chronic cerebrospinal venous insufficiency and iron …...2010/08/10 · chronic cerebrospinal...

Chronic cerebrospinal venous insufficiency and iron depositionon susceptibility-weighted imaging in patients with multiple sclerosis: a pilot case-control studyR. ZIVADINOV 1, 2, C. SCHIRDA 1, M. G. DWYER 1, M. E. HAACKE 3, B. WEINSTOCK-GUTTMAN 2, E. MENEGATTI 4, M. HEININEN-BROWN 1, C. MAGNANO 1, A. M. MALAGONI 4, D. S. WACK 1,D. HOJNACKI 2, C. KENNEDY 1, E. CARL 1, N. BERGSLAND 1, S. HUSSEIN 1, G. POLONI 1,I. BARTOLOMEI 4, F. SALVI 4, P. ZAMBONI 41Buffalo Neuroimaging Analysis Center, University at Buffalo, Buffalo, NY, USA 2The Jacobs Neurological Institute, University at Buffalo, Buffalo, NY, USA3Wayne State University, MR Research Facility, Department of Radiology, Detroit, MI, USA4Vascular Diseases Center, University of Ferrara-Bellaria Neurosciences, Ferrara and Bologna, Italy

Aim. Chronic cerebrospinal venous insufficiency (CCSVI) isa vascular phenomenon recently described in multiple scle-rosis (MS) that is characterized by stenoses affecting themain extracranial venous outflow pathways and by a highrate of cerebral venous reflux that may lead to increasediron deposition in the brain. Aim of this study was to inve-stigate the relationship between CCSVI and iron depositionin the brain of MS patients by correlating venous hemody-namic (VH) parameters and iron concentration in deep-graymatter structures and lesions, as measured by susceptibi-lity-weighted imaging (SWI), and to preliminarily define therelationship between iron measures and clinical and othermagnetic resonance imaging (MRI) outcomes.Methods. Sixteen (16) consecutive relapsing-remitting MSpatients and 8 age- and sex-matched healthy controls (HC)were scanned on a GE 3T scanner, using SWI. Results. All 16 MS patients fulfilled the diagnosis of CCSVI(median VH=4), compared to none of the HC. In MS patients,the higher iron concentration in the pulvinar nucleus of thethalamus, thalamus, globus pallidus, and hippocampus wasrelated to a higher number of VH criteria (P

100%, and negative predictive value 100%), andwas confirmed using selective venography.1-3

A role for CCSVI in MS is consistent not onlywith the well known perivenular distribution of MSlesions, but also with recent studies that have found:1) a central vein in the long axis of inflammatoryMS lesions using ultra-high field MRI 6-9 and 2)abnormally high levels of redox active metals, par-ticularly iron, identified with an MRI techniquecalled susceptibility-weighted imaging (SWI).7, 10

MRI, particularly the new iron-sensitive SWItechnique, stands up as an essential tool in under-standing the link between CCSVI and MS patho-biology.11, 12 SWI is a new neuroimaging techniquethat uses tissue magnetic susceptibility differencesto generate a unique contrast, different from thatof spin density, T1, T2, and T2*.11, 12 SWI involvesthe use of both magnitude and phase images froma high-resolution, three-dimensional (3D) fullyvelocity-compensated gradient recalled echo(GRE) sequence.11, 13 Phase masks are createdfrom the MR phase images, and multiplying thesewith the magnitude images increases the con-spicuousness of the smaller veins and other sourcesof susceptibility effects (such as iron deposits),which are depicted using minimal intensity pro-jection (minIP). The use of SWI in MS is in itsinfancy 6-10 and more studies between SWI andclinical and other conventional and non-conven-tional MRI outcomes need to be conducted.

It is not clear whether the venous reflux of themain extracranial CS venous routes that interferewith normal CS venous outflow may cause chron-ic iron deposition in the CNS. According to thehypothesis that has been recently postulated,4, 5 ironstorage in the CNS can precede inflammation andtrigger an inflammatory chain reaction. Therefore,investigating the relationship between accumula-tion of iron deposits on SWI and the presence andseverity of CCSVI could be a very important steptoward better understanding of MS pathogenesis.

The aim of this study was to preliminarily inves-tigate the relationship between CCSVI and irondeposition in the brains of MS patients and HCby correlating the number of VH criteria and theVH insufficiency severity score (VHISS) to irondeposition in deep-gray matter (DGM) structuresand lesions, as measured by SWI. We also aimedto explore the relationship between iron deposi-tion measures and demographic, clinical and otherconventional MRI outcomes in patients with MS.

Materials and methods

Study population and design

We enrolled 16 consecutive MS patients accord-ing to the McDonald Criteria,14 and a group ofeight HC matched for age and sex. The patientsand controls were recruited from the Bellaria Hos-pital, Bologna, Italy (8 patients, 4 controls) andthe Jacobs Neurological Institute, University atBuffalo, NY, USA (8 patients, 4 controls). The inclu-sion criteria required a relapsing remitting (RR)MS disease course,15 an Expanded Disability Sta-tus Scale (EDSS) 16 between 0-5.5, age 18-65 years,disease duration between 5 and 10 years, beingon treatment with currently FDA-approved dis-ease-modifying treatments, and having normalrenal function (creatinine clearance of >58mL/min). Exclusion criteria included an acuterelapse and/or steroid treatment within the 30days preceding study entry, pre-existing medicalconditions associated with brain pathology (e.g.,neurodegenerative disorder, positive history ofalcohol abuse, etc.), and abnormal renal function.

