Vertebral Artery Ostial Stenosis: Prevalence by Digital Subtraction Angiography, MR Angiography, and CT Angiography

Clinical Investigative Study

Vertebral Artery Ostial Stenosis: Prevalence by DigitalSubtraction Angiography, MR Angiography, and CT Angiography

Sathish Kumar Dundamadappa, MD, Keith Cauley, MD, PhDFrom the Department of Radiology, University of Massachusetts, Worcester, MA.

Keywords: Vertebral artery ostialstenoses, prevalence, MRA, CTA, DSA.

Acceptance: Received August 4, 2011,and in revised form October 11, 2011.Accepted for publication November 6,2011.

Conflict of Interest: None.

Correspondence: Address correspon-dence to Sathish Kumar Dundamadappa,MD, Department of Radiology, Univer-sity of Massachusetts, 55 Lake Av-enue, North Worcester, MA 01605. Email:[email protected].

J Neuroimaging 2012;XX:1–8.DOI: 10.1111/j.1552-6569.2011.00692.x

A B S T R A C T

BACKGROUND AND PURPOSE(1) To determine the prevalence of vertebral arterial ostial stenosis (VOS) as determinedby the “gold standard” of digital subtraction angiography (DSA). (2) To learn the cor-relation between vertebral ostial stenosis and study indication. (3) To determine theability of contrast-enhanced magnetic resonance angiography (CE MRA) and computedtomographic angiography (CTA) to reflect the true prevalence of vertebral ostial stenosisas determined by DSA.METHODSThree hundred and twenty-nine patients who underwent DSA had recorded evaluation of443 vertebral artery origins. Cases were categorized by patient age and study indication.Similar numbers of CTA and MRA studies were assessed.RESULTSThe prevalence of VOS in the study population was 5.4%. VOS was not observed in pa-tients under 40 years of age, and was seen in 12.5% of patients over 70 years. CE MRAdemonstrated decreased signal at the vertebral origins consistent with stenosis in 20%of patients. CTA estimated VOS at .8%, and yielded 7.3% of studies, which were nondiag-nostic for VOS.CONCLUSIONThe prevalence of VOS as determined by DSA is low and increases with patient age andcorrelates with factors such as anterior infarct (18.4%), posterior infarct (33.3%), carotidatherosclerosis (30.8%), and vertebrobasilar insufficiency (33%). Patients being evalu-ated for reasons less closely correlated with atherosclerotic disease, such as arteriove-nous malformation (AVM) or hemorrhage showed a lower prevalence of VA stenosis (brainaneurysm or AVM 5/121, 4.1%, brain hemorrhage 5/153, 3.3%). Routine clinical MRAsignificantly overestimates VOS prevalence, and findings suggest that CTA underestimatesthe degree and prevalence of VOS.

IntroductionVertebral artery stenosis may account for up to 20% of pos-terior circulation strokes.1-4 This fact together with advancesin interventional techniques has led to an increased interestin accurate diagnostic imaging and endovascular treatment ofthese lesions. Relative to carotid artery disease, the role of var-ious noninvasive angiographic imaging techniques is less welldefined for vertebral artery disease. Contrast-enhanced mag-netic resonance angiography (CE MRA) is widely consideredto be a safe, convenient, and noninvasive vascular imagingmodality, offering high sensitivity and specificity for evalua-tion of carotid artery stenosis.5-7 For this reason, CE MRAof the cervical vasculature has become a common imagingstudy, which entails a report on the posterior circulation bythe interpreting radiologist. Others have observed that MRAis not reliable for evaluation of the proximal vertebral arteries,

with tendency to overestimate the degree of stenosis.8,9 Nev-ertheless, the technique is advocated for study of the proximalvertebral arteries,8,10-12 and has even been used to evaluatethe prevalence of proximal vertebral stenosis in the generalpopulation.13

As decreased signal at the vertebral origins is often observedon CE MRA studies of the cervical vasculature, and is often seenin cases where true stenosis is unlikely, we wished to investigatethe prevalence of vertebral artery ostial stenosis (VOS) by the“gold-standard” technique of digital subtraction angiography(DSA). We undertook a retrospective study whereby all DSAcases seen at this institution over a 2-year period were reviewed,and VOS prevalence was calculated. Because DSA is an inva-sive procedure with measurable risk, it will never be possible todesign a prospective DSA study of normal volunteers. Instead,we investigated the accuracy of MRA by looking at a large

