FromRRthofMD

ThisofH(a

Patterns of chronic venous insufficiency in the duralsinuses and extracranial draining veins and theirrelationship with white matter hyperintensities forpatients with Parkinson’s diseaseManju Liu, MS,a Haibo Xu, MD, PhD,b Yuhui Wang, MD, PhD,b Yi Zhong, MS,c Shuang Xia, MD,d

David Utriainen, BS,c Tao Wang, MD,e and E. Mark Haacke, PhD,a,c,f,g Detroit, Mich; and Wuhan andTianjin, China

Background: Idiopathic Parkinson’s disease (IPD) remains one of those neurodegenerative diseases for which the causeremains unknown. Many clinically diagnosed cases of IPD are associated with cerebrovascular disease and white matterhyperintensities (WMHs). The purpose of this study was to investigate the presence of transverse sinus and extracranialvenous abnormalities in IPD patients and their relationship with brain WMHs.Methods: Twenty-three IPD patients and 23 age-matched normal controls were recruited in this study. They hadconventional neurologic magnetic resonance structural and angiographic scans and, for blood flow, quantification ofthe extracranial vessels. Venous structures were evaluated with two-dimensional time of flight; flow was evaluated withtwo-dimensional phase contrast; and WMH volume was quantified with T2-weighted fluid-attenuated inversion recovery.The IPD and normal subjects were classified by both the magnetic resonance time-of-flight and phase contrast images intofour categories: (1) complete or local missing transverse sinus and internal jugular veins on the time-of-flight images;(2) low flow in the transverse sinus and stenotic internal jugular veins; (3) reduced flow in the internal jugular veins;and (4) normal flow and no stenosis.Results: Broken into the four categories with categories 1 to 3 combined, a significant difference in the distribution of theIPD patients and normal controls (c2 [ 7.7; P < .01) was observed. Venous abnormalities (categories 1, 2, and 3) wereseen in 57% of IPD subjects and in only 30% of controls. In IPD subjects, category type correlated with both flowabnormalities and WMHs.Conclusions: From this preliminary study, we conclude that a major fraction of IPD patients appear to haveabnormal venous anatomy and flow on the left side of the brain and neck and that the flow abnormalities appear tocorrelate with WMH volume. Studies with a larger sample size are still needed to confirm these findings. (J Vasc Surg2014;-:1-10.)

Clinical Relevance: Magnetic resonance imaging is a powerful tool to study the venous structural abnormalities, toquantify blood flow, and to quantify the number and volume of brain white matter lesions. Correlations between thevenous anatomical and flow abnormalities and white matter hyperintensities in the patients with Parkinson’s disease werefound in this study. These findings may prove to be an important means to subclassify patients with idiopathicParkinson’s disease. Study of those patients with idiopathic Parkinson’s disease with venous structural and flow abnor-malities may help understand more about the etiology or progression of the disease.

Idiopathic Parkinson’s disease (IPD) is the secondmost common neurodegenerative disease after Alzheimer’sdisease, and it affects roughly 0.1% to 0.3% of the popula-tion.1 The main known risk factor is age. The etiology ofIPD remains unknown. In general, patients with Parkin-son’s disease (PD) show a loss of dopaminergic neurons

the Department of Biomedical Engineeringa and Department ofadiology,g Wayne State University, Detroit; the Department ofadiologyb and Department of Neurology,e Union Hospital, Wuhan;e Magnetic Resonance Innovations, Inc, Detroitc; the DepartmentRadiology, Tianjin First Central Hospital, Tianjind; and the

agnetic Resonance Imaging Institute for Biomedical Research,etroit.f

work was supported, in part, by National Natural Science FoundationChina (grant number: 81171386 and 30770623) and the Nationaleart, Lung, and Blood Institute of the National Institutes of Healthward number R42HL112580).

in the substantia nigra pars compacta, a reduction of dopa-mine levels in the striatum over time,2 and accumulation ofintraneuronal inclusions called Lewy bodies and Lewyneurites.3 Although PD is clinically a motor disorder andhas a good response to dopaminergic therapy, in theadvanced stages of PD, most of the motor disability

Author conflict of interest: Prof Haacke is the President of Magnetic Reso-nance Innovations, Inc.

Additional material for this article may be found online at www.jvascsurg.org.Reprint requests: E. Mark Haacke, PhD, 440 E Ferry St, Detroit, MI 48202(e-mail: nmrimaging@aol.com).