The Italian patients and controls were requiredto travel to Buffalo, NY, where all clinical andinstrumental study assessments were conductedover a period of four days. The Italian researchgroup conducted the VH/Doppler assessment andthe Buffalo research group conducted clinical andMRI examinations. All investigators conductingassessments were blinded to the clinical, demo-graphic, and subject group (MS or HC) charac-teristics to the extent possible given the disabili-ties present among MS patients and cultural dif-ferences. The clinical, VH, and MRI assessmentswere conducted on the same day for each subject.All patients underwent a complete physical andneurological examination, EDSS and MultipleSclerosis Functional Composite (MSFC) assess-ments,17 followed by ECD and MRI. The studywas approved by the Human Subjects InstitutionalReview Board of the University at Buffalo andItaly (for the Italian patients). A written informedconsent was obtained from all subjects.

Echo-color-Doppler assessments of cerebral venoushemodynamics

A combined transcranial and extracranial ECDprovides validated measures of VH parametersand enables the assessment of CCSVI. Cerebral

Vol. 29, No. 2 INTERNATIONAL ANGIOLOGY 159

160 INTERNATIONAL ANGIOLOGY April 2010

venous return was examined using the echo-colorDoppler (ECD Esaote-Biosound My Lab 25)equipped with 2.5 and 7.5-10 Mhz transducers(Genoa, Italy), with the subject positioned on abed tilted at 90° and 0°, and the vessels insonat-ed with an angle of 60°, as previously described.2

We focused on the detection of 5 anomalous VHcriteria affecting CS return, as previouslydescribed.2, 3 These included: 1) reflux in the IJVsand/or in the VVs assessed in both sitting andsupine positions; 2) reflux in the deep cerebralveins (DCVs); 3) b-mode detection of stenoses inthe IJVs in the form of annuli, webs, septa, or mal-formed valves; 4) absence of Doppler signal in theIJV and/or in the VVs, even after forced deepbreaths, in both sitting and supine positions and5) presence of a negative difference between supineand upright position in the cross-sectional area(CSA) of the IJV. The total number of VH criteriawas summed up. Operationally, a subject was con-sidered CCSVI positive if two or more proposedVH criteria were fulfilled.2, 3

We also calculated VH insufficiency severityscore (VHISS), as previously described.18 TheVHISS is defined as a sum of the scores con-tributed by each individual VH criterion. The for-mula for VHISS calculation is: VHISS=VHISS1 +VHISS2 + VHISS3 + VHISS4 + VHISS5. The sub-scripts in this formula indicate the subscores forthe 5 VH criteria. The VHISS is an ordinal mea-

sure of the overall extent and number of VH flowpattern anomalies, with a higher value of VHISSindicating a greater severity of VH flow patternanomalies. The minimum possible VHISS valueis 0 and the maximum 16.

MRI acquisition and analysis

IMAGE ACQUISITION

All subjects were examined on a 3T GE SignaExcite HD 12.0 Twin Speed 8-channel scanner(General Electric, Milwaukee, WI, USA), with amaximum slew rate of 150T/m/s and maximumgradient amplitude in each orthogonal plane of50mT/m (zoom mode). A multi-channel head andneck (HDNV) coil manufactured by GE was usedto acquire the following sequences: 2-dimension-al (2D) multi-planar dual fast spin-echo (FSE) pro-ton density (PD) and T2-weighted image (WI);Fluid-Attenuated Inversion-Recovery (FLAIR); 3Dhigh resolution (HIRES) T1-WI using a fast spoiledgradient echo (FSPGR) with magnetization-pre-pared inversion recovery (IR) pulse; SWI (sus-ceptibility weighted imaging) and SE T1-WI withand without using a single dose intravenous bolusof 0.1 mMol/kg gadolinium (Gd)-DTPA 5 min afterinjection. No Gd active lesions were found in thisstudy sample. All scans were prescribed in an axial-oblique orientation, parallel to the subcallosal line.

One average was used for all pulse sequences.

Figure 1.—Reconstruction of the SWI data was done off-line. Following channel recombination, the magnitude image wasprocessed using the non-parametric non-uniform intensity normalization program N3 (A). The high-pass filtered phase image(B) was used to generate a phase mask that was multiplied 4 times onto the magnitude image to generate the SWI image (C).


With the exception of SWI, all sequences wereacquired with a 256x192 matrix (freq x phase),field-of-view (FOV) of 25.6cm x 19.2cm (256x256matrix with Phase FOV=0.75), for an in-plane res-olution of 1 mm x 1 mm. For all 2D scans (PD/T2,FLAIR and SE T1), 64 slices were collected, thick-ness of 2 mm, and no gap between slices. For the3D HIRES IR-FSPGR, 184 locations wereacquired, 1 mm thick, providing for isotropic res-olution.

Other relevant parameters were as follows: fordual FSE PD/T2, echo and repetition times (TEand TR) TE1/TE2/TR=9/98/5300ms, flip angle(FA)=90, echo train length ETL=14, acquisitiontime (AT)=5:08 (min:s); for FLAIR, TE/TI/TR=120/2100/8 500 ms (inversion time, TI), FA=90, ETL=24,AT=6:49; for SE T1-WI, TE/TR=16/600 ms, FA=90,AT=6:11; for 3D HIRES T1-WI, TE/TI/TR=2.8/900/5.9 ms, FA=10, AT=9:18.

SWI was acquired using a 3D flow-compensat-ed GRE sequence with 64 locations, 2mm thick-ness, a 512x192 matrix, FOV= 25.6cm x 19.2cm(512x256 matrix with PhaseFOV=0.75), for an in-plane resolution of 0.5 mm x 1 mm, flip angleFA=12, TE/TR=22/40 ms. Raw (k-space) data wassaved and transferred to an off-line Linux work-station for post-processing using in-house devel-oped software written in Matlab (MathWorks Inc.,Novi, MI, USA) (Figure 1).

