Embolization-induced angiogenesis in cerebral arteriovenous malformations

Journal of Clinical Neuroscience 21 (2014) 1866–1871

Contents lists available at ScienceDirect

Journal of Clinical Neuroscience

journal homepage: www.elsevier .com/ locate/ jocn

Review

Embolization-induced angiogenesis in cerebral arteriovenousmalformations

http://dx.doi.org/10.1016/j.jocn.2014.04.0100967-5868/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 434 924 2203; fax: +1 434 982 5753.E-mail address: tjb4p@hscmail.mcc.virginia.edu (T.J. Buell).

Thomas J. Buell ⇑, Dale Ding, Robert M. Starke, R. Webster Crowley, Kenneth C. LiuUniversity of Virginia, Department of Neurosurgery, P.O. Box 800212, Charlottesville, VA 22908, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 20 December 2013Accepted 5 April 2014

Keywords:AngiogenesisArteriovenous malformationEmbolizationHypoxiaInflammationStroke

Endovascular occlusion of cerebral arteriovenous malformations (AVM) is often utilized as adjunctivetherapy in combination with radiosurgery or microsurgery. Evidence supports that partial occlusion ofAVM via endovascular embolization leads to increased angiogenesis. This phenomenon may be acontributing factor to the decreased efficacy of AVM radiosurgery following embolization. We reviewthe literature for potential mechanisms of embolization-induced angiogenesis. A comprehensive litera-ture search was performed using PubMed to identify studies that sought to elucidate the pathophysiologybehind embolization-induced angiogenesis. The terms ‘‘arteriovenous malformation’’, ‘‘embolization’’,and ‘‘angiogenesis’’ were used to search for relevant publications individually and together. Three distinctmechanisms for embolization-induced angiogenesis were described in the literature: (1) hypoxia-mediated angiogenesis, (2) inflammatory-mediated angiogenesis, and (3) hemodynamic-mediatedangiogenesis. Embolization-induced angiogenesis of cerebral AVM likely results from a combination ofthe three aforementioned mechanisms. However, future research is necessary to determine the relativecontribution of each individual mechanism to overall post-embolization AVM neovascularization.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Cerebral arteriovenous malformations (AVM) represent a rarepathology of the intracranial vasculature in which arteries connectdirectly to veins without an intervening capillary bed, thereby pre-cluding the development of normal brain parenchyma [1]. Thediagnosis of an AVM is confirmed by angiography, which demon-strates early venous drainage [2]. The most frequent clinical man-ifestations are due to rupture of the AVM nidus whereas lesscommon symptoms are related to arterial steal or venous hyper-tension [3–4]. In order to reduce the morbidity and mortality asso-ciated with AVM hemorrhage, obliteration of the nidus may beachieved with microsurgical resection, radiosurgery, or endovascu-lar embolization alone or in combination [5–6]. While completeAVM obliteration may be achieved in a small proportion of care-fully selected cases with embolization alone, embolization takeson an adjuvant role in the vast majority of AVM treatment plans[7–12]. The goal of embolization varies according to the subse-quent therapy but high flow feeding arteries harboring perinidal

or intranidal aneurysms are frequently targeted. Endovascularocclusion of deep feeding arteries, which are difficult to accessuntil the later stages of AVM resection, may specifically benefitsurgical cases [13]. For radiosurgery, one of the primary goals ofprior embolization is volume reduction of large AVM, which cannotbe safely targeted without incurring a high risk of radiation-induced adverse effects [14]. However, embolization decreasesthe radiosurgical obliteration rates [15–23].

The explanation as to why previously embolized AVM havelower obliteration rates following radiosurgery includes: (1) embo-lization converts a fairly uniform geometric target into a morepoorly defined target with irregular components, resulting ingreater error when designing the radiosurgical treatment plan;(2) embolized AVM partially re-canalize; and (3) ethylene vinylalcohol copolymer (Onyx, ev3 Endovascular, Irvine, CA, USA) mayabsorb or scatter the radiation beams, although a recent in vitrostudy did not find any significant radiation attenuation [21,24].Akakin et al. hypothesized that embolization-induced angiogenesis(EIA) contributed to the lower obliteration rates of previouslyembolized AVM treated with radiosurgery [25]. Using a rat corneamodel, embolized AVM tissue was found to exhibit higherangiogenic activity than untreated or radiosurgically treated AVMtissue. Another study by Sandalcioglu et al. found significantlyhigher endothelial cell proliferation, measured by elevated Ki-67

T.J. Buell et al. / Journal of Clinical Neuroscience 21 (2014) 1866–1871 1867

expression, in embolized AVM compared to non-embolized AVM[26]. In this review, we aim to define the mechanisms of EIA inAVM in order to determine potential targets of future AVMtherapies.