The editors and reviewers of this article have no relevant financial relationshipsto disclose per the JVS policy that requires reviewers to decline review of anymanuscript for which they may have a conflict of interest.

0741-5214/$36.00Copyright � 2014 by the Society for Vascular Surgery.http://dx.doi.org/10.1016/j.jvs.2014.02.021

1

JOURNAL OF VASCULAR SURGERY2 Liu et al --- 2014

symptoms do not respond to dopaminergic therapyanymore.3 There are also many nonmotor problems, suchas cognitive impairment, autonomic dysfunction, neuro-psychiatric symptoms,4 and fatigue,5 for patients in bothearly and advanced stages. These findings suggest thatthe dopaminergic system may not be the only systeminvolved in the PD process.

It has been shown that there is an increased presenceof white matter hyperintensity (WMH) in IPD patients.6

The suggested causes of these deficits include ischemia andvenous insufficiency from periventricular venular collageno-sis.7-9 WMH might cause or exacerbate motor or cognitivefeatures of PD,10 especially in the presence of gray mattervascular lesions involving the substantia nigra or striatum,11

corticostriatal-thalamocortical loop disruption, damage tothe interhemispheric connections of the corpus callosum,10

or disruption of the subcortical afferents. If typical late-onsetPD patients have a history of minor stroke, ischemic heartdisease, or diabetes mellitus, they show more severe clinicalfeatures.12 One group found that the severity of WMH atbaseline was the best predictor of PD progression.13

In the last few years, there has been an increasing inter-est in the role of veins in neurodegenerative diseases,14 andmore attention has been paid to the extracranial veins, suchas the internal jugular veins (IJVs) and the azygos veins, asbeing potential sources of venous hypertension.14-16

Obstructed venous outflow in the extracranial veins wasreported to correlate significantly with hypoperfusion inthe brain parenchyma, which could contribute to hypoxiaand axonal damage.17,18 In addition, venous hypertensionin the dural sinuses inhibits the absorption of cerebrospinalfluid through arachnoid villi. Some studies showed an asso-ciation between the venous outflow disturbances with lownet cerebrospinal fluid flow in patients with multiple scle-rosis (MS).19,20 An important aspect of the venous abnor-malities is that they are potentially treatable withpercutaneous transluminal angioplasty (PTA).21 The firstapplication of PTA in the major cerebral veins was by Zam-boni et al22 in 2009 in MS patients with chronic cerebro-spinal venous insufficiency (CCSVI). Since then, a fewstudies have shown improvement in neurologic outcomesand some quality of life parameters in MS patients whounderwent PTA.22-24

The use of magnetic resonance angiography and phasecontrast flow quantification for the study of the vasculaturein patients with PD is a novel concept spurred, in part, byour recent work in MS patients.25 By the application offlow encoding in the phase contrast sequence, the intensityof the phase images is directly proportional to the speed.26

The phase intensity in radians is then scaled to velocityby the relationship v ¼ phase (Venc/p), in which Vencstands for velocity encoding. By positioning these two-dimensional (2D) phase contrast magnetic resonance imag-ing (MRI) flow slices roughly perpendicular to the vesselsin the neck, flow in all major vessels can be quantified.Using these morphologic and functional MRI techniques,we can obtain not only the anatomic vascular informationbut also the quantitative arterial and venous blood flow.

To this end, we proposed this preliminary study with agroup of 23 IPD patients to see if some of these patientshave abnormal structure or flow either intracranially orextracranially. The outcomes from this study have thepotential to open new doors in studying the vascular etiol-ogy of IPD.

METHODS

Recruitment of patients and controls

From May 2011 to March 2013, 40 IPD patients wererecruited and scanned. Fifteen patients did not receive acomplete set of scans because of motion or terminationdue to the patient’s discomfort. Two IPD patients wereexcluded because they were scanned after taking medica-tion. By excluding those cases, we finally included 23IPD cases, and for this reason 23 age-matched healthy sub-jects were included in this study. They were all recruitedand imaged at Wuhan Union Hospital, China. The patientswith clinically definite IPD were diagnosed by a neurologistat Wuhan Union Hospital on the basis of the UnitedKingdom PD Society Brain Bank (UKPDSBB) criteria.The patients and controls were imaged under internalreview board-approved protocols.

