggested that the partial embolization might result in complete occlusion and improve survival rates, comparing with conservative treatment in the long-term (205). Flores et al. (206) proposed to utilized the palliative embolization to symptomatic bAVMs in Spetzler-Martin grade IV or V, that are inadvisable for surgical resection or SRS. Although the palliative embolization might improve the clinical manifestation of patients by changing the hemodynamics of lesion (207, 208), it was accused of increasing the risk of hemorrhage in large bAVMs (75). The role of palliative embolization remains controversial.
Recommendation
Palliative embolization remains controversial. The selective embolization of high-flow feeding arteries might be considered to postpone the progression of seizures or neurological deficits in surgical/SRS-inadvisable bAVMs, while potentially increasing the risk of hemorrhage (Class IIb; Level of Evidence B).
Targeted Embolization to “Weak Point”
Targeted embolization aims at eliminating the “weak point” of bAVMs, including intranidal or flow-related aneurysms, high-flow arteriovenous fistulas, and venous flow obstruction, and other anigoarchitectural features with high rupture risks (3, 101, 106, 209). Targeted embolization is proposed to perform when definitive managements were infeasible or excessively risky. Targeted embolization was reported to be performed to the ruptured bAVMs in the acute phase (210), and restoration stage (211). The effect of embolization in the management of ruptured bAVMs has not been well-established. The targeted embolization was supposed to reduce the incidence of bleeding after radiosurgery (106).
Recommendation
Targeted embolization is recommended for bAVMs with a “weak point” when definitive treatments are infeasible or too risky (Class I; Level of Evidence B).
Pre-microsurgical Embolization
Pre-microsurgical embolization has been used as the most common adjunct to improve the therapeutic safety and efficiency of bAVMs (79, 80, 199, 212). The pre-microsurgical embolization resulted in permanent morbidity of 2.5% and a mortality of 2.0% (201, 213). The strategy was proposed to facilitate the subsequent microsurgery in the following aspects: (1) occluding the supply arteries and lowering the size of bAVMs, to minimize the risk of bleeding, (2) eliminating the deep perforators that are inaccessible for surgeries, (3) embolizing the flow-related aneurysms (56, 82, 195, 206, 214, 215).
The prior embolization before microsurgery was proposed to minimize the risk of NPPB by gradually reducing the blood flow before surgery, and normalize the hemodynamics for high-flow or large lesions (202, 216–218). Whereas, the vigilance of the potential risks (e.g., hemorrhage, infarction, or seizures) in-between the treatments should be considered as concerns (219, 220). The subsequent microsurgery could be performed consecutively after the embolization in one session, which was known as a hybrid-modality treatment (80). It made the manipulations of embolization more flexible, and overcame the disadvantages and limitations when solely performed. It had been shown that the staged paradigm increased the expenditure of treatments (200, 221). The hybrid-modality treatment might reduce the frequency and duration of anesthesia and operation by a single treatment, from a health-economics perspective. However, the optimal paradigm of preoperative embolization and subsequent microsurgery remained unclear (82). The previous series had performed a paradigm with an interval between each embolization therapy for 16–42 days, and an interval before surgery for 1–42 days (79, 212, 222, 223). Nataraj et al. (224) supported a prompt microsurgical resection after endovascular intervention for a lower rate of mortality and morbidity.
Recommendation
The pre-microsurgical embolization is probably recommended for the bAVMs with large size, or deep perforating feeding arteries, or inaccessible locations for surgeries, or concomitant flow-related aneurysms (Class IIa; Level of Evidence B).
Pre-radiosurgical Embolization
The pre-radiosurgical embolization has been proposed to decrease large bAVMs to a suitable size for the subsequent SRS. It was particularly recommended for the lesions ≥3 cm in diameter, or the lesions with relevant aneurysms or high-flow fistulas. A pre-radiosurgical embolization could minimize the risk of hemorrhage before the definitive obliteration by SRS, especially effective for ruptured bAVMs in the posterior fossa (1, 3, 42, 225, 226). The multimodality treatment, consisting of prior embolization and subsequent SRS, was reported to achieve a complete obliteration rate of more than 60% in large bAVMs (227, 228). The efficiency of the subsequent SRS might be improved if the volume of the residual nidus is no more than 10 cm3 after embolization (77). The drawbacks induced by prior embolization needed to be considered. Embolic agents might shield the nidus from the radiation as protectants and make the outlines obscure with subsequent targeting inaccurately (229). Additionally, embolization might decrease the rate of obliteration by facilitating angiogenesis (230). The recanalization of embolized arteries might result in delayed recurrence (77). Due to the limitations, this paradigm might worsen the outcome of bAVMs (231–233).