2. Methods

Utilizing PubMed, we performed a comprehensive literaturesearch for all research studies attempting to elucidate the mecha-nisms of EIA in cerebral AVM. The terms ‘‘arteriovenous malforma-tion’’, ‘‘embolization’’, and ‘‘angiogenesis’’ were used to search forrelevant publications individually and together. The referencesfrom additional relevant studies were reviewed and providedadditional clinical and basic science experimental data. Overall,10 studies were reviewed (Table 1). We found three distinctmechanisms that may explain the pathophysiology behind EIA.

3. Results and discussion

The following sections review three potential mechanisms forEIA: (1) hypoxia-mediated angiogenesis, (2) inflammatory-mediated angiogenesis, and (3) hemodynamic-mediated angiogen-esis (Fig. 1).

3.1. Hypoxia-mediated angiogenesis after embolization

Hypoxia-induced expression of pro-angiogenic factors, notablyvascular endothelial growth factor (VEGF), is a potential mecha-nism for post-embolization angiogenesis (Fig. 2). Pro-angiogenicfactors such as VEGF and matrix metalloproteinases are highlyexpressed in cerebral AVM and may determine their clinical course[27]. Incomplete embolization of an AVM results in local regionalhypoxia within the obliterated portion of the nidus [28]. It hasbeen shown that this hypoxic stress can initiate a cascade of eventsleading to increased VEGF expression. Vascular tissue has the abil-ity to adjust to the changing metabolic demands in its microenvi-ronment. In order to match supply to demand under times ofhypoxic stress, the tissue may elicit compensatory angiogenesis.The transcription rate of VEGF mRNA is increased within hoursafter hypoxic stress. Furthermore, there is an increase in VEGFmRNA stability resulting in a four-fold increase in its half-life[29]. This increased VEGF then can bind to receptor Flk-1 leadingto angiogenesis and vascular proliferation until the metabolicdemands of the tissue are satisfied [30–31].

Several studies support the hypothesis that EIA is mediatedthrough hypoxia-induced upregulation of VEGF. Sure et al. studiedthe impact of preoperative embolization on the endothelial expres-sion of the ligand VEGF and its receptor, Flk-1. Using immunohis-tochemical analysis, the group found that VEGF expression wasincreased in embolized AVM compared to non-embolized AVM.VEGF was expressed in 72% of AVM embolized before surgery. Incontrast, only 28% of patients with AVM that had not been previ-

Table 1Mechanisms of embolization-induced angiogenesis

Hypoxia-mediated Sure et al. 2001 [28]Sure et al. 2001 [32]Sure et al. 2004 [33]Kim et al. 2008 [34]

Inflammatory-mediated Chen et al. 2006 [39]Yao et al. 2007 [40]Starke et al. 2010 [41]

Hemodynamic- mediated Malek et al. 2000 [46]Abumiya et al. 2002 [45]Ali et al. 2013 [48]

ously embolized exhibited VEGF expression. Interestingly, Flk-1expression was detected in 80% of AVM regardless of whetherembolization was performed [28].

In another immunohistochemical analysis of the angiogenicactivity of embolized versus non-embolized AVM, Sure et al. foundsupport for the same hypothesis: local regional hypoxia withinpartially obliterated (embolized) AVM induces VEGF expressionand subsequent neovascularization [32]. In their analysis of 30 sur-gical patients with symptomatic AVM, VEGF expression was signif-icantly higher in embolized AVM compared to non-embolizedcounterparts. Specifically, 77% (17/22) of previously embolizedAVM exhibited VEGF expression compared to only 25% (2/8) ofnon-embolized AVM (p = 0.0086). Flk-1 was expressed in 86%(19/22) of patients with embolized AVM compared to 75% (6/8)of patients without prior embolization. Of note, all patients whohad been embolized prior to surgery were operated on within1 week [32].