Patients who fulfilled the UKPDSBB criteria wereincluded. However, patients who had any of the followingconditions were excluded from this study: any element ofthe exclusion criteria listed in UKPDSBB; other neurologicdisorders, such as Huntington’s disease, MS, and normalpressure hydrocephalus; drug-induced parkinsonism;hypoxia; arteriosclerotic disease; and hypertension or dia-betes because excessive WMH may show in those patients.The following conditions were excluded for normal con-trols: history of cardiovascular, neurologic, or psychiatricconditions; head trauma; hypertension; diabetes; anddrug or alcohol problems.

All patients and controls consented to be subjects in thisstudy. A 3T Siemens scanner with a 16-channel head/neckcoil arrangement was used to acquire the data (SiemensTRIO). Patients underwent conventional clinical imagingas well as angiographic (arterial and venous) and flow quanti-fication imaging. The imaging parameters for each sequenceare listed in the Table. The magnetic resonance images ofthe patients were acquired before medication was taken.

Data processing and analysis

Data processing was done with our in-house softwareSignal Processing in Nuclear Magnetic Resonance (SPIN,Detroit, Mich). WMHs were evaluated from the 2Dfluid-attenuated inversion recovery (FLAIR) images. Thetotal volume of WMH was calculated semiautomaticallyby SPIN. As the FLAIR data were collected with differentresolutions, all the data were normalized to 1 � 1 � 5 mm3

(transverse in-plane) before the volume quantification wasdone. In addition, the MRI visual rating scale proposedby Scheltens et al27 was also used for evaluation of theWMH level, which takes the number, size, and locationof the WMH into account. The modified visual rating

Fig 1. Phase contrast flow quantification images without (A and B) and with (C andD) vessel contours used for bloodflow quantification. A, and C, are magnitude images. B, and D, are phase images. Vessels of interest for this caseinclude internal jugular veins (IJVs; solid down arrows), external jugular veins (up chevron arrows), common carotidarteries (notched down arrows), and vertebral arteries (up solid arrows). In this case, the right IJV has circulatory flow(blood flowing toward the brain) as shown by the black arrow in the phase image.

Table. Magnetic resonance imaging (MRI) parameters for different sequences

2D TOF-MRVT2-FLAIR

(21 PD patients)T2-FLAIR

(2 PD patients, 23 normal controls) PCFQ at C6/C7

Orientation Transverse Transverse Sagittal TransverseTR, ms 26 8500 6000 42.25TE, ms 5.02 93 396 4.13Flip angle 60� 130� 120� 25�Resolution, mm3 0.5 � 0.5 � 2.5 0.43 � 0.43 � 5 0.5 � 0.5 � 1 0.57 � 0.57 � 4Bandwidth, Hz/pixel 217 287 781 531

An arterial saturation band with a width of 40 mm and a separation of 10 mm from the excited slice was applied during two-dimensional time-of-flightmagnetic resonance venography (2D TOF-MRV) acquisition. A maximum encoding velocity (Venc) of 50 cm/s was used for phase contrast flowquantification (PCFQ). FLAIR, Fluid-attenuated inversion recovery; PD, Parkinson’s disease; TE, echo time; TR, repetition time.

JOURNAL OF VASCULAR SURGERYVolume -, Number - Liu et al 3

criteria are presented in the Appendix (online only). The2D time-of-flight (TOF) data were used to evaluate thevenous structures in the head and neck. A saturationband was applied to suppress the arterial flow in the 2DTOF images. The maximum intensity projection of thewhole series was generated in the coronal view from the2D TOF coverage. The major veins of interest for thestructural analysis included the transverse sinuses and theextracranial veins in the neck.

The phase contrast flow quantification images wereused to analyze the through-plane blood flow in the lowerneck (C6/C7). Thirty time points were collected for eachcardiac cycle. Cardiac gating was achieved by pulse

triggering. Background muscle was used to monitor thebaseline to ensure that there was no bulk drift over time.We chose 50 cm/s as the Venc because this gave muchbetter signal-to-noise ratio for venous flow than100 cm/s, and flow in the arteries could still be unwrappedwhen aliasing occurred, which was done automatically withSPIN. The vessels of interest (Fig 1) included the internaland external jugular veins, vertebral veins, deep cervicalveins, common carotid arteries, and vertebral arteries.The vessel lumen was segmented from the magnitude im-ages (Fig 1, A and C) semiautomatically owing to the pres-ence of sufficiently high contrast and copied onto thephase images (Fig 1, B and D).