LESION MEASURES

The T2- and T1- lesion volumes (LVs) were mea-sured using a semi-automated edge detection con-touring/thresholding technique previouslydescribed.19

GLOBAL AND TISSUE SPECIFIC ATROPHY MEASURES

For brain extraction and tissue segmentationinto gray matter (GM) and white matter (WM),the SIENAX cross-sectional software tool wasused, with corrections for T1-hypointensity mis-classification.20 Normalized volumes of the wholebrain (NBV), GM volume (NGMV), neocorticalvolume (NCV) and WM volume (NWMV) wereacquired as previously described.21

SWI ANALYSIS

An overview of SWI processing and the analy-sis steps involved are provided in Figure 2.

Data for each channel was reconstructed byzero-filling each slab location (slice) to 768x576to produce images with an interpolated resolu-tion of 0.33 mm x 0.33 mm. To ensure proper com-position of multi-channel data for the magnitudeand phase images, a channel re-centering and nor-malization process was employed.13 Followingchannel recombination, the magnitude image wasprocessed using the non-parametric non-uniform

Tissue segmentation:FIRST and manual ROIs

Conventionallesion analysis

Regional iron quantification SWI lesion analysis

Overlaps with conventionallesion analysis

Mean phase value Volume of high-iron voxels Lesion volumes

SWI pre-processing

Unwarping andco-registration

SWI acquisition

Figure 2.—Overview of MRI processing steps. Some intermediate steps are omitted or simplified for clarity.


intensity normalization program N3.22 Phaseimages were high-pass filtered 12 using a 64x48Hanning window to better visualize iron deposits(Figure 1). The high-pass filtered phase image wasused to generate a phase mask. SWI images wereobtained by multiplying the phase mask 4 timesonto the inhomogeneity-corrected magnitudeimage. Correction for geometric field-induced dis-tortions was carried out via a non-linear imageunwarping technique.23 The images were then co-registered into a corresponding subject-specifichigh-resolution upsampled FLAIR space using arigid-body linear image registration technique.24In addition, the phase image was linearly trans-formed and thresholded to create a phase mask.This phase mask was then composited with themagnitude image to create an SWI image as pre-viously described.10, 25

A manual region-of-interest approach was usedto identify SWI magnitude-visible lesions and SWIphase-visible lesions on corresponding images

(Figure 3). Phase-visible lesions were further sub-divided into nodular, ring-shaped, scattered andcortical categories (Figure 4). Nodular lesions weredefined as focal areas of homogeneously abnor-mal phase hypointensity. Ring-shaped (annular)lesions were defined as focal areas of abnormalphase hypointensity surrounding an area of isoin-tensity. Scattered lesions were defined as focalareas of phase hypointensity interspersed withinan area of isointensity and showing a clear spa-tial coherence. Finally, cortical lesions were definedas areas of focal phase hypointensity of any pat-tern contained within the cortex. All lesion maskswere co-registered into the subject-specific upsam-pled FLAIR space, at which point spatial overlapmaps were calculated. These were used to calcu-late mean magnitude and phase values for eachlesion type and each intersection of lesion types.

The high-resolution T1-weighted image wasthen segmented and parcellated into various tis-sue classes and regions. Subcortical GM struc-

Figure 3.—A manual region-of-interest approach was used to identify SWI magnitude (light blue) visible lesions and SWIphase (green) visible lesions on corresponding images. Phase-visible lesions were further subdivided into nodular, ring-sha-ped, and scattered categories. All lesion masks were co-registered into the subject-specific upsampled FLAIR space, at whichpoint spatial overlap T2 lesion (dark blue) and T1 lesion (purple) maps were calculated. These were used to calculate meanmagnitude and phase values for each lesion type and each intersection of lesion types.


tures were segmented using a combination ofsemi-automated edge-contouring and fully-auto-mated model-based registration approaches. Mostregions were identified automatically usingFMRIB’s integrated registration and segmenta-tion tool (FIRST);26 specifically, the thalamus, cau-date, putamen, globus pallidus, hippocampus,amygdala, and nucleus accumbens were identi-fied in this way. Regions not yet supported byFIRST, such as the red nucleus, pulvinar nucleusof the thalamus, and substantia nigra, were iden-tified semi-automatically using a technique sim-ilar to that used for the lesion delineation describedabove. All regions were identified separately ineach hemisphere. In addition, the images weredeskulled 27 and segmented into tissue-specificcompartments via an expectation maximizationoptimized hidden Markov random field statisti-cal model (FMRIB’s FAST tool).28 As part of thissegmentation process, T2 and T1 lesion maskswere also co-registered into the space of the high-resolution T1 and used to correct any lesion mis-classification due to T1 hypointensity. The lesionmasks were used to distinguish between wholebrain, GM and WM tissue compartments and nor-mal-appearing tissue compartments. As with thelesion measurements, mean phase and magnitudevalues were calculated for each subcortical regionand each tissue compartment.

In addition to mean phase iron data (Figures 5and 6), mean and volume values of those voxelswith phase values indicative of abnormal iron con-

tent (Figure 7) were calculated to yield high-irontissue volume (HITV) and high-iron tissue meaniron concentration (HITMIC). Voxel classificationwas performed via simple thresholding at a region-specific cut-point corresponding to the phase valuetwo standard deviations below the mean valuefound in the same area in a HC population. Theiron volume fraction (IVF) was obtained by divid-ing the HITV of each region by the total volumeof that region. The normative reference data points(both means and standard deviations) for thesemeasurements were pre-calculated on a group of60 HC with various age duration and no historyof neurological disease (unpublished data). Pro-cessing for that group was identical to the othersteps described above. More negative mean phaseand more positive HITMIC values represent high-er iron concentration within a region. Mean val-ues are presented in parts per billion (ppb). High-er HITV values indicate a greater dispersion ofiron within a region, and are presented in mm3.