Sure et al. examined a larger cohort of patients (n = 56) andexpanded the study of angiogenesis-related proteins to includehypoxia-inducible factor-1a (HIF-1a) in addition to VEGF and Flk-1. Immunohistochemical analysis of resected AVM tissue was con-sistent with previous findings: AVM treated with embolizationbefore microsurgical resection exhibited significantly higher localexpression of VEGF and HIF-1a than resected AVM without priorembolization. The differential expression of Flk-1 was similar tothat of VEGF and HIF-1a but insignificant. The inclusion of HIF-1ain the study was important because it supports the hypothesis thatVEGF upregulation is secondary to hypoxia rather than othermechanisms [33].

We have previously found that serum VEGF levels decreasedfollowing embolization and resection of AVM [34]. Others havefound increased local expression of VEGF in AVM following surgi-cal removal [28,32–33]. This sequence of events may be due toincreased expression of VEGF locally within AVM that creates asystemic sink leading to decreased overall levels. This may leadto further local angiogenesis within the AVM.

3.2. Inflammatory-mediated angiogenesis after embolization

The cascade of inflammatory events that occurs within embol-ized AVM tissue may be another mechanism for EIA (Fig. 3). Thereis supporting evidence that inflammation is a major contributingcause for AVM pathogenesis and progression [35]. After emboliza-tion, a variety of inflammatory changes take place within AVM tis-sue [36–37]. In one case series with 23 patients who had AVMembolized with Onyx, histopathological examination revealedmild acute inflammation in specimens resected 1 day after embo-lization, chronic inflammation in specimens resected >4 days afterembolization, and evidence of angionecrosis in two patients [38].

After embolization induces an inflammatory reaction within theAVM vessel wall, the tissue may develop an intrinsically higherpotential for angiogenesis. Chen et al. concluded that the main

25 patients; 18 with embolization, 7 no embolization30 patients; 22 with embolization, 8 no embolization56 patients; 35 with embolization, 21 no embolization13 patients; 12 with embolization, 1 no embolization

Cell culture and animal studyCell culture15 patients; 15 with embolization

Cell cultureCell cultureCell culture

Fig. 1. Transarterial embolization (internal carotid artery) of a cerebral arteriovenous malformation with Onyx (ev3 Endovascular, Irving, CA, USA). (A) Local hypoxia causesincreased vascular endothelial growth factor (circles) that diffuse outward (arrows) and stimulate vascular sprouting. (B) Acute and chronic inflammatory infiltrates in thevessel wall and lumen after embolization. Interleukin-6 is the main inflammatory cytokine that is also pro-angiogenic. (C) Increased blood flow and shear stress in non-obliterated vessels may cause angiogenesis by increasing vascular endothelial growth factor and fetal liver kinase-1 expression. (This figure is available in colour at http://www.sciencedirect.com.)

1868 T.J. Buell et al. / Journal of Clinical Neuroscience 21 (2014) 1866–1871

cytokine bridging inflammatory processes to downstream angio-genesis is interleukin-6 (IL-6) [39]. In cell culture and animal stud-ies, the authors found that locally increased intranidal IL-6 mRNAlevels were strongly correlated with increased mRNA levels ofmatrix metalloproteinase-3 (MMP-3), MMP-9, and MMP-12 –enzymes involved in proteolysis and important for cerebral vascu-lar generation. The correlation between increased expression ofIL-6 and MMP supports the role of IL-6 as an upstream angiogenicmolecule [39]. In another study involving cerebral AVM tissue andIL-6, Yao et al. also concluded that IL-6 plays a pivotal role in theangiogenic cascade [40]. We have also found that systemic MMP-9 expression is increased after both embolization and resectionof AVM [41]. Together, these studies support the hypothesis thatpost-embolization inflammation within the vessel walls of anembolized AVM nidus can stimulate a pro-angiogenic cascade viaIL-6-mediated signaling pathways.

3.3. Hemodynamic-mediated angiogenesis

An alteration in the intraluminal hemodynamic shear forceswithin AVM feeding or draining vessels may contribute to thepathophysiology of EIA (Fig. 4) [33]. Prior studies have demon-strated that fluid shear forces in arteries and veins affect vascularsprouting and proliferation [42–43]. Hemodynamic forces are

known to play an integral role in regulating blood vessel growthand structure. This is manifested by the increased size of high-flowarterial AVM feeders and conversely by the regression of arteries inlow-flow states [44].

