The Role of Veins in Arteriovenous Malformation and Fistula, Pathophysiology and Treatment
The article explained how the cerebral veins and patterns of venous drainage affects the formation and development of brain arterial-venous malformations (AVMs) and dural arterial-venous fistulas (DAVFs) and also the role it plays in their natural histories. Most clinical relevantly, it proved the involvement and importance of veins by explaining the effectiveness of transvenous approach in the endovascular treatment of brain AVMs and DAVFs in selective cases and introduced some keynotes in the curative embolization of AVMs and DAVFs via venous approach.
KeywordsCerebral veins AVM DAVF Endovascular approach
Brain arteriovenous malformation (AVM) and dural arteriovenous fistula (DAVF) are different kinds of abnormal arteriovenous shunts that share some similarities in pathophysiological process and outcome. An AVM is located within the brain parenchyma, while a DAVF is located within the dura matter. Venous abnormalities are thought to be associated with these lesions. The way of venous drainage may play an important role in their pathophysiology and outcome, as strengthened by the fact that proper embolization through venous approach appeared to be one of the most curative treatments. A thorough understanding of the venous angioarchitectural features and clinical course is of prime importance in the decision making and management of these lesions.
1 Etiology: The Roles of Vein
Etiologies of AVMs and DAVFs are unclear. Whether the AVM is a congenital disease or an acquired disease remains controversial. Recent studies suggest that the initiation and progression of AVMs require interplay among several factors, including (1) homozygous loss of function of causative genes in somatic endothelial cells, (2) angiogenic stimulation (response to injury), (3) participation of bone marrow-derived cells (BMDCs), (4) alteration of monocyte/macrophage function, and (5) hemodynamic changes. Venous sinus occlusion is frequently seen in patients with DAVFs. Some have argued that DAVFs precede the venous sinus occlusion, which occurs secondarily as a result of gradual and progressive hypertrophy of the lining of the sinus due to exposure to high blood flow and/or pressure of the fistula akin to the venous-occlusive phenomenon seen in the cortical veins draining a cerebral parenchymal AVM. On the other hand, some have proposed that the venous sinus occlusion precedes and is directly responsible for the development of the fistula. Venous sinus thrombosis results in the release of angiogenic factors from the organizing thrombus, which subsequently leads to the invasion of small dural arteries and the formation of small dural arteriovenous shunts 
Another theory suggests that DAVFs arise from naturally occurring dormant channels between dural arteries and sinuses, which will open when the sinus is occluded and venous pressure increases . Venous hypertension may lead to tissue hypoxia and increased production of angiogenic factors, which promotes endothelial proliferation and neoangiogenesis [3, 4, 5]
Histologic studies suggest that microscopic thrombosis always presents and plays an important role in the formation of DAVFs through releasing many kinds of growth factors. In immunohistochemical studies, the expression of basic fibroblast growth factor and vascular endothelial growth factor was significantly elevated in the wall of the dural sinuses in patients with DAVFs .
2 Nature History: The Roles of Vein
Although the correlations between the etiologies of AVMs and DAVFs and venous abnormalities have not yet been determined, the characteristics of venous drainage may play important roles in their pathophysiology and outcome.
For the management of unruptured AVMs, the most important thing is to identify the risks of rupture. In addition to the existence of associated proximal intracranial aneurysms, older age, pregnancy and female sex, the characteristics of venous drainage, including the obstructed properties of drainage veins (stenosis or occlusion), the number of draining veins, the location of drainage veins( superficial and deep drainage or mixed type)and the presence of venous aneurysms or varices, are thought to be associated with rupture. The correlations between the number of drainage veins and the risk of nidus rupture are still not clear as different conclusions were drawn from different researches. Deep venous drainage from an AVM was identified as increasing the risk of AVM hemorrhage. AVMs with deep drainage have a higher propensity to bleed compared with that with superficial drainage. However it is difficult to explain the great difference between the deep drainage and the superficial drainage in fluid dynamics. An explanation for this association is that the limitation in venous outflow will result in an increased venous low wall shear stress and eventually lead to increased pressure within the AVMs. Pan et al. identified that angioarchitectural characteristics were associated with the initial hemorrhagic event of supratentorial AVMs. They found that supratentorial AVMs with perforating feeders and deep venous drainage have a higher risk of hemorrhage. However, in contrast with many previous reports, their study showed that the location of AVMs was not associated with hemorrhagic presentation in adjusted analyses. The correlations between deep location and initial hemorrhage in univariate analysis might be caused by the involved perforating feeders and deep draining vein in the deep located AVMs . Lv et al. found that the annual risk of hemorrhage from AVMs was 1.9%. The angioarchitectural characteristics of AVMs associated with hemorrhage including deep and infratentorial AVM location, AVM size <3 cm, single arterial feeder, single draining vein, combined deep and superficial drainage, presence of varices in the venous drainage, and coexisting aneurysms .