Statistical analysis

Statistical analysis was performed using the Sta-tistical Package for the Social Sciences (SPSS), ver-sion 16.0. The age and proportions of females andmales in the MS and HC groups were assessed withthe Student’s t-test and Fisher Exact test, respec-tively. Because not all of the MRI distributions weredistributed normally as assessed by the Komologrov-Smirnov test (P


between MS patients and HC were assessed usinggeneral linear model (GLM) analysis adjusted forgender and age. A post-hoc Bonferroni correctionwas performed for all analysis of covariance (ANCO-VA) calculations in order to correct for multiplecomparisons directly using the SPSS-related pro-cedure. The non-parametric Mann-Whitney test wasused to assess the whether the VH, VHISS and otherclinical and MRI parameters differed between theMS and HC groups. Correlation analysis (Spear-man or Pearson, as appropriate) was used to assessthe relationship between SWI measures and VH,VHISS, and clinical and other MRI parameters. Nocorrection for multiple comparisons was performedfor correlation analyses. Given the exploratory natureof the study and low sample size, we consideredp


Figure 6.—Mean phase iron concentration maps in healthy control (HC) subjects (A) and multiple sclerosis (MS) patients(B) in the pulvinar nucleus of the thalamus. The iron concentration map is displayed on a scale ranging from dark (low ironconcentration) to bright (high iron concentration) colors. MS patients show more bright areas in the the pulvinar nucleusof the thalamus, as well as increased intensities.

Figure 7.—High-iron voxels identified in deep gray matter structures by thresholding of the mean phase iron concentrationmaps at two standard deviations below the healthy control mean (red color) are significantly more prevalent in multiplesclerosis patients (A) than in normal controls (B).


significantly higher (P


detected in phase lesions (5.5 ppb) as expected,and there was no difference between the iron con-centration in the intersection of phase lesions thatoverlapped with T2, T1 or magnitude lesions.

Relationship between CCSVI and SWI measures

Tables 4 and 5 show the correlation analysisbetween Doppler (number of VH criteria andVHISS) and MRI (HITMIC and IVF) measure-ments. No relationship was observed for the HCbetween these variables.

In MS patients, increased HITMIC of the rightglobus pallidus, right and left pulvinar nucleusof the thalamus, and left hippocampus was relat-ed to a higher number of VH criteria. The rightglobus pallidus was the only measure that wasassociated with VHISS. Significant correlationswere found between higher IVF of the right thal-amus, the right and left pulvinar nucleus of thethalamus and the right and left hippocampus

and higher number of VH criteria. Higher IVFof the right and left pulvinar nucleus of the thal-amus and the right hippocampus was related tohigher VHISS.

There was a significant association between thehigher number of VH criteria and higher meanphase iron concentration of overlapping T2 (r=-0.64, P=0.007) and T1 (r=-0.56, P=0.023) phaseLVs. The nodular mean phase iron concentrationshowed the highest correlation with the numberof VH criteria (-0.75, P


showed significant correlations with increasedage, lower MSFC and higher EDSS (Table IV). Theglobus pallidus was related to increased age at

diagnosis, and the pulvinar nucleus of the thala-mus to longer disease duration.

In MS patients, IVF did not show a correlation

TABLE II.—Iron deposition in deep-gray matter structures on susceptibility-weighted imaging in relapsing-remitting mul-tiple sclerosis patients and healthy controls.

HITV in MS HITV in HC IVF in MS IVF in HC(mm3) (mm3)

MS: multiple sclerosis; HC: healthy controls; HITMIC: high-iron tissue mean iron concentration; IVF: iron volume fraction. All values are represented by mean and SD. Iron concentrations are expressed in parts per billion (ppb). More negative iron concentration values representhigher iron concentration, whereas more positive values represent lower concentration. Thresholded phase iron volumes are represented in mm3. Thre-sholding mean phase value voxel classification was performed via simple thresholding at a region-specific cut-point representing the phase value two stan-dard deviations below the mean value found in the same area in a HC population. IVF was obtained by dividing the thresholded phase iron concentrationvolume of a particular region with the total volume of that region. These normative data points (both means and standard deviations) were pre-calculatedon a group of 60 HC with no history of neurological disease (unpublished data). The differences between MS patients (N.=16) and healthy controls (N.=8) were assessed by general linear model (GLM) analysis adjusted for gender andage. A post-hoc Bonferroni correction was performed for all analysis of covariance (ANCOVA) calculations in order to correct for multiple comparisonsdirectly using a SPSS-related procedure.*P


with age, age at diagnosis, or EDSS (except for theright red nucleus), but lower IVF of various DGMregions showed significant correlations with longerdisease duration and lower MSFC (Table V).

The SWI lesion outcomes were significantly relat-ed only to disease duration. Higher mean phase ironconcentrations of overlapping T1 (r=-0.78, P


to the HITMIC measurements. There was also arelationship between a higher number of VH cri-teria and higher mean phase iron concentrationof overlapping T2, T1 phase lesion volumes. Thesefindings suggest that CCSVI may be an importantmechanism related to iron deposition in brainparenchyma of MS patients with a predilectionfor DGM and phase lesions visible on SWI.