Copyright ◦C 2012 by the American Society of Neuroimaging 1

cohort of DSA patients. As an invasive procedure, DSA patientsgenerally represent a “sicker” patient population than MRApatients. To address this issue, DSA patients were grouped ac-cording to study indication, to learn the prevalence of VOS inpatients with high probability of atherosclerotic disease basedon age or study indication (ie, history of stroke), versus thoseof lower probability, reflecting a more general population (ie,history of aneurysm). A similar number of cervical computedtomographic angiography (CTA) studies were reviewed overthe same time period, and the prevalence of VOS was calcu-lated. Only a small number of patients received both DSA andMRA, which were of diagnostic quality for assessment of thevertebral origins by both studies (6 patients) or DSA and CTA(55 vertebral origins), permitting direct comparison in muchsmaller cohorts.

Materials and MethodsPatients

All DSA studies performed at this institution between January 1,2008 and December 12, 2009 were reviewed. To be includedin the study, a DSA must include an interpretable archivedimage of the vertebral artery origin. With an interest in iso-lated ostial disease, extensive vertebral artery disease or casesof extreme hypoplasia or vertebral occlusion were excluded(8 cases). Cases of dissection involving the V1 segment were alsoexcluded (11 cases). These resulted in exclusion of the patientfrom the database (19 patients). Cases of VOS that had beenpreviously treated (stented) were counted as positive (3 cases).Three hundred and twenty nine patients and 443 vertebral ar-teries were included in the study.

DSA patients were categorized, based on previous imag-ing studies and clinical indication for DSA into seven groups:(1) Trauma (n = 12); (2) Aneurysm or arteriovenous malforma-tion (AVM) without hemorrhage (n = 5); (3) Known carotiddisease (n = 13); (4) Intracranial hemorrhage (±aneurysm;n = 153); (5) Anterior circulation infarction (n = 87); (6) Poste-rior circulation infarction (n = 21); and (7) Symptomatic with-out bleed or infarct (n = 25). The seventh group was furthersubdivided into patients at higher risk for posterior circula-tion disease (symptoms of vertebrobasilar insufficiency (n = 2),suspected vasculitis (n = 2), diagnosed cerebral venous throm-bosis (n = 2), suspected basilar artery disease (n = 1)), andthose at lower risk (syncope (n = 1), headache (n = 5), symp-toms of epistaxis (n = 2) preoperative for meningioma resec-tion (n = 1)). DSA cases were also profiled according to patientage.

A comparable number of MRA and CTA studies were re-viewed from the same time period. Two hundred and fortyeight MRA studies were reviewed. Four studies were excludedas uninterpretable due to poor contrast timing (3) or patientmotion (1). Arteries occluded due to dissection or thrombosis(23 arteries) were excluded. Four hundred and sixty-five verte-bral arterial (VA) origins were included in the CE MRA study.Two hundred and forty-nine CTA studies were reviewed. Ar-teries occluded due to dissection or thrombosis (7 arteries) wereexcluded. Four hundred and ninety-four VA origins were in-cluded in the CTA study.

DSA Technique

DSA (Easyvision; Philips; Best, the Netherlands) was performedwith femoral artery catheterization. Angiography of the verte-bral arteries was performed with aortic, right innominate arteryor left subclavian artery injection at the frame rate of 2/secondfor 4 seconds duration. A total of 9 mL of the contrast medium(Omnipaque 300; Nycomed, Oslo, Norway) was injected atan approximate rate of 6 mL/second via 5-F catheter, handinjection.

Retrospective review of all DSA studies over a 2-year periodat this institution was conducted to assess prevalence of isolatedostial stenosis of the vertebral artery. Inclusion criteria were:(1) Aortic, subclavian or innominate contrast injection permit-ting assessment of the vertebral origin. (2) Recorded diagnosticquality DSA image of the vertebral origin. Cases were excludedfrom the study if there was other DSA apparent disease of thevertebral artery, such as stenosis, occlusion, or dissection (ie,ostial stenosis was not isolated). Individual patient data wasrecorded only once in the 2-year period (ie, repeated studieson the same patient were not recorded). Three cases were re-ferred specifically for evaluation of known VOS, as determinedby outside studies. These cases are addressed specifically in thediscussion.