Intraluminal shear stress has been shown to regulate the geneexpression of angiogenic proteins in endothelial cells [45]. Maleket al. demonstrated that fluid shear stress rapidly increased levelsof VEGF mRNA expression in exposed brain microvasculature [46].In another study, Abumiya et al. found increased Flk-1 mRNAexpression in response to fluid shear stress in human umbilicalvein endothelial cells [45]. Therefore when blood flow increasesin non-occluded portions of an AVM after partial embolization,increasing shear stress may contribute to neovascularization byupregulating endothelial cell VEGF or Flk-1 expression [45]. Wehave also found that shear stress may increase MMP-9 as well asother pro-inflammatory mediators that may further stimulateangiogenesis [47–48].

3.4. Endoglin

Endoglin (CD105), a transforming growth factor (TGF)-B1receptor-associated glycoprotein expressed on endothelial cells,has been studied in mouse models of cerebrovascular dysplasia

Fig. 2. Hypoxia-mediated angiogenesis. Incomplete embolization of an arteriove-nous malformation (AVM) results in local regional hypoxia within the obliteratedportion of the nidus. This causes increased expression of hypoxia-inducible factor-1a (HIF-1a), vascular endothelial growth factor (VEGF), and fetal liver kinase-1(FLK-1) and stimulates angiogenesis.

Fig. 3. Inflammatory-mediated angiogenesis. Histopathological examinationrevealed acute inflammation in arteriovenous malformations (AVM) resected1 day after embolization, chronic inflammation >4 days after embolization, andevidence of angionecrosis in some patients. This inflammatory reaction within theAVM vessel walls may increase intranidal interleukin-6, upregulate matrixmetalloproteinase expression, and stimulate angiogenesis.

Fig. 4. Hemodynamic-mediated angiogenesis. Increased blood flow and shearstress in the non-occluded portions of an arteriovenous malformation (AVM) maycontribute to neovascularization by upregulating endothelial cell vascular endo-thelial growth factor (VEGF) and fetal liver kinase-1 (FLK-1) expression. Shear stressmay also increase matrix metalloproteinase-9 as well as other pro-inflammatorymediators which may further stimulate angiogenesis.

T.J. Buell et al. / Journal of Clinical Neuroscience 21 (2014) 1866–1871 1869

[49–51]. A soluble form of endoglin (sEng) has been implicated inthe pathogenesis and progression of cerebral AVM. Chen et al.found that human cerebral AVM specimens have elevated meansEng levels [50]. The authors also showed that sEng may induce

pathological vascular remodeling through increased MMP activity.However, the study found no significant difference in sEng AVMtissue levels when comparing embolized and non-embolized AVM.

3.5. Contribution of EIA to AVM radiosurgery outcomes

While the effect of pre-radiosurgical AVM embolization onobliteration rates remains controversial, mounting evidence sug-gests that embolization has a deleterious effect on AVM radiosur-gery outcomes [52–54]. However, it is unknown if the embolicagent itself alters radiosurgery outcomes. Onyx has largely sup-planted other permanent embolic agents, such as N-butyl cyanoac-rylate (NBCA), for endovascular AVM occlusion [9]. Whendecreased radiosurgical obliteration rates of Onyx-embolizedAVM were first noted, it was initially hypothesized that Onyxmay scatter or absorb radiation beams, thereby attenuating thedose delivered to the patent nidus [21]. While a recent study byBing et al. did not report any dose attenuating effects of Onyx, itsconclusions were based on an in vitro AVM model [24]. To ourknowledge, there is no in vivo evidence to support or refute theeffect of specific embolic agents, such as Onyx or NBCA on AVMradiosurgery outcomes.

Embolization results in a multitude of biological and hemody-namic alterations to an AVM nidus. The contribution of EIA todecreased radiosurgical obliteration rates has not been rigorouslystudied. Therefore, it is unclear if the phenomenon of EIA resultsin a clinically or radiographically significant degree of angiogene-sis. Although the mechanisms underlying the negative effect ofembolization on AVM radiosurgical obliteration rates are not com-pletely understood, it is unlikely that induction of angiogenesisalone is adequate to account for the lower obliteration rates ofradiosurgically treated embolized AVM. However, even a fractionalimprovement in the radiosurgical obliteration rates of embolizedAVM may contribute to improved patient outcomes. Therefore,further investigation into the effect of EIA on AVM radiosurgeryoutcomes appears warranted.