The source of iron deposition in MS has not beencompletely elucidated.29 Previous studies in MShave identified iron accumulation in DGM 30 andplaques,10 but they did not establish whether irondeposition was a primary phenomenon in MSpathology or secondary to chronic inflammationin MS. Iron deposition in MS patients may derivefrom myelin/oligodendrocyte debris, destroyedmacrophages, or can be the product of hemorrhagesfrom damaged brain vessels.4, 29 In addition, ironoverload may also lead to oxidative mitochondri-al injury through Fenton reaction and release ofphospholipid-rich cellular membrane elements inMS.31 The mechanism of direct damage to the CNSby iron might be related to oxidative stress and thegeneration of toxic free radicals.31 Recently, it has

been proposed that iron deposits in MS are a con-sequence of altered CS return and chronic insuf-ficient venous drainage.4, 5 According to thishypothesis, an excessive amount of iron (result ofCS venous reflux) may cause the damage to theblood-brain-barrier and consequent disturbedmicrocirculation, leading to erythrocyte extrava-sation as a primary source of iron stores. Histo-logical and SWI studies confirm erythrocyteextravasation in brain plaques of MS patients andthe presence of iron-laden macrophages at theperivenular level.5-10, 32-34 In histological studies,an iron increase was reported in the DGM in MSpatients compared to HC.32, 34 It has been observedthat the cell involved in iron overload with thegreatest effect on immunity is the macrophage,and there is a close relationship between iron andthe major cells of adaptive immunity, the T lym-phocytes, since they are major players in recyclingof the iron from hemoglobin.35 Therefore, ironmay be a powerful chemotactic stimulus thatattracts macrophages and contributes to or caus-es initial activation of T-cell autoimmunity. Theresults from this pilot study suggest that CCSVI-

TABLE IV.—Relationship between thresholded high-iron mean iron concentration (HITMIC) on susceptibility-weightedimaging and venous hemodynamic, demographic, clinical, and conventional MRI outcomes in sixteen patients withmultiple sclerosis.

MSFC EDSS T2-LV T1-LV NBV NGMV NCV NWMV

VH: venous hemodynamic criteria (0-5); VHISS: venous hemodynamic insufficiency severity score (0-16); MSFC: Multiple Sclerosis Functional Composi-te; EDSS: Expanded Disability Status Scale; LV: lesion volume; NBV: normalized brain volume; NGMV: normalized gray matter volume; NCV: normalizedneocortical volume; NWMV: normalized white matter volumeThe correlation analysis was performed using Spearman and Pearson correlation coefficients, as appropriate.*P


related CS blood flow disturbances may lead toiron deposition in DGM structures and in lesions,as measured by SWI. Whether CCSVI is a prima-ry cause of iron deposition in MS that may pre-cede T cell activation still remains to be elucidat-ed in future studies. Certainly, the preliminaryfindings from this pilot study are encouraging andneed to be extended to a larger sample of MSpatients, HC and patients with other inflamma-tory and non-inflammatory neurological disor-ders, to different disease subtypes and differentdisease forms. Such studies are currently underway in our Centers.

Iron is a paramagnetic substance that reducesT2 relaxation time on MRI, resulting in hypointen-sity on T2-weighted images. Bakshi et al. showedalmost a decade ago the presence of DGM T2hypointensity in patients with MS.36, 37 Theseauthors recently reported that the DGM T2hypointensity is present even at the first symptomonset.38 Various imaging techniques have beenused to evaluate the amount of DGM iron depo-sition, including T2 hypointensity,36, 37 relaxome-try,39, 40 magnetic field correlation 41 and SWI.7, 10-12 SWI is a unique MRI technique that offers away to visualize tissues affected by iron deposi-

tion in the form of ferritin, deoxyhemoglobin orhemosiderin.11 The MS patients enrolled in thisstudy had a disease duration of approximately 7.5years and showed 5-40% higher iron deposition(according to different iron measures and regions)(Table II) compared to age- and sex-matched HC(Table II, Figures 5-7). Therefore, measurementof iron deposition in the DGM structures of MSpatients on SWI may be an important biomarkerof the disease process and can predict iron con-tent more accurately than other available imag-ing techniques.7, 10-12

In the present study, the differences betweenMS patients and HC in DGM structures were inves-tigated by four different SWI measures, includ-ing mean phase iron concentration, HITMIC, HITVand IVF (Table II, Figures 5-7). Thresholding wasperformed at a region-specific cut-point repre-senting the phase value two standard deviationsbelow the mean value found in the same area inHC, as previously proposed.42 This SWI HITMICmeasure differentiated MS and HC for the pulv-inar nucleus of the thalamus, the thalamus, globuspallidus and hippocampus regions (Figure 7). Italso yielded somewhat higher correlation coeffi-cients with clinical and MRI outcomes compared

TABLE V.—Relationship between iron volume fraction (IVF) on susceptibility-weighted imaging and venoushemodynamic, demographic, clinical and conventional MRI outcomes in sixteen patients with multiple sclerosis.

Regional HITMIC VH VHISS Age Age at Diseasediagnosis duration

0.75***0.73***0.47*0.47*

0.4*

0.49*0.42*

0.48*

0.46*052**0.43*0.51*

0.46*0.47*0.41*

Right caudate Left caudateRight putamenLeft putamen Right globus pallidus Left globus pallidusRight thalamusLeft thalamus Right pulvinar thalamus Left pulvinar thalamus Right hippocampusLeft hippocampusRight amygdala Left amygdala Right accumbens Left accumbens Right red nucleus Left red nucleusRight substantia nigra Left susbstantia nigra

VH: venous hemodynamic criteria (0-5); VHISS: venous hemodynamic insufficiency severity score (0-16); MSFC: Multiple Sclerosis Functional Composi-te; EDSS: Expanded Disability Status Scale; LV: lesion volume; NBV: normalized brain volume; NGMV: normalized gray matter volume; NCV: normalizedneocortical volume; NWMV: normalized white matter volume. The correlation analysis was performed using Spearman and Pearson correlation coefficients,as appropriate. *P


to the IVF. To the best of our knowledge, this isthe first study in which the HITV (Figure 7) wasquantified. No differences were found betweenMS patients and HC. It is widely known that DGMatrophy is largely present from the earliest clini-cal phases in patients with MS.43 Therefore, thecontent of iron deposits in those structures couldbe affected by the decreased volume of the samestructures. In order to normalize the HITV for thevolume of the DGM structures, we created theIVF. IVF differentiated the MS and HC for all DGMstructures more significantly than HITMIC and,additionally, detected differences between the twogroups in the putamen. IVF also correlated some-what more strongly and consistently with theCCSVI criteria and disease duration, but showedweaker correlations with clinical and MRI out-comes. These findings suggest that HITMIC andIVF SWI measures are detecting somewhat dif-ferent aspects of iron deposition in patients withMS and should be complementary.