Magnetic Resonance Angiographic Technique

All CE MRA examinations were performed using GE 1.5-T Signa systems (GE Medical Systems, Milwaukee, WI,USA) with a combination of phased-array head and neckcoils and high-performance gradients (maximum gradientstrength 33 mT/m, slew rate 120 mT/m/ms. The CE MRAwas performed with 3-dimensional fast-spoiled gradient echo(FSPGR) acquisition in the coronal plane (TR/TE 5.05/1.27 ms,flip angle 45◦, number of signal averages (NSA) 1, matrix128 × 512, zero filling in the slice direction, bandwidth62.5 KHz, slice thickness 1.0 mm, 80 partitions, field of view30 cm, scan time 1 minute 14 seconds). The nomimal voxelsize was .78 × 1.07 × 1.0 mm with a value of .84 mm.3

Contrast material (Gadopentetate Dimeglumine [Magnevist,Schering, Berlin, Germany] or Gadobenate Dimeglumine [Mul-tihance, Bracco, Milan, Italy]) was infused through a 22-gauge venous angiocatheter in the anticubital fossa by using apower injector (Spectris; Medrad, Pittsburgh, PA, USA). Con-trast at .1 mmol/kg body weight was injected at a rate of2.0 mL/second. Scan delay was individually calculated per-forming a bolus test. Each bolus was immediately followed bya 60-mL saline flush.

The images were postprocessed with a Windows Advan-tage workstation (GE Medical Systems). Magnetic resonanceangiograms were generated in lateral rotations by using a max-imum intensity projection (MIP) algorithm and projections ev-ery 10◦, from –90◦ to +90◦. We also used base partitions toanalyze stenosis. All views were evaluated.

Spiral Computed Tomographic Angiographic Technique

CTA studies were performed on Philips Brilliance 64-sliceCT scanners. X-ray tube voltage and current were 120 kevand 370 mAs. A 80 mL Isovue 370 was administered with a

2 Journal of Neuroimaging

Table 1. Prevalence of VOS Within the General (DSA) Population

All Cases VOS %

Total 443 24 5.4R 188 11 5.8L 255 13 5.1

Avg age 54.6 60<40 51 0 .040-49 108 5 4.650-59 123 6 4.060-69 97 5 5.170-79 52 7 9.6>79 12 1 8.3

DSA-VOS correlation with age P < .0001.

Medrad power injector at 4 mL/second. An automated trackingbolus technique was used, which included a 40 mL saline flushat 4 mL/second. Axial images were obtained at .9 mm with.45 mm overlap. Reconstructions were at 20/3 mm for ax-ial and 10/3 mm for coronal and sagittal maximal intensityprojections.

Image Analysis

Measurements were based on the North American Symp-tomatic Carotid Endarterectomy Trial (NASCET) measure-ment technique, stenosis was measured by comparing the nar-rowest diameter of the residual lumen at the vertebral origin tothe lumen of the immediately adjacent normal-appearing ves-sel. Because of the narrow dimensions of the vertebral artery,stenosis was considered to be “positive” when narrowing wasmeasured at >50%, consistent with several recent studies ofvertebral artery disease.13,14 Two neuroradiologists evaluatedthe degree of vascular stenosis with consensus grading.

Statistical Analysis

The relationship between patient age and vertebral origin dis-ease was analyzed by χ2 test, with significance of P < .05.One goal of this study is to assess vertebral origin disease byeach angiographic method, with sufficient case numbers to ar-rive at statistically meaningful prevalence numbers. Numbersare as reported (Table 5), with prevalence percentage and 95%confidence intervals.

