These guidelines were developed by consensus of a multidisciplinary panel of specialists interested in the evaluation and treatment of patients with arteriovenous malformations (AVMs) of the CNS. The reporting criteria described will serve as a template for trial design and for clinical investigators who wish to report on endovascular therapy of cerebral AVMs. Direct comparison of various treatment paradigms is important to standardization of care, maximization of good treatment outcomes, assessment of new methods and technologies.
Statistics from Altmetric.com
The goal of this article is to provide consensus recommendations for reporting standards, terminology and written definitions when reporting on the radiological evaluation and endovascular treatment of cerebral arteriovenous malformations (AVMs). These criteria can be used to design clinical trials, to provide uniformity of definitions for appropriate selection and stratification of patients, and to allow analysis and meta-analysis of reported data. This article must not be construed as defining the standard of care. These reporting standards represent an ideal and are intended primarily for use in research protocols and possibly for quality assessment in individual practice. These definitions represent recommendations for constructing useful research datasets. The primary intent is to facilitate production of scientifically rigorous results capable of reliable comparisons between and among similar studies. In some cases, the definitions contained here are recommended by consensus of a panel of experts for consistency in reporting and publication.
AVMs are an uncommon neurovascular disorder, and methods used for evaluation and treatment are highly variable. There is no consensus on best therapy, while rates of procedural complications vary from 2% to 40%. This has led some investigators to question the validity and efficacy of AVM treatment.1 ,2 The goal in developing these reporting standards is to arrive at a multidisciplinary consensus on how series of endovascular therapy for AVMs are reported. It is our intent that adoption of these standards will facilitate comparison of various series in the literature and assist in evaluation of new technologies or techniques in the endovascular management of AVMs.
Working group composition
This working group is composed of members from the Society of NeuroInterventional Surgery, Joint Section on Cerebrovascular Neurosurgery of the American Association of Neurological Surgeons and Congress of Neurological Surgeons, the Society of Vascular and Interventional Neurology and the American Society of Neuroradiology.
For the purposes of this document, AVM is defined as a high flow, pial, congenital vascular malformation that is based predominantly in neuronal tissue of the CNS. Acquired dural arteriovenous fistulae/malformations, carotid–cavernous fistulas, Vein of Galen malformations, traumatic arteriovenous fistulae or extracranial vascular malformations with intracranial involvement or venous drainage represent a heterogenous grouping of distinctly separate pathological entities with different natural history and treatment considerations and are thus excluded.
A computerized search of the National Library of Medicine database of literature (Pubmed) from January 2005 to December 2009 was performed with the goal of capturing published English and non-English language series of endovascular therapy for cerebral AVMs with at least 100 patients treated. Other historical literature was included by expert consensus.
Patient and lesion characteristics
A variety of patient and lesion specific characteristics may be important in the diagnosis and treatment triage of patients with AVMs. Grading scales have previously been developed for surgical3 and endovascular4 treatment risk assessment. By consensus, we have identified and summarized the demographic characteristics which may prove crucial in a broader evaluation (box 1). We will elaborate on a few specific points.
Patient and lesion characteristics suggested to be reported
Demographics (age, gender)
Presenting symptom (ie, hemorrhage, seizure, headache, focal neurologic deficit, other or incidental)
Any arteriovenous malformation related hemorrhage (ie, after diagnosis but prior to treatment initiation)
Any history of seizure disorder.