Different SWI measures applied to this pilotstudy identified the pulvinar nucleus of the thal-amus, the thalamus, globus pallidus and hip-pocampus as the DGM structures that differenti-ated MS patients from HC, and that were associ-

ated with impaired VH, clinical and MRI out-comes. The particular involvement of theseregions, but not of some other explored DGMregions, indicates that irrigation of the vascularpathway from the IJV to these regions should beinvestigated in more detail. This work is underway in a collaboration between our Centers.

A recently published study detected SWI lesionsnot seen on conventional MRI and classified themin six different categories.10 We used a manualregion-of-interest approach that identified SWImagnitude and phase lesions (Figure 3). Phase-visible lesions were further subdivided into nodu-lar, ring-shaped, scattered and cortical subcate-gories (Figure 4). Magnitude lesions were alsoassessed. We attempted to evaluate the lesionsassociated with veins 10 but determined that theassessment was not reliable in most of the scans.All lesion masks were co-registered into the sub-ject-specific upsampled FLAIR space and an over-lap with respect to T2, T1 and magnitude inter-sections of phase lesions was automaticallyobtained. The mean number of phase and mag-nitude lesions was more than 50% lower com-pared to T2 and T1 lesions. The phase LV was 87%lower compared to T2-LV, whereas the magnitude

TABLE IV.—Relationship between thresholded high-iron mean iron concentration (HITMIC) on susceptibility-weightedimaging and venous hemodynamic, demographic, clinical, and conventional MRI outcomes in sixtelerosis.

MSFC EDSS T2-LV T1-LV NBV NGMV NCV NWMV

-0.38*-0.38*-0.36*

-0.42*-0.41*-0.49*-0.42*

-0.56**

0.71***

-0.45*-0.39*

-0.5**-0.49*-0.52**

-0.43*-0.43*

-0.52**-0.56**-0.52**-0.42**

-0.46*-0.42*-0.44*-0.47*

-0.43*

-0.47*-0.48*-0.51*-0.43*


LV was 43% lower compared to T2-LV (Table III).The overlap between phase and magnitude lesionsand T2 and T1 lesions ranged from 17.5% to 28.3%.This indicates that phase and magnitude lesionslargely do not overlap with T2 and T1 lesions. Themost representative phase lesion type was nodu-lar. As expected, the highest iron concentrationwas detected in the phase lesions and in the over-lapping intersections of phase T2, T1 and magni-tude lesions. The iron concentration did not dif-fer between different types of phase lesions ortheir overlaps. The iron concentration was sig-nificantly lower when only T2, T1 and magnitudelesions volumes were considered without the over-laps with the phase lesion masks (Table III). Ofall the lesion types measured in the study (num-ber, volume and iron concentration), the phaseiron concentration was the only measure to showa relationship with CCSVI, clinical and MRI out-comes. The higher phase iron concentration ofoverlapping T2, T1 and nodular phase lesions wasrelated to a higher number of VH criteria, indi-cating that focal iron deposition outside of theDGM structures may be directly related to restrict-ed CS venous outflow. This was not observed inHC. The higher mean phase iron concentrationwas also strongly related to longer disease dura-tion and other conventional MRI outcomes, includ-ing LVs and whole brain and GM atrophy vari-ables. All in all, these preliminary findings sug-gest that phase lesions and their overlaps with T2,T1 and magnitude lesions represent an importantnew lesion category in MS patients. Given encour-aging correlations between these lesions and VH,clinical and MRI outcomes, further studies areneeded to explore these associations on a largersample size.

One of the secondary aims of this study was toobtain preliminary correlations between SWI DGMiron measures and clinical and MRI outcomes.The HITMIC of the caudate, putamen, thalamus,pulvinar nucleus of the thalamus, hippocampusand red nucleus showed a robust relationship withEDSS and MSFC. While the IVF of the sameregions was related to MSFC, the relationship withEDSS was significant only for the red nucleus.This suggests that measurement of iron concen-tration on SWI may be an important predictor ofdisability in patients with MS and confirms pre-vious data obtained by other iron-related imag-ing techniques.36, 44 The HITMIC of various regions

was related to age, while IVF was not, indicatingthat higher iron concentration occurs as part ofthe aging process. The decreased IVF showed arobust correlation with longer disease duration,suggesting higher accumulation of iron in DGMstructures increases as the disease continues.

In general, the caudate, putamen, thalamus,pulvinar nucleus of the thalamus, hippocampusHITMIC and IVF were related to whole brain andGM atrophy, with particular robustness in corti-cal regions. These interesting findings warrantfurther investigation in both cross-sectional andlongitudinal studies.

One of the main shortcomings of the study isrepresented by the absence of patients with inflam-matory and non inflammatory neurological dis-orders in the control group. In fact, this drawbackdid not allow us to establish whether the presenceof increased brain iron deposition and CCSVI isrestricted to MS patients or shared with otherinflammatory neurological conditions. We are inthe process of collecting these data.

Conclusions

In conclusion, the findings from this pilot studysuggest that CCSVI may be an important mecha-nism related to iron deposition in the brainparenchyma of MS patients. In turn, iron depo-sition, as measured by SWI, is a modest-to-strongpredictor of disability progression, LV accumula-tion and atrophy development in patients withMS.