ResultsOur study was motivated by the frequent impression of sig-nificant narrowing of the vertebral origins seen on CE MRA.We sought to determine the “true” prevalence of VOS as de-termined by the “gold standard” of DSA, and to compare thatvalue to the prevalence as determined by CE MRA and CTA.As assessed in “all-comers” to DSA, VOS was seen in 5.4% ofcases, and correlated with patient age (P < .05, Table 1). Toaccount for selection bias, the DSA population was stratifiedinto risk groups based on indication for the study (Table 2).Prevalence among groups ranged from 3.3% in the group be-ing evaluated for intracranial hemorrhage to 54% for the groupbeing evaluated for symptoms of vertebrobasilar insufficiency,and showed clear correlation with known risk factors and study

Table 2. Prevalence of Vertebral Ostial Stenosis as a Function ofIndication for DSA

Patient Groups VOS

Trauma (n = 12) 2 (16.6%)Aneurysm/AVM (no hemorrhage; n = 121) 5 (4.1)Known carotid disease (n = 13) 4 (30.8)Hemorrhage ± aneurysm (n = 153) 5 (3.3)Anterior infarct (n = 87) 16 (18.4)Posterior infarct (n = 21) 7 (33.3)Vertebral, VB disease (n = 36) 12 (33.3)

DSA-VOS correlation with indication P < .0001.

Table 3. Prevalence of VOS as a Function of Patient Age Within theGeneral MRA and CTA Population

All Cases (#vert art) VOS %

MRA 465 93 20.0Avg Age 56.8 64.8<40 77 6 7.840-49 81 8 9.950-59 112 18 16.160-69 79 20 25.370-79 68 26 30.2>79 48 15 31.25

All Cases VOS %

CTA 494 4 .8Avg Age 58.1 64.75<40 66 1 1.540-49 72 050-59 118 060-69 100 2 2.070-79 67 0>79 71 1 1.4

MRA-VOS correlation with age P = .0752 (borderline significant). #vert art =number of vertebral arteries.

Table 4. Prevalence of VOS as a Function of Indication for MRA andCTA

Patient Groups VOS-MRA VOS-CTA

Question stroke n = 198 46 (23.2 %) n = 348 2Known stroke n = 70 14 (20.0%) n = 20Known carotid disease n = 9 2 (22.2%) n = 19 1Hemorrhage n = 7 2 (28.6%) n = 23Vertebral, VB disease n = 12 0 n = 12Other (headache, collagen

vascular disease,multiple sclerosis, etc)

n = 169 29 (17.2%) n = 72 1

MRA-VOS correlation with indication P = .3115 (not significant).

indication. To address the possibility of selection bias for VOSin the MRA and CTA population, these data were reviewedin terms of correlation with patient age and study indication(Tables 2-4).

CE MRA demonstrated reduced signal at the vertebral ori-gins, consistent with VOS, in 20% of cases (Table 5). CTAstudies showed .8% prevalence of VOS in diagnostic studies;7.3% of CTA studies were nondiagnostic for VO disease (36 of494 vertebral arteries), largely secondary to poor visualization

Dundamadappa and Cauley: VOS—Prevalence 3

Table 5. Prevalence of Vertebral Ostial Stenosis as Assessed by DSA,MRA, and CTA, with Confidence Intervals

StenoticVertebral Vertebral Prevalence

Patients Arteries Origins (±95% CI)

DSA 329 443 24 5.4% ± 2.1%MRA 244 465 93 20% ± 3.0%CTA 249 494 4 .8% ± .06%

A B

Fig 1. Fifty seven-year-old woman evaluated for subarachnoid hem-orrhage of unknown etiology. (A) CE MRA of the neck was obtained.The study is obliqued to the right. Note signal void at the left verte-bral artery origin (arrow). DSA study obtained 3 weeks later (B, leftsubclavian injection) shows normal appearance to the left vertebralartery origin.

due to streak artifacts from incoming contrast bolus and patientbody habitus.

Imaging studies were reviewed to identify patients who hadreceived a DSA study and had also received a CTA or MRAwithin the 2-year period of inclusion. Eleven patients obtainedboth DSA and MRA, with study of 14 vertebral origins. Twostudies demonstrated narrowing or signal loss by MRA, withno evidence of narrowing by DSA (see Figs 1 and 2). Nooverlapping study recorded narrowing of the vertebral originby DSA. Forty nine patients were studied by DSA and CTA,with data obtained on 61 vertebral arteries. Only two casesdemonstrated vertebral ostial stensosis on DSA. Neither CTAstudy recorded significant narrowing (see Fig 3). No overlap-ping study recorded narrowing of the vertebral origin on CTA.One patient had all three angiographic studies—DSA, CTA, andMRA of the cervical vessels performed within a 2-day period.These studies showed signal loss at the left vertebral origin on

MRA, which could be interpreted as stenosis; DSA and CTAstudies demonstrated a tortuous but entirely patent proximalleft vertebral artery (Fig 2).