1870 T.J. Buell et al. / Journal of Clinical Neuroscience 21 (2014) 1866–1871

3.6. Future directions

Future research is still needed to elucidate how each mecha-nism contributes to the overall result of post-embolization angio-genesis. However, since each of these mechanisms can result inincreased VEGF expression, an interesting study may combineanti-VEGF therapies with endovascular embolization. In animalstudies, Walker et al. demonstrated that anti-VEGF monoclonalantibody bevacizumab (Avastin, Genentech, South San Francisco,CA, USA) attenuates an AVM dysplastic vasculature. The future ofendovascular embolization for cerebral AVM may incorporate suchanti-VEGF treatment [55]. Avastin-eluting stents have already beentried in cardiology for coronary artery disease with reported suc-cess [56]. Such combination treatment should continue to be stud-ied in intracranial pathology as well. For example, a pilot studycould be carried out in humans with a VEGF inhibitor both givenas an oral medical therapy and/or intra-arterially during pre-radi-osurgery embolization. In addition to reducing EIA, anti-VEGF ther-apy may even reduce AVM post-radiosurgery complications suchas the rare but serious chronic, encapsulated intracerebralhematoma [57].

In addition to anti-VEGF therapy for EIA, another potentialtreatment modality would focus on MMP. MMP are involved inthe degradation of the vascular matrix and are a critical step incerebral angiogenesis and remodeling. Hashimoto et al. foundincreased local MMP-9 expression in AVM tissue (endothelialcell/peri-endothelial cell layer and infiltrating neutrophils) aftermicrosurgical resection regardless of whether the patients werepretreated with embolization [58]. Tetracycline derivatives suchas doxycycline may have therapeutic potential to act as anti-angio-genic agents by reducing MMP-9 activity [59–61]. An interestingstudy would use the rat cornea model from Akakin et al. and havea fourth experimental group: embolization plus doxycycline(ex vivo or oral pretreatment) [25]. Since doxycycline suppressescerebral MMP-9 and angiogenesis in animal models, one could rea-sonably predict similar results in human embolized AVM tissue[61].

3.7. Study limitations

This review is limited by the relatively sparse amount of avail-able literature regarding EIA. The combined data supporting thephenomenon of EIA in AVM is not substantial, and this should betaken into account when considering the hypotheses we haveput forth. Additionally, the three mechanisms we delineated,namely EIA secondary to hypoxia, inflammation, and hemody-namic stress, are likely interrelated to an extent. Whether inflam-mation is a byproduct of hypoxia, or vice versa, has not been wellstudied. Indeed, vascular inflammation induced by embolizationis likely associated with, or may underlie, hypoxia and hemody-namic-mediated angiogenic mechanisms. While we defined threedistinct mechanisms of EIA, there is likely overlap among them.However, due to the limited evidence supporting any one of themechanisms, the degree and significance of the lap among themechanisms is currently unknown.

4. Conclusions

Mounting evidence provides support that EIA could be contrib-uting to the lower obliteration rates observed in AVM that receiveembolization prior to radiosurgery. Current mechanistic explana-tions for EIA in cerebral AVM include: (1) hypoxia-mediated angi-ogenesis, (2) inflammatory-mediated angiogenesis, and (3)hemodynamic-mediated angiogenesis. Most likely, each mecha-nism contributes to EIA of cerebral AVM, but the patient and

AVM characteristics which predispose specific lesions to one EIApathway versus another are incompletely understood. Futureresearch is still needed to elucidate how each mechanism contrib-utes to the overall result of neovascularization. Targeting AVMwith anti-VEGF or MMP-reducing treatments delivered duringpre-radiosurgery embolization may be a promising avenue.

Conflicts of Interest/Disclosures

The authors declare that they have no financial or other con-flicts of interest in relation to this research and its publication.

References

[1] McCormick WF. The pathology of vascular (‘‘arteriovenous’’) malformations. JNeurosurg 1966;24:807–16.