Eloquent regions of basal ganglionic and thalamic AVMs could be treated with embolization in conjunction with SRS. Complete obliteration was observed in 14.3% and improved disabling in more than 1/3 of patients (234). Also, selected brainstem AVMs could be treated with embolization combined SRS, while the selection criteria had yet to be determined. Favorable outcomes were potentially comparable with general bAVMs under precision techniques (235).
Recommendation
Pre-radiosurgical embolization may be considered for reducing the size of bAVMs, particularly for large lesions (>3 cm), or occluding bAVM-related aneurysms or high-flow fistulas, whereas the effectiveness remains uncertain (Class IIb; Level of Evidence B).
Seizure Control
Endovascular embolization had been reported to be less effective in controlling bAVM-induced seizures. Hyun et al. (236) investigated 399 bAVM patients with long-term follow-up after embolization, and found out that only half of the patients achieved seizure-free status after embolization, compared with 78 and 66% in the surgical and SRS groups, respectively. The duration of seizure-free status was 8.1 months in the embolization group and 20.5 months in the SRS group. A meta-analysis demonstrated that embolization resulted in the highest morbidity of new-onset seizures (39.4%), compared with other treatments (111).
Recommendation
Embolization is not indicated for seizure control (Class III; Level of Evidence A).
Embolic Agents
Varieties of embolic agents are available in the endovascular treatment of bAVMs, consisting of adhesive, non-adhesive, and solid ones.
The adhesive embolic agents include N-butyl 2-cyanoacrylate (NBCA) and NBCA metacryloxysulfolane (NBCA-MS). Both NBCA and NBCA-MS have been proved to be efficient and safe in treating bAVMs with sophisticated endovascular skills (237–239). Compared with NBCA, the main advantage of NBCA-MS has a longer polymerization time (NBCA vs. NBCA-MS = 15–40 s vs. 60–90 s), which provides a sufficient and precious time window for its diffusion in bAVMs (240).
The non-adhesive embolic agents mainly include Ethylene vinyl alcohol (EVOH) and ethylene vinyl copolymer (EVAL). Comparing with NBCA, EVOH achieved indifferent results in its safety validations (241, 242), and was proved to be higher efficiency in occluding arterial pedicles (221). The comparative study on EVAL has not been widely conducted. Although the non-adhesive ones have better performance in endovascular embolization, both EVOH and EVAL have to be used accompany with dimethyl sulfoxide (DMSO), the latter of which might induce a series of side effects due to its vascular toxicity (243).
The success in embolization relies on the appropriate choice of embolic agent. the polymerization speed of NBCA could be reduced by adding lipiodol. NBCA is usually used in a concentration of 16–50%. Higher concentration results in higher polymerization speed. Commercially available EVOH is premixed in different concentrations, including 6 and 8%. Higher concentration results in higher polymerization speed. Choo et al. (244) reported their experience of using EVOH and NBCA in high concentration and coil to an embolize dural arteriovenous fistula. EVOH embolization was proved superior to NBCA and coil embolization in completely obliterating DAVFs.
Recommendation
Non-adhesive embolic agents are probably recommended to the adhesive ones for the complete embolization of bAVMs (Class IIa, Level of Evidence C).
For bAVMs with high flow capacity, a non-adhesive embolic agent in high concentration is probably recommended (Class IIa, Level of Evidence C).
Stereotactic Radiosurgery
Stereotactic radiosurgery (SRS) has been widely accepted as an effective treatment for patients with small bAVMs, especially for those with deep location or eloquence involved (44, 245, 246). SRS leads to proliferation of endothelial cells, progressive, concentric vessel wall thickening over years, and eventually endoluminal occlusion of the bAVM nidus (247). Obliteration of the bAVM is the goal for SRS. The disadvantage of SRS compared with microsurgery or embolization was the latency interval between treatment and obliteration, which differs from 6 months to several years (135, 248). Patients were remained at the risk of hemorrhage and delayed presentation of procedural complications during the latency interval. The annual risk of post-SRS hemorrhage was reported as 1.1% (249). Actuarial obliteration rates after SRS were related to multiple independent variables, and generally ranged from 27 to 62% within 3–10 years of treatment according to a multicenter retrospective cohort (248). The Spetzler-Martin grading scale was the most commonly used system for stratifying bAVMs, there were some grading scales used to predict SRS outcomes for bAVMs such as VRAS and RBAS (44, 46).