DiscussionIntra-arterial angiography is considered to be the “gold-standard” for detection of atherosclerotic change.15,16 Diag-nostic angiography is an invasive and time-consuming proce-dure with measurable patient risk, however. Developments inmagnetic resonance methods, particularly CE MRA togetherwith the neurovascular head and neck coil have been shownto be of high sensitivity and specificity for study of carotid dis-ease,10,17-19 and is now commonly used in the clinical setting.The ability of CE MRA to accurately evaluate the vertebralarteries has been more controversial.8,13,14,17,19-21 Routine CEMRA of the neck often demonstrates decreased signal at theVA origins, even in patients with little risk for arteriosclero-sis or vasculopathy. This appearance can lead to an incorrector equivocal evaluation of vertebral origin disease. As verte-bral interventions are becoming more common, we wishedto address the reliability of CE MRA for assessment of ver-tebral ostial stenosis.

Because our interest is specifically in the signal loss seen onCE MRA at the VA origins, we wished to learn the prevalenceof isolated vertebral artery ostial stenosis. As DSA is the goldstandard for angiographic assessment, we investigated VA ostialdisease in the DSA population. Previous studies have addressedthe prevalence of proximal VA stenosis, though few studieslimit their investigation to ostial disease. Previous large cohortstudies based on catheter angiographic findings have describedthe proximal VA as the second most common site of stenosisfollowing the extracranial ICA in patients with stroke, estimatedat approximately 20% of patients.22−24 VA origin stenosis wasfound to be much less common in asymptomatic patients withrisk factors for atherosclerosis, at 2.3%.25 These studies evaluateproximal vertebral artery disease, are limited to subpopulationsof patients, and do not quantify the extent of stenosis. They,therefore, do not reflect the true prevalence of vertebral ostialstenosis, which we might use to compare with the signal lossseen in this location on CE MRA studies.

Because DSA is an invasive procedure with measurable risk,it will never be possible to design a prospective DSA studyof normal volunteers. Others have addressed this problem byperforming prospective MRA study of patients who have al-ready undergone DSA,21 or carried out retrospective study ofpatients which have been imaged by both techniques.26 Suchapproaches (1) result in small patient cohorts, (2) are biased to-ward the DSA population, and (3) cannot address questions ofdisease prevalence outside of the DSA population. We insteadstratified the DSA population by study indication, to show thatnot only is the overall prevalence of VOS in this population lowat 5.4%, but that this number is elevated because of subpopu-lations of “sicker” DSA patients, with high pretest probabilityof significant atherosclerotic disease, as determined by studyindication.

Our study reviewed 443 vertebral arteries in 329 patientsundergoing DSA to find 5.4% of cases with VA origin stenosis.The prevalence of VA stenosis increased with age, is not seen in


A B C

Fig 2. Sixty two-year-old man evaluated for right cerebellar infarction. (A) CE MRA of the neck shows signal void at the left vertebral origin.(B) DSA study and (C) CTA study obtained 2 days later shows that the proximal vertebral artery is patent. Loss of signal at the proximalvertebral artery on CE MRA study correlated with VA tortuosity seen on DSA study.

A B

Fig 3. Seventy one-year-old man evaluated for right frontal lobe infarction. CTA of the neck is without evidence of vertebral stenosis. DSAstudy the following day demonstrated vertebral origin stenosis.

patients under 39 years of age, and is seen in 12.5% of patientsover the age of 69 (Table 1). This population was further strat-ified into study indications to learn the prevalence of diseasein patient subpopulations (Table 2). A population prevalenceof 33% was seen in the posterior infarct group, 30.8% in thegroup with known carotid disease, and 54% in the group beingevaluated for symptoms of vertebrobasilar insufficiency. Muchlower prevalence of vertebral ostial stenosis was seen with thegroups being evaluated for aneurysm or AVM (4.1%), or in-tracerebral hemorrhage (3.3%). Similar observations have beenmade by others,13,24 although one study did not concur withthese correlations.14