[2] Novakovic RL, Lazzaro MA, Castonguay AC, et al. The diagnosis andmanagement of brain arteriovenous malformations. Neurol Clin2013;31:749–63.

[3] Arteriovenous malformations of the brain in adults. N Engl J Med1999;340:1812–8.

[4] Ding D, Liu KC. Orbital venous congestion: rare manifestation of an intracranialarteriovenous malformation. J Clin Neurosci 2014;21:522–4.

[5] Ding D, Yen CP, Xu Z, et al. Radiosurgery for patients with unrupturedintracranial arteriovenous malformations. J Neurosurg 2013;118:958–66.

[6] de Oliveira E, Tedeschi H, Raso J. Comprehensive management of arteriovenousmalformations. Neurol Res 1998;20:673–83.

[7] Spetzler RF, Martin NA, Carter LP, et al. Surgical management of large AVM’s bystaged embolization and operative excision. J Neurosurg 1987;67:17–28.

[8] Gobin YP, Laurent A, Merienne L, et al. Treatment of brain arteriovenousmalformations by embolization and radiosurgery. J Neurosurg 1996;85:19–28.

[9] Saatci I, Geyik S, Yavuz K, et al. Endovascular treatment of brain arteriovenousmalformations with prolonged intranidal Onyx injection technique: long-termresults in 350 consecutive patients with completed endovascular treatmentcourse. J Neurosurg 2011;115:78–88.

[10] Katsaridis V, Papagiannaki C, Aimar E. Curative embolization of cerebralarteriovenous malformations (AVMs) with Onyx in 101 patients.Neuroradiology 2008;50:589–97.

[11] van Rooij WJ, Sluzewski M, Beute GN. Brain AVM embolization with Onyx.AJNR Am J Neuroradiol 2007;28:172–7 [discussion 178].

[12] Panagiotopoulos V, Gizewski E, Asgari S, et al. Embolization of intracranialarteriovenous malformations with ethylene-vinyl alcohol copolymer (Onyx).AJNR Am J Neuroradiol 2009;30:99–106.

[13] Brown Jr RD, Flemming KD, Meyer FB, et al. Natural history, evaluation, andmanagement of intracranial vascular malformations. Mayo Clin Proc2005;80:269–81.

[14] Flickinger JC, Kondziolka D, Lunsford LD, et al. Development of a model topredict permanent symptomatic postradiosurgery injury for arteriovenousmalformation patients. Arteriovenous Malformation Radiosurgery StudyGroup. Int J Radiat Oncol Biol Phys 2000;46:1143–8.

[15] Ding D, Yen CP, Starke RM, et al. Radiosurgery for ruptured intracranialarteriovenous malformations. J Neurosurg 2014 [Epub ahead of print].

[16] Ding D, Yen CP, Xu Z, et al. Radiosurgery for low-grade intracranialarteriovenous malformations. J Neurosurg 2014 [Epub ahead of print].

[17] Ding D, Yen CP, Starke RM, et al. Radiosurgery for Spetzler-Martin Grade IIIarteriovenous malformations. J Neurosurg 2014;120:959–69.

[18] Ding D, Yen CP, Starke RM, et al. Outcomes following single-sessionradiosurgery for high-grade intracranial arteriovenous malformations. Br JNeurosurg 2013 [Epub ahead of print].

[19] Ding D, Starke RM, Yen CP, et al. Radiosurgery for cerebellar arteriovenousmalformations: does infratentorial location affect outcome? World Neurosurg2014 [Epub ahead of print].

[20] Ding D, Yen CP, Xu Z, et al. Radiosurgery for primary motor and sensory cortexarteriovenous malformations: outcomes and the effect of eloquent location.Neurosurgery 2013;73:816–24 [discussion 824].

[21] Andrade-Souza YM, Ramani M, Scora D, et al. Embolization beforeradiosurgery reduces the obliteration rate of arteriovenous malformations.Neurosurgery 2007;60:443–51 [discussion 451–2].

[22] Kano H, Kondziolka D, Flickinger JC, et al. Stereotactic radiosurgery forarteriovenous malformations after embolization: a case-control study. JNeurosurg 2012;117:265–75.

[23] Starke RM, Yen CP, Ding D, et al. A practical grading scale for predictingoutcome after radiosurgery for arteriovenous malformations: analysis of 1012treated patients. J Neurosurg 2013;119:981–7.