Functional status at presentation (typically presented as modified Rankin scale)
Side of brain
Frontal, parietal, temporal, occipital lobe, cerebellum, brainstem/pons, basal ganglia/thalamus
Nidal size, given as mean, and also classified using size subgroups of (<3 cm, 3–6 cm and >6 cm)
Aneurysms (circle of Willis, proximal feeding artery, distal feeding artery, intranidal)
Superficial vs deep vs mixed venous drainage
Venous outlet stenosis or ectasi
Presenting symptoms and prior hemorrhage
Approximately 50% of AVMs present with hemorrhage.5 The remainder present with seizure, focal neurologic deficit, headache or are occasionally identified as incidental findings on non-invasive imaging studies. Prior hemorrhage significantly increases future hemorrhage risk in AVMs.6–8 In addition, the natural history of previously ruptured AVMs may be different from unruptured lesions and, as such, may influence treatment decisions.1 ,2
The patient's clinical status at presentation and prior to treatment initiation should also be recorded, preferably by a neuroscience specialist not directly involved in the care of the patient. The patient's functional status can be reported using the modified Rankin score (mRS) and Barthel Index.9
Initial evaluation most commonly includes cross sectional imaging studies, such as CT or MRI. All patients with AVMs identified or suspected on non-invasive imaging should undergo selective catheter angiography for complete evaluation prior to any form of treatment. The spatial and temporal resolution of catheter angiography still represents the standard of care for complete AVM evaluation and treatment planning. If the size of the AVM is to be measured on cross sectional imaging, it can be done so in three standard planes (cranio-caudal, anterior-posterior and transverse) (see supplementary figure 1, available online only). Any evidence of current or prior hemorrhage on cross sectional imaging should also be noted. In addition, MRI evaluation whenever possible is recommended to better localize the nidus with respect to adjacent brain tissue.
All patients with suspected AVMs should undergo selective catheter angiography by a physician trained in cerebral angiography,10 and ideally the same physician who would be performing the endovascular therapy (see supplementary figure 2, available online only). This would typically entail selective internal carotid and vertebral angiograms, performed ideally in a biplane angiographic suite in order to minimize contrast dose. External carotid angiograms should also be considered, especially on large AVMs that may parasitize the dural supply. The size of the nidus can be measured on the angiogram using either external fiducial markers, as has been described for intracranial aneurysms,11 or using calibrated measurement software in newer angiographic suites, although the accuracy of this technique has been called into question.12 If the nidus has an elongated configuration which would clearly lead to underestimation in size if measured in traditional orthogonal planes, it would be appropriate to measure along the long axis and two planes orthogonal to that measurement (see supplementary figure 2, available online only). For larger AVMs with supply from multiple circulations, it may be easier to measure the overall nidus volume on cross sectional imaging (see supplementary figure 3, available online only).
Contrast injection rates
If using a power injector, typical injection rates for selective carotid angiograms range from 5 to 8 ml/s, with a total volume ranging from 7 to 12 ml.13 Vertebral angiograms typically have injection rates ranging from 3 to 6 ml/s and total volumes from 6 to 9 ml. In cases of some large AVMs with very high flow fistulous components, it may be necessary to increase the injection rate at the discretion of an experienced angiographer.
Image acquisition, archiving and storage
The image acquisition rate should ideally be at least 3 frames/s during the arterial phase, and may need to be higher in order to evaluate fistulous components of the AVM. It is also critical that filming continue through the venous phase for the normal brain circulation. Oblique, magnified or three-dimensional acquisitions should be obtained at the discretion of an experienced angiographer to complete the evaluation. The facility should also have the capability to export subtracted images in a de-identified digital imaging and communications in medicine (DICOM) format to facilitate core laboratory review for future multicenter trials.
Risks and complications
The risk of cerebral angiography at high volume centers is exceedingly low,14–16 and a large meta-analysis showed the rate of permanent disabling stroke to be 0.07% in patients undergoing angiography for evaluation of an intracranial vascular lesion.17 The angioarchitectural characteristics pertinent to the AVM should be documented in the final report.