When similar study was made of patients who had under-gone CE MRA, VA origin signal loss with the appearance of

VA origin stenosis was seen in 20% of cases. In two illustrativecases, DSA showed no stenosis, and the CE MRA study demon-strated significant signal loss at the vertebral origins (Figs 2and 3). These data argue that CE MRA significantly overesti-mates the extent of VA origin disease, giving a falsely elevatedestimation of disease prevalence, with the potential to yield afalse diagnosis of stenosis at the vertebral origin in a significantpercentage of cases. As an invasive study with associated risks,DSA is ordered only by subspecialists, after noninvasive imag-ing workup. MRA and CTA, on the other hand, are orderedroutinely in the acute setting by primary care and emergencymedicine physicians, as well as by subspecialists. These studiescan be used as front-line diagnostic tools without previous workup. Used to evaluate virtually any focal of nonfocal neurologic


Fig 4. Seventy-one-year-old man with bilateral carotid disease and right MCA infarction. CTA (A) is without significant vertebral stenosis.Subsequent DSA study (B) demonstrates significant osteal stenosis.

symptom or sign with the indication “stroke,” these studies areless amenable to classification by indication, though such an at-tempt was made (Table 4). If MRA-VOS was reflective of trueatherosclerotic narrowing, we would expect a greater correla-tion with “known stroke” than with “question stroke,” thoughno such correlation was seen.

Others have argued for the accuracy of CE MRA for as-sessment of vertebral disease.17,19,27 These studies utilize smallretrospective cohorts, and focus on the entire vertebrobasilarsystem, rather than the vertebral origins. One previous studydetails the prevalence of vertebral ostial disease using CE MRA,reporting significant (>50%) ostial disease in the general Koreanpopulation at 12.9% with prevalence of 44% in one subpopula-tion (posterior infarct group).13 These investigators argue thatthe CE MRA prevalence estimate is validated by a smallercohort of correlating DSA studies. These higher prevalencenumbers may reflect differences in the patient population, ordifferences in CE MRA protocol. The CE MRA protocol usedin our study and those of this prior study are similar. Further, ourprevalence numbers determined by DSA argue that the higherprevalence numbers obtained by MRA in both our study andin the previous study represent overestimation of disease preva-lence due to general MRI limitations rather than to differencesin MR protocol. Our data suggest that CE MRA significantlyoverestimates the prevalence of vertebral ostial narrowing. Thisobservation was recently made for CE MRA at 3T, where witha false positive rate of 52.5% for significant VOS.28

The limitations of CE MRA for evaluation of vertebral ori-gins is likely manifold. The small diameter and often tortuouspath of the proximal vertebral artery poses a challenge to thespatial resolution capabilities of magnetic resonance imaging.

Calcification at the vertebral origin also contributes to suscep-tibility effects and local signal loss. One study reports that anintravoxel dephasing effect around the stenotic segment of theartery results in overestimation of stenosis.8 Our small voxelsize should serve to minimize this effect. Cardiac and respira-tory motion together with possible local tortuosity of the verte-bral origin also likely contribute to decreased spatial resolutionof small vascular structures arising near the aorta and contributeto overestimation of vascular narrowing.20 These latter factorsare age-dependent and may offer explanation for some of thecorrelation between MRA-VOS and patient age (Table 3).

To further investigate the prevalence of VA ostial narrowingin the general population, a review of CTA findings was alsoconducted. We found that ostial disease was infrequent in theCTA population. In a small number of cases where both DSAand CTA were performed, two cases of VOS diagnosed at DSAwere not evident at CTA (see Figs 3 and 4). Lower prevalencenumbers together with the two illustrative cases described sug-gests that CTA can underestimate the prevalence of VOS. Far-res and colleagues5 propose that the multiplanar capability ofCTA offers an imaging advantage over DSA. The CTA recon-struction software used in that study afforded a MIP in a rangeof 180◦, every 10◦, as opposed to the 3-plane MIPs in the currentstudy.