Stenosis or occlusion of drainage veins of AVMs can lead to impairment of drainage veins. Previous studies have showed that stenosis of draining vein with over 50% shrinkage in the diameter identified from at least two projections in angiography would have impact on blood flow. Most of the venous stenosis occurs in the junctions where large drainage veins flow into the dural sinus, such as the junction of vein of Galen and straight sinus or the junction of cortical veins and superior sagittal sinus. The etiology of venous stenosis or occlusion is not completely clear. Albert et al. considered it as a anatomic variation rather than an abnormity; but Vinuela et al. thought that it might be resulted from injury of local endothelial cells caused by turbulent blood flow, and the changes of local pressure and shear stress that further lead to thrombosis and luminal stenosis. Quisling and Mickle et al. put forwarded that partial venous stenosis might be vascular contractile responses to high blood flow or venous hypertension. Willinsky et al. proposed that the mechanical kinking of tentorial marginal vein might be the cause that why the AVMs located in the temporal lobe and the posterior fossa AVMs were more prone to bleeding.
So far, there had been no histopathological study unveiling the exact rupture point of brain AVMs, but it is generally considered to be located in the venous part of the nidus. This may be due to the secondary retrograde venous hypertension after venous return impairment because the venous hypertension further causes high pressure in local AVM mass and eventually leads to the rupture of weak venous vessel wall. Miyasaka et al. confirmed the high pressure state of the anterior segment of the drainage venous stenosis using the intraoperative manometry (mean venous pressure of measured AVMs can be as high as 34 mmHg). While from the clinical point of view, the results of related researches concerning the correlation between venous drainage impairment and the history of AVM bleeding were contradictory. Willinsky et al. found that 52% of the bleeding AVMs were in combination with venous stenosis, but Yong et al. failed to confirm the correlations between AVM rupture and venous stenosis.
Although some clues of angioarchitectural features of AVMs and their relationship with clinical presentations has been found, these data was mostly from single center retrospective studies. In the future, more prospective studies focusing on angioarchitectural risk factors for hemorrhage and clinical long-term outcome will be expected.
DAVFs that acquire cortical venous drainage can behave aggressively presenting as intracranial hemorrhage or non-hemorrhagic neurologic deficits such as progressive dementia, seizures, or ataxia due to cerebral edema or ischemia. In a study on the natural history of DAVFs, the annual rupture and mortality rate were 1.6% and 2.3%, respectively . Intracranial hemorrhage is believed to occur from rupture of fragile parenchymal veins exposed to increased pressure from retrograde venous reflux. On the other hand, parenchymal ischemia is thought to occur from venous congestion and hypertension, which impedes exchange of oxygen and removal of metabolic byproducts within the surrounding parenchyma.
3 Venous Drainage and Classification Systems
Noninvasive angiographies, including CTA, CTV, MRI as well as MRA/MRV, are first-line diagnostic tools for DAVFs. Digital subtraction angiography remains the gold standard for detecting and evaluating a suspected DAVF. The hallmark of DAVFs is the presence of early venous drainage due to an abnormal communication within the dura mater between meningeal arterial branches and a venous sinus and/or a subarachnoid vein. A cerebral angiogram can provide information regarding the fistula location, the arterial supply, patency of the venous sinuses, and the pattern of venous drainage.
Cognard classification of intracranial dural arteriovenous fistulas
Anterograde drainage into venous sinus
Drainage into main sinus with reflux into secondary sinus (retrograde sinus drainage)
Drainage into main sinus with reflux into cortical veins
Drainage into main sinus with reflux into secondary sinus and cortical veins
Direct cortical venous drainage without ectasia
Direct cortical venous drainage with venous ectasia (>5 mm and 3 mm larger than diameter of draining vein)
Drainage into the spinal perimedullary veins
Borden classification of intracranial dural arteriovenous fistulas
Anterograde drainage into the dural sinus/meningeal veins
Anterograde drainage into dural sinus and retrograde drainage into cortical veins
Isolated retrograde drainage:
• Drains directly into cortical veins
• Trapped segment of sinus with reflux into cortical veins
• Venous varix/dural lake with reflux into cortical veins
The Spetzler-Martin AVM grading system was designed to assess the risk of neurological deficits after open surgical resection (surgical morbidity), and was based on the characteristics of the AVM itself. It allocates points for features of intracranial AVMs including size, eloquent location, and venous drainage . This system was not intended to characterize risk of hemorrhage.