Lesion anatomy and angioarchitecture
Any series of AVM embolization would be incomplete without a discussion of the pertinent angioarchitectural characteristics, including nidal size, presence of deep venous drainage and involvement of eloquent brain parenchyma, used to arrive at the Spetzler–Martin grade.3 Lesions with deep brain involvement, basal ganglia or thalamic lesions or those with exclusively deep venous drainage appear to have a higher hemorrhage rate.8 ,18–20 Infratentorial lesions may also have higher risk of hemorrhagic presentation.21 In addition, lesions with large nidal size may have a higher prospective hemorrhage rate,19 as may Spetzler–Martin grade IV and grade V AVMs,22 although this has been debated.23 By contrast, AVMs with small nidal size have been shown to have high nidal pressures, which may be related to an increased risk of hemorrhage in small AVMs.24
AVM related aneurysms occur in approximately 15% of patients, and have been shown to independently increase the risk of future hemorrhage.8 ,25–27 These aneurysms occur along the course of arteries that supply AVMs and are typically considered related to high blood flow. In addition, aneurysms may occur in the nidus of AVMs with variable incidence (see supplementary figure 4, available online only). Differences in reported incidence of nidal aneurysms may be related to technical factors in the acquisition of catheter arteriography. High frame rate acquisition or superselective arteriography may be necessary to identify nidal aneurysms in some AVMs (see supplementary figure 5, available online only). Flow related aneurysms can also regress with treatment of the AVM itself.
The venous anatomy of brain AVMs is often implicated in hemorrhage20 and is likely a critical feature that has been under evaluated. In addition to describing the drainage as deep, superficial or mixed, the presence of venous outlet stenoses or ectasia should be recorded as evidence of outflow obstruction. Focal venous dilatations, often termed a venous varix or venous aneurysm, can also be seen, and their presence should prompt investigation into venous outflow obstruction (see supplementary figure 6, available online only). If a venous outlet stenosis or occlusion results in a change from deep to superficial drainage (ie, secondary to Straight sinus occlusion), this would still be termed deep venous drainage.
There are a few specific subtypes of AVM arterial supply that are distinct angiographically which deserve mention. If the predominant supply to an AVM is by small, indirect branches from a pial vessel that continues on to supply normal brain tissue distally, such vessels are termed en passage feeders. This type of supply is often seen in sylvian fissure AVMs (see supplementary figure 7, available online only) and can be more difficult to safely embolize. In addition, medium to large AVMs often recruit collateral supply from a different circulation through pial anastamoses, which has been termed ‘angiomatous change’25 (see supplementary figure 8, available online only). This feature has been suggested to be protective against hemorrhage, as has an arterial ‘borderzone’ location of AVM nidus.28 It is also critical to recognize angiomatous change and not mistake it as a component of the nidus, so as to not overestimate the size of the nidus for surgical or radiosurgical treatment.
It is also helpful to characterize the nidus as compact or diffuse (see supplementary figure 9, available online only). The presence of a diffuse nidus, especially one associated with a deep perforating artery supply, has been associated with worse surgical outcomes.29 Recently, an additional entity has been described, termed ‘proliferative angiopathy’.30 This distinct lesion has multiple enlarged vessels which are primarily on the brain surface rather than intermixed with neuronal tissue, with a markedly disorganized pattern, and overall low shunt volume relative to the size of the lesion. The natural history of this entity has been suggested to be different than that of typical pial AVMs with a distinct nidus, and recognition of this subtype may have implications with respect to adequate patient counseling.
Clinical and hospital support for embolization
AVM embolization should be performed in a center that is experienced in the treatment of these complex lesions. Ideally, all patients are discussed by a multidisciplinary team capable of offering embolization, microsurgical resection and stereotactic radiosurgery. Embolization should be performed at a hospital which has neurosurgical support available, should emergent hematoma evacuation, ventriculostomy placement or AVM resection become necessary during the course of treatment. Post-procedure, patients should be admitted to a dedicated neurological intensive care unit whenever possible.
Procedural details of AVM embolization
Endovascular treatment of cerebral AVMs is commonly referred to as ‘embolization’. The procedure may be intended as primary or definitive treatment. Alternatively, the embolization procedures may be performed as palliative or adjunctive to other procedures, most commonly in advance of surgical resection. Procedural details to be collected are listed in box 2. The rationale for reporting the major details of embolization are highlighted below.