We found that the most prevalent artifact of CTA was high-density incoming contrast on the side of injection limiting eval-uation of the adjacent proximal vertebral artery, occurring in7.3% of cases. For this reason, studies designed to assess thevertebral origins may benefit from contrast injection througha foot vein.22 Unlike MRA, this artifact clearly rendered thestudy uninterpretable, rather than having the potential to lead


to an incorrect diagnosis. Following tight-bolus contrast injec-tion with normal saline “chaser” is designed to minimize thisproblem, which remains a persistent limitation of CTA for studyof proximal cervical vascular disease.

A small number of patients were referred to DSA for knownvertebral disease (3 patients were followed for previous knownvertebral origin stenosis with stent), or for suspected VA diseasebased on symptoms or previous imaging findings. We only in-cluded DSA cases when DSA images of the vertebral originswere archived. In all cases, when vertebral artery origin stenosiswas diagnosed at DSA, the corresponding image was archived.However, in many cases no specific comment was made re-garding the vertebral origin in the DSA report, and no verte-bral origin image was recorded. Each of these biases, which wehave discussed, serve to increase our measured disease preva-lence numbers at DSA. We contend that because only higherrisk patients undergo DSA, findings at DSA represent an upperlimit of disease prevalence in the general population, therefore,our calculated DSA population prevalence number (5.4%) rep-resents the upper limit of disease prevalence. We show thatsubgroups of patients have higher prevalence of vertebral ostialstenosis, with the highest prevalence being found to correlatewith vertebrobasilar insufficiency (12/36 or 33%) and knownposterior infarct (7/21 or 33%).

Twenty percent of all CE MRA cases demonstrated signif-icant signal loss at the vertebral origin, giving the appearanceof greater than 50% luminal narrowing in each case. This sug-gests that CE MRA significantly overestimates the prevalenceof proximal vertebral artery disease, and has the potential toyield an incorrect diagnosis of proximal vertebral artery steno-sis in a significant percentage of cases. This is largely due tothe fact that technical limitations of MRA at the vertebral ori-gins can give rise to an appearance that cannot be distinguishedfrom a true stenosis, and this type of artifact does not occurwith CTA or DSA. CTA demonstrated few cases of vertebralostial stenosis, with a low-recorded prevalence of .8%. Preva-lence data together with a few illustrative cases where CTA andDSA were obtained on the same patient suggests that CTA canunderestimate vertebral origin disease. DSA continues to be thegold standard for stenosis at the vertebral origin.

References1. Caplan LR, Amarenco P, Rosengart A, et al. Embolism from verte-

bral artery origin occlusive disease. Neurology 1992;42:1505-1512.2. George B, Laurian C. Vertebro-basilar ischaemia. Its relation to

stenosis and occlusion of the vertebral artery. Acta Neurochir (Wien)1982;62:287-295.

3. Koroshetz WJ, Ropper AH. Artery-to-artery embolism causingstroke in the posterior circulation. Neurology 1987;37:292-295.

4. Pessin MS, Daneault N, Kwan ES, et al. Local embolism fromvertebral artery occlusion. Stroke 1988;19:112-115.

5. Borisch I, Horn M, Butz B, et al. Preoperative evaluation of carotidartery stenosis: comparison of contrast-enhanced MR angiographyand duplex sonography with digital subtraction angiography. AJNRAm J Neuroradiol 2003;24:1117-1122.

6. Remonda L, Senn P, Barth A, et al. Contrast-enhanced 3D MR an-giography of the carotid artery: comparison with conventional dig-ital subtraction angiography. AJNR Am J Neuroradiol 2002;23:213-219.

7. Romero JM, Ackerman RH, Dault NA, et al. Noninvasive evalua-tion of carotid artery stenosis: indications, strategies, and accuracy.Neuroimaging Clin N Am 2005;15:351-365, xi.

8. Cosottini M, Calabrese R, Puglioli M, et al. Contrast-enhancedthree-dimensional MR angiography of neck vessels: does dephas-ing effect alter diagnostic accuracy? Eur Radiol 2003;13:571-581.

9. Randoux B, Marro B, Koskas F, et al. Proximal great vessels of aor-tic arch: comparison of three-dimensional gadolinium-enhancedMR angiography and digital subtraction angiography. Radiology2003;229:697-702.

10. Alvarez-Linera J, Benito-Leon J, Escribano J, et al. Prospective eval-uation of carotid artery stenosis: elliptic centric contrast-enhancedMR angiography and spiral CT angiography compared with digi-tal subtraction angiography. AJNR Am J Neuroradiol 2003;24:1012-1019.