Spetzler-Martin AVM Grading System
Size of nidus
1 Point = small (<3 cm)
2 Points = medium (3–6 cm)
3 Points = large (>6 cm)
Eloquence of adjacent brain
0 Point = noneloquent (frontal and temporal lobe, cerebellar hemispheres)
1 Point = eloquent (sensorimotor, language, visual cortex, hypothalamus, thalamus, brain stem, cerebellar nuclei, or regions directly adjacent to these structures)
0 Point = superficial only
1 Point = deep
Grade 1: 0%
Grade 2: 5% minor deficit, 0% major deficit
Grade 3: 12% minor deficit, 4% major deficit
Grade 4: 20% minor deficit, 7% major deficit
Grade 5: 19% minor deficit, 12% major deficit
4 Treatment: Transvenous Approach
Owing to better understanding of the pathophysiology of intracranial DAVFs as well as the development in catheter and embolic material technology, endovascular treatment has become the first choice for most intracranial DAVFs. The goals of endovascular treatment of intracranial DAVFs varies depending on clinical presentation, location, and angioarchitecture of the fistula. Careful analysis and fully understanding of the angioarchitecture are critically prerequisites for treatment planning.
Pretreatment angiographic evaluation should confirm the exact location of arteriovenous shunting, arterial supply (dural and pial), flow characteristics, and venous drainage of the DAVF. Particular attention should be paid to the venous drainage of normal brain during the late venous phase of angiography. The venous outflow should be scrutinized for the presence of a parallel venous pouch or compartmentalization of the involved dural sinus, which is often present in a DAVF associated with a major intracranial venous sinus. It is of paramount importance to identify whether the involved sinus is functionally isolated or still serves as a venous drainage conduit for normal brain. Complete and persistent obliteration of the DAVF is achieved by occlusion of its proximal venous drainage. Based on the route of access, endovascular approaches to DAVFs can be schematically divided into transarterial, transvenous, combined.
Transvenous approach remains a standard procedure in the endovascular treatment of DAVFs of the cavernous sinus . Furthermore, certain DAVF associated with a functionally isolated segment of the superior sagittal sinus and transverse/sigmoid junction harboring the fistulous connection may be accessible via a transvenous route. A transvenous approach may also be more suitable if multiple small arterial feeders shunt into a widely dispersed segment of the dural sinus wall . A standard trans-femoral or trans-jugular venous approach is frequently used. The major routes available for the access to the cavernous sinus include anteriorly the superior ophthalmic and facial veins, superiorly the superficial middle cerebral veins and sphenoparietal sinus, posteriorly the petrosal sinuses, and inferiorly the pterygoid plexus. If venous access to the fistulous is difficult, surgical exposure of the superior ophthalmic veins may be entertained. Access to DAVFs of the transverse/sigmoid junction may be established via the ipsilateral, or a patent contralateral, transverse sinus through the torcula. Transvenous embolization has traditionally been performed with detachable and fibered coils that promote thrombosis, and less frequently, with liquid embolic agents.
Nonselective sacrifice of a dural sinus that serves as a functional venous outlet for normal brain tissue may lead to venous ischemia and/or hemorrhage and could trigger the de novo development of DAVFs . Moreover, excessive sinus packing, frequently necessary in the treatment of cavernous DAVFs, may also be responsible for the exacerbation of cranial neuropathies and ophthalmoplegia . In most cases, however, meticulous DAVF angioarchitectural analysis will frequently uncover a parallel venous pouch that acts as the recipient of arterial inflow, and which is angiographically discrete from the involved sinus.
A DAVF associated with an isolated sinus is characterized by a network of small arterial feeders opening into the wall of a completely thrombosed sinus that prohibits antegrade or retrograde venous access. Transvenous access may therefore be difficult to achieve.
Transvenous embolization of cavernous DAVFs is associated with a high occlusion rate and symptom alleviation . In a series of 141 patients, complete DAVF interruption was achieved in 81% of patients, whereas cranial nerve palsies and diplopia improved slowly (65%) or did not change (11%). A potential complication of transvenous cavernous DAVF embolization, especially when attempting access through a thrombosed venous tributary, such as the inferior petrosal sinus, is vessel injury. In a series of 56 patients, 5.4% patients had venous perforation and subarachnoid hemorrhage. The complication was managed with prompt coil occlusion of the rupture site without neurologic sequelae . The reported rate of recanalization of cavernous DAVFs after complete obliteration on immediate angiography is low.
The angioarchitecture, location, and symptomatology of an intracranial AVM are determinant risk factors in the type of treatment. The 3 available treatment options for intracranial AVMs include microsurgery, endovascular embolization and radiosurgery. Drawing lessons from the successful experience of transvenous embolization of DAVFs, some doctors also began to try to treat AVMs through transvenous embolization, and most of the therapeutic effects were satisfactory. Massoud et al. confirmed that controlled hypotension and artery balloon occlusion could promote the glue diffuse from the vein to the artery in vitro and in animal models by Trench technique (the transvenous retrograde nidus sclerotherapy under controlled hypotension, TRENSH). But the greatest challenge of transvenous embolization of the AVM is how to accomplish total occlusion of the AVM mass before effecting the outflow (adequate nidal penetration).