Procedural details and follow-up to be recorded
Indication for embolization (presurgical, preradiosurgical, curative, targeted, palliative)
No of embolization sessions (overall, mean per patient)
Anesthesia (conscious sedation vs monitored anesthesia care vs general anesthesia)
Embolic agent(s) used
Per cent nidal reduction after all embolizations
Post-procedural BP control (mean arterial pressure targets)
Post-procedure steroids (if routine, typical taper)
Stable cure=follow-up angiography at minimum of 6 months showing persistent nidal occlusion
Non-neurologic (but not resulting in a permanent deficit or death)
Clinical outcome measure
Rationale for embolization
AVM embolization is performed for a variety of reasons. Adjunctive embolization as a preoperative technique is often performed to minimize blood loss, to reduce operative time and to facilitate resection of larger lesions.31–33 For patients undergoing radiosurgery, embolization has been used to reduce overall nidal volume or target angioarchitectural components of the lesion that may represent a source of hemorrhage or potential hemorrhage.34 ,35 Curative embolization has historically been achieved in only a minority of AVMs. However, there is renewed interest in using certain embolic agents to attempt to achieve a higher proportion of endovascular cures,36 ,37 although some have questioned the advisability of this approach38 ,39 due to concerns of increasing hemorrhagic risk. The goals of the embolization procedure are dependent on the indication, and as such goals should be specifically and prospectively identified and stated. For example, deep nidal penetration of embolic materials may not be as critical in the preoperative embolization of AVMs as it may be for pre-radiosurgical or curative embolization.
Embolization sessions per patient
Staged nidal volume reduction has often been performed for adjunctive embolization of larger AVMs, owing to a reportedly lower rate of post-procedural hemorrhage. Heidenreich et al showed a higher rate of hemorrhage when greater than 60% of an AVM was embolized in one session.40 The interval between sessions should also be mentioned.
Embolic agents used to treat AVMs have changed over time. Historically, a variety of agents had been used off label, including silk suture and barium pellets. The use of liquid agents began in earnest in the 1980s, and over two decades of experience exists with cyanoacrylate derivatives. Currently, liquid embolic agents are used almost exclusively in the nidal embolization of brain AVMs. The two Food and Drug Administration approved agents in the USA are n-butyl cyanoacrylate (TruFill; Codman Neurovascular, Raynham, Massachusetts, USA) and ethylene vinyl alcohol copolymer (Onyx; Covidien Inc., Mansfield, Massachusetts, USA). These two agents have significantly different handling characteristics. For example, Onyx has been touted as having the potential for greater control of injection and greater nidal penetration.37 ,41–46 Accordingly, the slow injection of Onyx results in longer procedure times, greater fluoroscopy use and patient radiation dose.47
Neurophysiologic monitoring and provocative testing
The use of neurophysiologic monitoring including EEG, somatosensory evoked potentials and motor evoked potentials is sometimes performed during embolization. In the anesthetized patient, this technique is often be combined with provocative testing, where anesthetic agents, including sodium amytal and lidocaine, are injected via a superselectively placed microcatheter to test for transient neurological deficits prior to any permanent embolization and occlusion of a cerebral artery.48–54 Similarly, the same testing technique may be performed in the awake patient with detailed neurological and cognitive testing.
AVM embolization can be performed with either conscious sedation with a nursing assistant, monitored anesthesia care or general anesthesia with the help of an anesthesiologist. Definitionally, detailed neurological testing is not possible under general anesthesia. While complications from general anesthesia are rare, especially in the patient age group that is often treated for AVMs, reporting of which type of anesthesia was used would be helpful. The effect of anesthetic choice on the safety of AVM embolization has not been well studied.
An important part of understanding the role of embolization is the ability to detail and stratify the type and severity of complications that occur. As such, complications should be stratified according to a standard classification system. By consensus, the following method has been defined for the classification of important complications. The cause of these clinical complications may be procedural (ischemic or hemorrhagic) or not directly related to the procedure. Regardless, all complications must be reported to generate a better understanding of the safety of embolization in comparison with natural history data.