[24] Bing F, Doucet R, Lacroix F, et al. Liquid embolization material reduces thedelivered radiation dose: clinical myth or reality? AJNR Am J Neuroradiol2012;33:320–2.

[25] Akakin A, Ozkan A, Akgun E, et al. Endovascular treatment increases butgamma knife radiosurgery decreases angiogenic activity of arteriovenousmalformations: an in vivo experimental study using a rat cornea model.Neurosurgery 2010;66:121–9 [discussion 129–30].

T.J. Buell et al. / Journal of Clinical Neuroscience 21 (2014) 1866–1871 1871

[26] Sandalcioglu IE, Asgari S, Wende D, et al. Proliferation activity is significantlyelevated in partially embolized cerebral arteriovenous malformations.Cerebrovasc Dis 2010;30:396–401.

[27] Hashimoto T, Wu Y, Lawton MT, et al. Coexpression of angiogenic factors inbrain arteriovenous malformations. Neurosurgery 2005;56:1058–65 [discussion1058–65].

[28] Sure U, Butz N, Schlegel J, et al. Endothelial proliferation, neoangiogenesis, andpotential de novo generation of cerebrovascular malformations. J Neurosurg2001;94:972–7.

[29] Stein I, Neeman M, Shweiki D, et al. Stabilization of vascular endothelialgrowth factor mRNA by hypoxia and hypoglycemia and coregulation withother ischemia-induced genes. Mol Cell Biol 1995;15:5363–8.

[30] Millauer B, Wizigmann-Voos S, Schnürch H, et al. High affinity VEGF bindingand developmental expression suggest Flk-1 as a major regulator ofvasculogenesis and angiogenesis. Cell 1993;72:835–46.

[31] Hashimoto T, Emala CW, Joshi S, et al. Abnormal pattern of Tie-2 and vascularendothelial growth factor receptor expression in human cerebral arteriovenousmalformations. Neurosurgery 2000;47:910–8 [discussion 918–9].

[32] Sure U, Butz N, Siegel AM, et al. Treatment-induced neoangiogenesis incerebral arteriovenous malformations. Clin Neurol Neurosurg 2001;103:29–32.

[33] Sure U, Battenberg E, Dempfle A, et al. Hypoxia-inducible factor and vascularendothelial growth factor are expressed more frequently in embolized than innonembolized cerebral arteriovenous malformations. Neurosurgery 2004;55:663–9[discussion 669–70].

[34] Kim GH, Hahn DK, Kellner CP, et al. Plasma levels of vascular endothelialgrowth factor after treatment for cerebral arteriovenous malformations.Stroke 2008;39:2274–9.

[35] Chen Y, Zhu W, Bollen AW, et al. Evidence of inflammatory cell involvement inbrain arteriovenous malformations. Neurosurgery 2008;62:1340–9 [discussion1349–50].

[36] Schmutz F, McAuliffe W, Anderson DM, et al. Embolization of cerebralarteriovenous malformations with silk: histopathologic changes andhemorrhagic complications. AJNR Am J Neuroradiol 1997;18:1233–7.

[37] Germano IM, Davis RL, Wilson CB, et al. Histopathological follow-up study of66 cerebral arteriovenous malformations after therapeutic embolization withpolyvinyl alcohol. J Neurosurg 1992;76:607–14.

[38] Jahan R, Murayama Y, Gobin YP, et al. Embolization of arteriovenousmalformations with Onyx: clinicopathological experience in 23 patients.Neurosurgery 2001;48:984–95 [discussion 995–7].

[39] Chen Y, Pawlikowska L, Yao JS, et al. Interleukin-6 involvement in brainarteriovenous malformations. Ann Neurol. 2006;59:72–80.

[40] Yao JS, Zhai W, Fan Y, et al. Interleukin-6 upregulates expression of KDR andstimulates proliferation of human cerebrovascular smooth muscle cells. JCereb Blood Flow Metab 2007;27:510–20.

[41] Starke RM, Komotar RJ, Hwang BY, et al. Systemic expression of matrixmetalloproteinase-9 in patients with cerebral arteriovenous malformations.Neurosurgery 2010;66:343–8 [discussion 348].