11. Hathout GM, Duh MJ, El-Saden SM. Accuracy of contrast-enhanced MR angiography in predicting angiographic stenosis ofthe internal carotid artery: linear regression analysis. AJNR Am JNeuroradiol 2003;24:1747-1756.

12. Willinek WA, Gieseke J, Conrad R, et al. Randomly seg-mented central k-space ordering in high-spatial-resolution contrast-enhanced MR angiography of the supraaortic arteries: initial expe-rience. Radiology 2002;225:583-588.

13. Kim SH, Lee JS, Kwon OK, et al. Prevalence study of prox-imal vertebral artery stenosis using high-resolution contrast-enhanced magnetic resonance angiography. Acta Radiol 2005;46:314-321.

14. Marquardt L, Kuker W, Chandratheva A, et al. Incidence andprognosis of > or = 50% symptomatic vertebral or basilar arterystenosis: prospective population-based study. Brain 2009;132:982-988.

15. Hill MD, Demchuk AM, Frayne R. Noninvasive imaging is improv-ing but digital subtraction angiography remains the gold standard.Neurology 2007;68:2057-2058.

16. North American Symptomatic Carotid Endarterectomy Trial Col-laborators. Beneficial effect of carotid endarterectomy in symp-tomatic patients with high-grade carotid stenosis. North AmericanSymptomatic Carotid Endarterectomy Trial Collaborators. N EnglJ Med 1991;325:445-453.

17. Phan T, Huston J III, Bernstein MA, et al. Contrast-enhanced mag-netic resonance angiography of the cervical vessels: experiencewith 422 patients. Stroke 2001;32:2282-2286.

18. Remonda L, Heid O, Schroth G. Carotid artery stenosis, oc-clusion, and pseudo-occlusion: first-pass, gadolinium-enhanced,three-dimensional MR angiography–preliminary study. Radiology1998;209:95-102.

19. Yang CW, Carr JC, Futterer SF, et al. Contrast-enhanced MR an-giography of the carotid and vertebrobasilar circulations. AJNR AmJ Neuroradiol 2005;26:2095-2101.

20. Krinsky G, Maya M, Rofsky N, et al. Gadolinium-enhanced 3DMRA of the aortic arch vessels in the detection of atheroscle-rotic cerebrovascular occlusive disease. J Comput Assist Tomogr1998;22:167-178.

21. Wright VL, Olan W, Dick B, et al. Assessment of CE-MRAfor the rapid detection of supra-aortic vascular disease. Neurology2005;65:27-32.

22. Farres MT, Grabenwoger F, Magometschnig H, et al. Spiral CTangiography: study of stenoses and calcification at the origin of thevertebral artery. Neuroradiology 1996;38:738-743.

23. Hass WK, Fields WS, North RR, et al. Joint study of extracranialarterial occlusion. II. Arteriography, techniques, sites, and compli-cations. JAMA 1968;203:961-968.

24. Wityk RJ, Chang HM, Rosengart A, et al. Proximal extracranialvertebral artery disease in the New England Medical Center Pos-terior Circulation Registry. Arch Neurol 1998;55:470-478.

25. Hennerici M, Aulich A, Sandmann W, et al. Incidence of asymp-tomatic extracranial arterial disease. Stroke 1981;12:750-758.


26. Wang S, Poptani H, Woo JH, et al. Amyotrophic lateral sclerosis:diffusion-tensor and chemical shift MR imaging at 3.0 T. Radiology2006;239:831-838.

27. Leclerc X, Martinat P, Godefroy O, et al. Contrast-enhanced three-dimensional fast imaging with steady-state precession (FISP) MR

angiography of supraaortic vessels: preliminary results. AJNR Am JNeuroradiol 1998;19:1405-1413.

28. Choi HS, Kim DI, Kim DJ, et al. Accuracy of 3 T MR angioraphyin vertebral artery stenosis and coincidence with other cerebrovas-cular stenoses. Neuroradiology 2010;52:893-898.


Documents

Vertebral Artery Ostial Stenosis: Prevalence by Digital Subtraction Angiography, MR Angiography, and CT Angiography