Small Size
Multiple studies were indicating that SRS appeared to be best suited for small volume bAVMs, which were <10 cm3 in volume or <3 cm in its maximum diameter (44, 231, 250–252). However, most of these studies were retrospective single-centered cohorts. Graffeo et al. (253) systematically reviewed eight studies with 1,102 bAVMs involved, and proposed that SRS appeared to be a safe, effective treatment for bAVMs in Spetzler-Martin grade II and might be considered a front-line treatment, particularly for lesions in deep or eloquent locations. A cohort study on SRS with 363 basal ganglia or thalamic bAVMs suggested its preference for the majority of basal ganglia and thalamic lesions. Another cohort study with 891 bAVMs (eloquence involved in 89.8%) in Spetzler-Martin grade III, suggested that the lesion with small sizes (maximum diameter <3 cm) had the best outcomes after single-staged SRS, even with critical structures involved (229).
Recommendation
SRS can be effective in treating small-sized (≤ 3 cm) bAVMs in deep eloquent areas, including those located in the basal ganglia, thalamus, corpus callosum cerebellum, and brainstem (Class IIa; Level of Evidence B).
Medium and Large Size
The medium and large-sized bAVMs referred to those with a maximum diameter of 3–6 and ≥6 cm, respectively. Those lesions usually belonged to Spetzler-Martin grade III–V.
The traditional paradigm of single-session radiosurgery was not usually used for bAVMs larger than 3 cm in diameter, because of its low total obliteration rate (254). In a retrospective multi-centered study, 233 bAVMs in Spetzler-Martin grade IV (94.4%) and V (5.6%) were treated with single-session SRS (229). A limited role of single-session SRS was suggested in the management of high-grade (IV–V) bAVMs and particularly in the ruptured ones (255). The benefit of SRS for medium-size bAVMs in Spetzler-Martin grade III (i.e., 3 cm < those <6 cm in maximum diameter) was also less evident.
Meanwhile, the bAVMs in Spetzler-Martin grade IV–V were usually with larger volume, more complex angioarchitectures, and frequently located in critical locations. There was no consensus on the optimal management of these high-grade bAVMs, SRS was proposed to be one of the treatments that could be utilized (72, 255). Staged SRS was optional for large bAVMs, but usually utilized in multimodality treatments with mixed results (76). The therapeutic paradigm could be staged by the dosage of radiation and volume of the lesion. A dose-staged SRS treated the entire volume with a repeated low-dosage SRS; and the volume-staged SRS provided a sufficient therapeutic dosage to the targeted volumes as a part of the lesion (256). A systematic review suggested that the volume-staged SRS could achieve a higher obliteration rate and similar complication rate compared with the dose-staged one in the treatment of bAVMs in large volume (>10 cm3) (257).
Recommendation
Single-session SRS is not recommended for patients with large-sized (>3 cm) bAVMs, especially for those which are ruptured (Class III; Level of Evidence B).
Staged SRS might be considered for treating large bAVMs, however, the effectiveness of staged SRS is unclear (Class IIb; Level of Evidence C).
Volume-staged SRS is probably recommended in preference to the dose-staged treatment (Class IIa; Level of Evidence B).
SRS After Endovascular Embolization
It had been reported that the prior embolization of bAVMs would lower the obliteration rates of SRS (231, 249, 255). The prior embolization was proposed to promote the angiogenesis of bAVMs, which might increase the radio-resistance of the lesion and decrease its obliteration rate (258). However, the multimodality treatment of SRS plus prior embolization had been proposed to benefit outcomes for high-grade bAVMs (259). The timing of SRS after embolization had not been determined, which could range from days to years (3 months in median) (225, 260). Referring to the stereotactic radiosurgery guideline for bAVMs, several weeks of latency after the prior embolization was considered beneficial to reduce the risk of post-radiosurgical complications (261).
Recommendation
The SRS subsequently to endovascular embolization can be generally beneficial to high-grade bAVMs, though the obliteration rates are lower in embolized lesions (Class IIa; Level of Evidence C).
Associated Aneurysm
The presence of an untreated bAVM-associated aneurysm was proposed to be a strong predictor for post-SRS hemorrhage (229, 262). AVM-associated aneurysms should be obliterated via microsurgery or endovascular surgery to reduce the hemorrhage risk during the latency interval (248).