Acknowledgements.—We would like to thank all subjectsparticipating in the study, and in particular Italian partici-pants who traveled to Buffalo, NY to undergo all study exa-minations over 4 days. This study was in part supported bythe Hilarescere Foundation and the Buffalo NeuroimagingAnalysis Center. The authors wish to thank Zohara Sternberg,a researcher at the Jacobs Neurological Institute in Buffalo,who suggested and helped arrange the first meeting betweenthe researchers involved in the present cooperative study. Inaddition, we thank other study contributors that were invol-ved in this project. We also would like to thank Eve Salczyn-ski for her technical assistance in the preparation of this manu-script.

References

1. Zamboni P, Galeotti R, Menegatti E, Malagoni AM, Gia-nesini S, Bartolomei I et al. A prospective open-label studyof endovascular treatment of chronic cerebrospinal venousinsufficiency. J Vasc Surg 2009;50:1348-58 e1-3.


2. Zamboni P, Galeotti R, Menegatti E, Malagoni AM, Tac-coni G, Dall’Ara S et al. Chronic cerebrospinal venousinsufficiency in patients with multiple sclerosis. J Neu-rol Neurosurg Psychiatry 2009;80:392-9.

3. Zamboni P, Menegatti E, Galeotti R, Malagoni AM, Tac-coni G, Dall’Ara S et al. The value of cerebral Dopplervenous haemodynamics in the assessment of multiplesclerosis. J Neurol Sci 2009;282:21-7.

4. Singh AV, Zamboni P. Anomalous venous blood flow andiron deposition in multiple sclerosis. J Cereb Blood FlowMetab 2009;29:1867-78.

5. Zamboni P. The big idea: iron-dependent inflammationin venous disease and proposed parallels in multiple scle-rosis. J R Soc Med 2006;99:589-93.

6. Ge Y, Zohrabian VM, Osa EO, Xu J, Jaggi H, Herbert J etal. Diminished visibility of cerebral venous vasculaturein multiple sclerosis by susceptibility-weighted imagingat 3.0 Tesla. J Magn Reson Imaging 2009;29:1190-4.

7. Hammond KE, Metcalf M, Carvajal L, Okuda DT, Srini-vasan R, Vigneron D et al. Quantitative in vivo magneticresonance imaging of multiple sclerosis at 7 Tesla withsensitivity to iron. Ann Neurol 2008;64:707-13.

8. Tallantyre EC, Brookes MJ, Dixon JE, Morgan PS, Evan-gelou N, Morris PG. Demonstrating the perivascular distri-bution of MS lesions in vivo with 7-Tesla MRI. Neuro-logy 2008;70:2076-8.

9. Tallantyre EC, Morgan PS, Dixon JE, Al-Radaideh A,Brookes MJ, Evangelou N et al. A comparison of 3T and7T in the detection of small parenchymal veins withinMS lesions. Invest Radiol 2009;44:491-4.

10. Haacke EM, Makki M, Ge Y, Maheshwari M, Sehgal V,Hu J et al. Characterizing iron deposition in multiple scle-rosis lesions using susceptibility weighted imaging. JMagn Reson Imaging 2009;29:537-44.

11. Haacke EM, Mittal S, Wu Z, Neelavalli J, Cheng YC.Susceptibility-weighted imaging: technical aspects andclinical applications, part 1. AJNR Am J Neuroradiol2009;30:19-30.

12. Haacke EM, Xu Y, Cheng YC, Reichenbach JR. Suscep-tibility weighted imaging (SWI). Magn Reson Med2004;52:612-8.

13. Hammond KE, Lupo JM, Xu D, Metcalf M, Kelley DA,Pelletier D et al. Development of a robust method for gene-rating 7.0 T multichannel phase images of the brain withapplication to normal volunteers and patients with neu-rological diseases. Neuroimage 2008;39:1682-92.

14. Polman CH, Reingold SC, Edan G, Filippi M, HartungHP, Kappos L et al. Diagnostic criteria for multiple scle-rosis: 2005 revisions to the “McDonald Criteria”. Ann Neu-rol 2005;58:840-6.

15. Lublin FD, Reingold SC. Defining the clinical course ofmultiple sclerosis: results of an international survey. Natio-nal Multiple Sclerosis Society (USA) Advisory Commit-tee on Clinical Trials of New Agents in Multiple Sclero-sis. Neurology 1996;46:907-11.

16- Kurtzke JF. Rating neurologic impairment in multiplesclerosis: an expanded disability status scale (EDSS). Neu-rology 1983;33:1444-52.

17. Cutter GR, Baier ML, Rudick RA, Cookfair DL, FischerJS, Petkau J et al. Development of a multiple sclerosisfunctional composite as a clinical trial outcome measu-re. Brain 1999;122(Pt 5):871-82.

18. Zamboni P, Menegatti E, Weinstock-Guttman B, Schir-da C, Cox JL, Malagoni AM et al. The severity of chroniccerebrospinal venous insufficiency in patients with mul-tiple sclerosis is related to altered cerebrospinal fluiddynamics. Funct Neurol 2009;24:133-8.

19. Zivadinov R, Rudick RA, De Masi R, Nasuelli D, UkmarM, Pozzi-Mucelli RS et al. Effects of IV methylpredniso-lone on brain atrophy in relapsing-remitting MS. Neu-rology 2001;57:1239-47.

20. Smith SM, Zhang Y, Jenkinson M, Chen J, Matthews PM,Federico A et al. Accurate, robust, and automated longi-tudinal and cross-sectional brain change analysis. Neu-roimage 2002;17:479-89.