Some lessons in transvenous endovascular therapy have been learned from transvenous embolization of peripheral AVMs. Beek et al  used femoral transvenous approach to embolize AVMs involving the mandible using coils. Similarly, van der Linden et al  described transvenous fashion to treat high flow AVMs in the hand, skull, neck, and pelvis using ethanol. In these procedures, the draining veins were occluded in these procedures with balloons, manual compressions, or tourniquets, and patients experienced a high complication rate. Massoud et al.  were the first to hypothesize a model for transvenous upstream embolization of nidal AVMs. One of the biggest concerns with any transvenous approach is to maintain venous egress until the nidus is eliminated. Based on the TRENSH model, this is ensured by allowing retrograde movement of the embolic agent without any significant reflux. Increased penetration of the nidus allows the closure of arteriovenous shunting. Build-up of arteriovenous shunting pressure can happen if the embolic agent blocks the venous drainage.
The key technique of transvenous embolization is to get close to the malformation mass as fast as possible through the vein end using microcatheter. Any method that can reduce the pressure of draining veins will increase retrograde diffusion from embolic material to nidus. Therefore, in order to judge the possible adverse outcome of venous occlusion, it is necessary to understand how the drainage veins of AVMs work. Venous anatomy can be highly variable and venous catheter navigation may involve extreme angles with tortuosities. Additionally, the cortical venous walls are usually thinner, with concern for a higher rate of perforation. The arterialized veins in AVMs, however, have much thicker walls and may be safer to catheterize [23, 24]. Improvements in microcatheter and wire technology can allow catheterization of cortical veins and the deep venous system. Still, it is important to define the venous anatomy firstly and to understand it in dual planes. Having an arterial pathway which includes an arterial guiding catheter and a micro-catheter for angiographic injection is extremely critical before any venous anatomy is navigated. A transjugular venous access and a triaxial catheter access system can support safe catheterization of the venous system. Using newer microcatheters with detachable tips can decrease the risk of veins tearing while retrieving the microcatheter after embolization [25, 26].
The embolic agents used for intracranial endovascular embolization include polyvinyl alcohol particles, ethanol, coils, silastic spheres, and silk sutures. In 2005, Onyx, an ethylene vinyl alcohol copolymer, was approved for intracranial AVMs treatment. It is a less adhesive embolic agent that laminates along the vessel wall, polymerizes by desiccation, and can be injected over minutes to over an hour to achieve more controlled embolization. The advent of Onyx has allowed a more controlled attempt at transvenous embolization in combination with the modulation of local arterial pressures. Any refluxed Onyx laminates along the vessel wall without completely blocking outflow. Onyx can be injected over time depending on changing AVM hemodynamics during embolization. As non-adhesive liquid embolic technology improves, the safety of using transvenous technique for AVMs embolization will be further established.
Serious consequences due to venous occlusion prematurely can be avoided if there is a plurality of drainage veins, but sufficient dispersion within the malformation mass becomes more difficult. Further work is needed on the exact parameters required to allow successful retrograde nidal penetration based on the size of the nidus and the number of draining veins. We believe that the transvenous approach to AVMs is a novel and an elegant approach to complex disease process. Even though it negates the principle of arterial devascularization known to AVM treatment, it aims to achieve the same end point. A successful transvenous embolization still aims at nidal obliteration before venous outflow obstruction. There are few research reports about the curative effect of transvenous embolization of AVMs ,and transvenous embolization is only applicable for specific cases, such as small AVMs (<2 cm), AVM presenting with known prior hemorrhage, surgery is not suitable or radiotherapy failure, lack of good artery route (such as passing blood-supply, lenticulostriate artery or choroid plexus blood-supply), ideally single venous drainage, only use Onyx or other non-adhesive embolic material, the use of controlled hypotension and transarterial balloon occlusion technique. But we believe that transvenous embolization of AVMs is an entirely novel therapeutic approach, and the goal of nidus obliteration can be accomplished by the method of reverse diffusion. Now the indication of the technique is relatively limited, but the application may gradually increase alone with the accumulation of experience.
Transvenous embolization of intracranial AVMs is a retrograde technique aiming at nidus obliteration. Newer embolic materials such as Onyx have allowed transvenous embolization to be used clinically in limited cases, increasing the clinical credibility of the TRENSH model. Indications for embolization via transvenous approach are limited, which may change as experience accumulates. Nevertheless, this technique will contribute to our understanding of AVM physiology and may have an impact on future endovascular therapeutic options for AVMs.