Recommendation
It is recommended to treat bAVM-associated aneurysms before SRS to reduce the risk of post-SRS hemorrhage (Class I; Level of Evidence B).
Conservative Treatment
Current conservative treatments cannot promote the obliteration of bAVMs, however, they were preferred reluctantly under a few certain conditions, especially for those located in critical locations. In a prospective study with 48 deep located bAVMs (in basal ganglia, thalamus, insula, etc.) the outcomes of 12-year follow-up indicated that conservative treatments resulted in better prognosis in unruptured bilateral thalamic bAVMs of Spetzler-Martin grade V (86). Another research by Potts et al. (223) also supported the conservative treatment to the unruptured thalamic bAVMs. For asymptomatic large (>6 cm) brain stem AVMs, Thines et al. (264) suggested that the surgical intervention would increase the risk of neurological deterioration by 16-fold at final follow-ups. Spetzler and Martin (51) supposed that the large diffusive bAVMs dispersing through critical areas were inappropriate for microsurgery alone.
Conservative treatments were proposed to be optimal in the management of unruptured eloquent bAVMs. ARUBA trial was the first randomized controlled trial focusing on these issues. Two-hundred and twenty-six adult patients (18 years or older) were recruited during 2007–2013 and randomly allocated to medical management alone (n = 110) or interventional therapy (n = 116) including resection, embolization, SRS, and multimodal approaches (97). The published results of ARUBA suggested that medical management resulted in lower risks of stroke or death in the 33 months of follow-up than interventional management (10.1 vs. 30.7%) in the patients with mRS ≤ 1, or bAVMs lower than Spetzler-Martin Grade IV (62% of ARUBA cases were in grade ≤ II), or bAVM sized <60 mm (62% of lesions in ARUBA sized <30 mm). It was halted because the interim results showed the superiority of the medical management group. It seemed that the conclusion of ARUBA strongly supported the conservative management to eloquent bAVMs, however, it was argued for its limitations. ARUBA trial was criticized for its low enrollment rate, insufficient sample size and follow-ups, high interventional hemorrhage rate, and lack of treatment stratification (265, 266). The study initially estimated that 800 patients would be selected based on statistical analysis. Due to the difficulty of enrollment, 223 cases were included in the analysis which affected the statistical results. The insufficient duration of follow-up might also omit the hemorrhagic risks in long term and over-amplified the short-term complications of the interventions (267). Studies of ARUBA-eligible patients had reported more favorable results and substantially less morbidity, compared to the outcomes of ARUBA (268–270). In the retrospective study of 142 ARUBA-eligible patients treated with embolization, surgery, and/or proton beam radiosurgery during 5 years of follow-up, the risk of stroke, death, and progressive symptoms are less in the intervention group. For those younger patients, conservative management may be inappropriate due to the high accumulative risk of hemorrhage, considering that the annual risk of hemorrhage may be as low as 1% or as high as 33%.
Anti-epilepsy drugs (AEDs) are the essential management for patients with bAVM-related epilepsies. Monotherapy of AEDs was taken by 57% of bAVM patients with epilepsy presentation. However, Josephson et al. (110) reported that AEDs had limited effect on reducing the seizure risk in patients with ruptured temporal bAVMs. If anticonvulsant therapy failed to control seizures, surgical management might be pursued (271).
No specific medicine has been applied for the treatment of bAVMs. Therapeutic strategies like anti-angiogenesis drugs, immunomodulatory drugs, and anti-inflammatory drugs which aim at preventing hemorrhage are still in the experimental stage (272). Headache occurred in ~5–14% of patients with bAVMs, and it could be unilateral or bilateral concurrent with migrainous features. No validated therapy has been applied to release the headache (273, 274).
Recommendations
Conservative managements are reasonable for large-sized (>6 cm) unraptured bAVMs of adult patients concurrent with one of the following conditions: (1) bilateral thalamic bAVMs with deep venous drainage that are deemed inoperable, (2) asymptomatic patients with unruptured bAVMs involving brain stem parenchyma, or (3) diffusive bAVMs dispersed through eloquent areas (Class IIa; Level of Evidence C).
The effectiveness of conservative management on ARUBA-eligible patients is uncertain, due to the limitations and disputed conclusions of the ARUBA trial (Class IIb; Level of Evidence B).
The monotherapy of AEDs can be useful to control the bAVM-related epileptic seizure diagnosed by an electroencephalograph (Class IIa; Level of Evidence C). The usefulness of prophylactic use of AEDs is uncertain (Class IIb; Level of Evidence C).