21. Zivadinov R, Weinstock-Guttman B, Benedict R, Tamaño-Blanco M, Hussein S, Abdelrahman N et al. Preservationof Gray Matter Volume in Multiple Sclerosis Patients withthe Met Allele of the rs6265 (Val66Met) SNP of Brain Deri-ved Neurotrophic Factor (BDNF). Hum Mol Genet2007;16:2659-68.

22. Sled JG, Zijdenbos AP, Evans A. C. A nonparametricmethod for automatic correction of intensity nonunifor-mity in MRI data. IEEE Trans Med Imaging 1998;17:87-97.

23. Klein A, Andersson J, Ardekani BA, Ashburner J, AvantsB, Chiang MC et al. Evaluation of 14 nonlinear defor-mation algorithms applied to human brain MRI regi-stration. Neuroimage 2009;46:786-802.

24. J Jenkinson M, Bannister P, Brady M, Smith S. Impro-ved optimization for the robust and accurate linear regi-stration and motion correction of brain images. Neuroi-mage 2002;17:825-41.

25. Sehgal V, Delproposto Z, Haacke EM, Tong KA, WycliffeN, Kido DK et al. Clinical applications of neuroimagingwith susceptibility-weighted imaging. J Magn Reson Ima-ging 2005;22:439-50.

26. Patenaude B, Smith S, Kennedy D, Jenkinson M. Impro-ved surface models for FIRST. Human Brian MappingConference 2008.

27. Smith SM. Fast robust automated brain extraction. HumBrain Mapp 2002;17:143-55.

28. Zhang Y, Brady M, Smith SM. Segmentation of brain MRimages through a hidden Markov random field modeland the expectation maximization algorithm. IEEE TransBiomed Imag 2001;20:45-57.

29. Stankiewicz J, Panter SS, Neema M, Arora A, Batt CE,Bakshi R. Iron in Chronic Brain Disorders: Imaging andNeurotherapeutic Implications. Neurotherapeutics2007;4:371-86.

30. LeVine SM. Iron deposits in multiple sclerosis and Alzhei-mer’s disease brains. Brain Res 1997;760:298-303.

31. Kalman B, Laitinen K, Komoly S. The involvement ofmitochondria in the pathogenesis of multiple sclerosis.J Neuroimmunol 2008;188:1-12.

32. Craelius W, Migdal MW, Luessenhop CP, Sugar A, Miha-lakis I. Iron deposits surrounding multiple sclerosis pla-ques. Arch Pathol Lab Med 1982;106:397-9.

33. Adams CW. Perivascular iron deposition and other vascu-lar damage in multiple sclerosis. J Neurol Neurosurg Psy-chiatry 1988;51:260-5.

34. Levine SM, Chakrabarty A. The role of iron in the patho-genesis of experimental allergic encephalomyelitis andmultiple sclerosis. Ann N Y Acad Sci 2004;1012:252-66.

35. Porto G, De Sousa M. Iron overload and immunity. WorldJ Gastroenterol 2007;13:4707-15.

36. Bakshi R, Benedict RH, Bermel RA, Caruthers SD, PuliSR, Tjoa CW et al. T2 hypointensity in the deep gray mat-ter of patients with multiple sclerosis: a quantitativemagnetic resonance imaging study. Arch Neurol2002;59:62-8.

37. Bakshi R, Dmochowski J, Shaikh ZA, Jacobs L. Gray mat-ter T2 hypointensity is related to plaques and atrophy inthe brains of multiple sclerosis patients. J Neurol Sci2001;185:19-26.

38. Ceccarelli A, Rocca MA, Neema M, Martinelli V, Arora A,Tauhid S et al. Deep gray matter T2 hypointensity is pre-sent in patients with clinically isolated syndromes sug-gestive of multiple sclerosis. Mult Scler 16:39-44.

39. Neema M, Goldberg-Zimring D, Guss ZD, Healy BC, Gutt-mann CR, Houtchens MK et al. 3 T MRI relaxometry


detects T2 prolongation in the cerebral normal-appea-ring white matter in multiple sclerosis. Neuroimage2009;46:633-41.

40. Khalil M, Enzinger C, Langkammer C, Tscherner M, Wal-lner-Blazek M, Jehna M et al. Quantitative assessment ofbrain iron by R(2)* relaxometry in patients with clini-cally isolated syndrome and relapsing-remitting multiplesclerosis. Mult Scler 2009;15:1048-54.

41. Ge Y, Jensen JH, Lu H, Helpern JA, Miles L, Inglese M etal. Quantitative assessment of iron accumulation in thedeep gray matter of multiple sclerosis by magnetic fieldcorrelation imaging. AJNR Am J Neuroradiol2007;28:1639-44.

42. Haacke EM, Ayaz M, Khan A, Manova ES, KrishnamurthyB, Gollapalli L et al. Establishing a baseline phase beha-

vior in magnetic resonance imaging to determine normalvs. abnormal iron content in the brain. J Magn ResonImaging 2007;26:256-64.

43. Zivadinov R, Bakshi R. Role of MRI in multiple sclerosisII: brain and spinal cord atrophy. Front Biosci 2004;9:647-64.

44. Tjoa CW, Benedict RH, Weinstock-Guttman B, FabianoAJ, Bakshi R. MRI T2 hypointensity of the dentate nucleusis related to ambulatory impairment in multiple sclero-sis. J Neurol Sci 2005;234:17-24.

Corresponding author: R. Zivadinov, MD, PhD, Departmentof Neurology, School of Medicine and Biomedical Sciences,Buffalo Neuroimaging Analysis Center, 100 High St., Buffa-lo, NY 14203, USA. E-mail: [email protected]

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org) /PDFXTrapped /Unknown

/Description >>> setdistillerparams> setpagedevice

chronic cerebrospinal venous insufficiency and iron …...2010/08/10 · chronic cerebrospinal...

Documents