[42] Song JW, Munn LL. Fluid forces control endothelial sprouting. Proc Natl AcadSci USA 2011;108:15342–7.

[43] Asano Y, Ichioka S, Shibata M, et al. Sprouting from arteriovenous shunt vesselswith increased blood flow. Med Biol Eng Comput 2005;43:126–30.

[44] Rossitti S, Svendsen P. Shear stress in cerebral arteries supplying arteriovenousmalformations. Acta Neurochir (Wien) 1995;137:138–45 [discussion 145].

[45] Abumiya T, Sasaguri T, Taba Y, et al. Shear stress induces expression ofvascular endothelial growth factor receptor Flk-1/KDR through the CT-rich Sp1binding site. Arterioscler Thromb Vasc Biol 2002;22:907–13.

[46] Malek A, Lee I, Izumo S, et al. Identification of vascular endothelial growthfactor (VEGF) as a flow-regulated mediator of angiogenesis. Neurosurgery2000;47:499–500.

[47] Starke RM, Ali MS, Jabbour PM, et al. Cigarette smoke modulates vascularsmooth muscle phenotype: implications for carotid and cerebrovasculardisease. PLoS One 2013;8:e71954.

[48] Ali MS, Starke RM, Jabbour PM, et al. TNF-a induces phenotypic modulation incerebral vascular smooth muscle cells: implications for cerebral aneurysmpathology. J Cereb Blood Flow Metab 2013;33:1564–73.

[49] Xu B, Wu YQ, Huey M, et al. Vascular endothelial growth factor inducesabnormal microvasculature in the endoglin heterozygous mouse brain. J CerebBlood Flow Metab 2004;24:237–44.

[50] Chen Y, Hao Q, Kim H, et al. Soluble endoglin modulates aberrant cerebralvascular remodeling. Ann Neurol 2009;66:19–27.

[51] Choi EJ, Walker EJ, Shen F, et al. Minimal homozygous endothelial deletion ofEng with VEGF stimulation is sufficient to cause cerebrovascular dysplasia inthe adult mouse. Cerebrovasc Dis 2012;33:540–7.

[52] Kano H, Kondziolka D, Flickinger JC, et al. Stereotactic radiosurgery afterembolization for arteriovenous malformations. Prog Neurol Surg2013;27:89–96.

[53] Schwyzer L, Yen CP, Evans A, et al. Long-term results of gamma knife surgeryfor partially embolized arteriovenous malformations. Neurosurgery2012;71:1139–47 [discussion 1147–8].

[54] Valle RD, Zenteno M, Jaramillo J, et al. Definition of the key target volume inradiosurgical management of arteriovenous malformations: a new dynamicconcept based on angiographic circulation time. J Neurosurg 2008;109:41–50.

[55] Walker EJ, Su H, Shen F, et al. Bevacizumab attenuates VEGF-inducedangiogenesis and vascular malformations in the adult mouse brain. Stroke2012;43:1925–30.

[56] Stefanadis C, Toutouzas K, Stefanadi E, et al. Avastin-eluting stent: long-termangiographic and clinical follow up. Hellenic J Cardiol 2008;49:188–90.

[57] Nakamizo A, Suzuki SO, Saito N, et al. Clinicopathological study on chronicencapsulated expanding hematoma associated with incompletely obliteratedAVM after stereotactic radiosurgery. Acta Neurochir (Wien) 2011;153:883–93.

[58] Hashimoto T, Wen G, Lawton MT, et al. Abnormal expression of matrixmetalloproteinases and tissue inhibitors of metalloproteinases in brainarteriovenous malformations. Stroke 2003;34:925–31.

[59] Hashimoto T, Matsumoto MM, Li JF, et al. Suppression of MMP-9 bydoxycycline in brain arteriovenous malformations. BMC Neurol 2005;5:1.

[60] Frenzel T, Lee CZ, Kim H, et al. Feasibility of minocycline and doxycycline useas potential vasculostatic therapy for brain vascular malformations: pilotstudy of adverse events and tolerance. Cerebrovasc Dis 2008;25:157–63.

[61] Lee CZ, Xu B, Hashimoto T, et al. Doxycycline suppresses cerebral matrixmetalloproteinase-9 and angiogenesis induced by focal hyperstimulationof vascular endothelial growth factor in a mouse model. Stroke 2004;35:1715–9.