Transient neurologic deficit
Transient neurologic deficits may occur following embolization, on the basis of local ischemia or edema adjacent to the nidus after embolization. Moreover, changes in local hemodynamics resultant from embolization can similarly lead to a transient deficit. We propose that any deficits that are completely resolved by 30 days post-embolization be characterized as transient.
Permanent neurologic deficit, non-disabling
Any deficit that persists after 30 days will be considered permanent. An mRS score ≤2 or Barthel Index ≥80 indicates a non-disabling deficit.
Permanent neurologic deficit, disabling
Any deficit that persists after 30 days will be considered permanent. An mRS score ≥3 or Barthel Index <80 indicates a disabling deficit.
Any complication, neurologic or otherwise, that results in death within 30 days of the procedure shall be considered a ‘procedure related’ death.
Any death occurring beyond 30 days of a treatment procedure during the study period should be recorded, with cause of death identified to the best ability, for epidemiological purposes. Comparison of life expectancy without and with treatment remains an important source of disagreement.
Non-neurologic complications occur in a minority of patients treated for AVM embolization. These must be classified according to a consensus method which has been used for other cerebral endovascular procedures11(see supplementary table 1, available online only).
Series describing adjunctive AVM embolization are typically reported as per cent nidal volume reduction. This is often given as an estimate based on the judgment of an experienced neurointerventionalist, which has been shown to have fairly reliable inter- and intraobserver reliability.55 Any residual shunting with visible nidus remaining constitutes an incomplete embolization.
If completion angiograms at the time of embolization show no residual nidal filling, the term ‘complete embolization’ is recommended instead of ‘cure’. The term ‘cure’ should only be used after delayed imaging (at least 6 months after the last procedure), and the term ‘cure’ should not be invoked without a selective catheter angiogram. Recanalization of completely embolized AVMs has occurred, especially following incomplete nidal penetration of liquid embolic agents.56 The catheter arteriogram must not show any residual nidus or arteriovenous shunting in order to be deemed an ‘angiographic cure’ (see supplementary figure 10, available online only). Follow-up imaging with CT or MRI is often performed to assess for ischemic or hemorrhagic complications but remains inadequate to define complete and durable nidal obliteration.
Clinical outcome measures
The patient's clinical and neurologic outcome should be assessed and monitored using the mRS, preferably by an independent neurologist. In patients who have a disabling deficit (mRS >2), ongoing evaluation using functional measures such as mRS and the Barthel Index may be helpful and should be recorded at regular intervals. Because AVM patients have been shown to improve, especially those with postoperative deficits, repeat evaluation at regular intervals is most appropriate, including 30 day, 90 day, 180 day, 1 year and annual intervals thereafter.57 When embolization is being performed as an adjunctive tool, neurological status must be recorded following the embolization procedure alone and after any additional treatment, such as surgery or radiosurgery. This is important in identifying deficits that arise from adjunctive embolization as a component of a broader treatment paradigm.57
In addition to reporting the results for the aggregate population, subgroups analysis is important. The severity of AVM related hemorrhages has been called into question.58 For this reason, it is critical that every effort is made to determine whether or not hemorrhage has occurred prior to first treatment. Classification of AVMs according to standard grading systems is helpful to assess the prognostic significance of these scales, as some have shown higher complication rates in high grade lesions.4 ,59
We have listed recent large AVM embolization series in table 1. The range of permanent disabling complications or death in these large series is quite varied, as are the definitions of complications themselves. We feel this further highlights the need to arrive at a standardized reporting schema.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Download Supplementary Data (PDF) - Manuscript file of format pdf
Additional figures are published online only. To view these files please visit the journal online (http://jnis.bmj.com).
Correction notice This article has been corrected since it was published Online First. The author Joshua Hirsch has been amended to Joshua A Hirsch.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.