JOURNAL OF VASCULAR SURGERY4 Liu et al --- 2014

The magnetic resonance flow technique we use gives apixel-by-pixel flow for each vessel. From each pixel we getflow as a function of the cardiac cycle. From these data, forall vessels, we can obtain integrated flow over the entirecardiac cycle (mL), volume flow rate (mL/s), and bothpositive and negative volume flow rates (mL/s). The pos-itive volume flow rate in the IJVs was used to establish thecriteria for circulatory flow. Circulatory flow represents thesituation in which there is simultaneously positive andnegative flow present within a vein. On the phase contrastimages, the vein with circulatory flow shows dark andbright signal areas (Fig 1, D), and on the basis of the pos-itive volume flow rate plot, the vein has at least four timepoints with positive flow higher than 2 mL/s and positiveflow (toward the brain) present for at least half of the car-diac circle. Example plots of the key parameters over a fullcardiac cycle are shown in Fig 2. Fig 2, A(3) shows the pos-itive flow plot for an example with circulatory flow. Thevenous flow was normalized by the total arterial flow(tA), which was defined as the sum of the flow in the com-mon carotid arteries and vertebral arteries. In addition,the ratio between the dominant and subdominant IJVs(Fd/Fsd) was calculated. Dominant IJV represents theIJV that carries the most outflow, and the subdominantIJV is the one with less outflow.

On the basis of the anatomic information from the 2DTOF and the quantitative IJV flow, the IPD patients andnormal controls were separated into four categories. Exam-ples from each of these IJV abnormalities are shown inFig 3, A-D. In defining the category types from a venousanatomy and flow perspective, flow thresholds and crosssection thresholds were defined. An Fd/Fsd ratio greaterthan 4 was taken to be abnormal. We determined theFd/Fsd threshold based on the receiver operating character-istic (ROC) curve analysis of the patients and controls, and4 represented the optimal cutoff to distinguish between thetwo populations. The work of Seoane et al28 suggested afactor of 3:1 for dominant to subdominant transverse sinusflow as a risk factor. IJVs are a continuation of the trans-verse sinus, so our result is not inconsistent with their find-ings. The threshold of normalized subdominant IJV flow(Fsd/tA) was calculated from the normal controls by sub-tracting the standard deviation from the mean (0.24 �0.10 ¼ 0.14). An IJV was called stenotic if the cross-sectional area was less than 12.5 mm2 for the upper neckor less than 25 mm2 for the lower neck.25 On the basisof these quantitative data and our general observations,we considered the following four categories as representa-tive of the data.

Category 1. All the following conditions must be met:(1) one or both transverse sinuses do not appear on theTOF; (2) one or both sigmoid sinuses are not shown onthe TOF; (3) absence or local absence of IJVs on theTOF; (4) Fd/Fsd at C6/C7 is greater than 4 and Fsd/tAis less than 0.14, or circulatory flow in one or both of theIJVs. In the example case shown in Fig 3, A(2), there isno left transverse sinus and no left sigmoid sinus on TOF.The upper left IJV is severely stenotic (arrow) and the top

part is absent, and some other small veins reconstitute theleft IJV at the lower level. In this case, Fd/Fsd ¼ 13.85.

Category 2. All the following conditions must be met:(1) one or both transverse sinuses do not appear on theTOF; (2) sigmoid sinuses appear on the TOF; (3) thereis banding in the IJVs (this represents slow or reflux bloodflow) or the presence of stenosis along the IJVs; (4) same ascategory 1, criterion 4. In the examples shown in Fig 3,B(1), the left transverse sinus is missing or invisible,although the left sigmoid sinus is present. A banding arti-fact is evident in almost the entire left IJV, which may becaused by an oscillatory or slow flow. The left sigmoid sinusand the left IJV are much narrower than the right. Thepresence of a transverse sinus cannot be ruled out withoutthe use of a contrast agent, but there may be reduced flowthat makes it difficult to visualize with 2D TOF. Never-theless, this slow flow may still be manifested in thereduced flow visualized here. In this case, Fd/Fsd ¼ 21.1.

Category 3. All the following conditions must be met:(1) sigmoid sinuses appear on the TOF; (2) same as cate-gory 1, criterion 4. In the example case shown in Fig 3,C, part of the left transverse sinus appears to be missingbecause of slow flow or just in-plane flow, the left sigmoidsinus is visible, and Fd/Fsd ¼ 4.39.

Category 4. All the following conditions must be met:(1) the transverse sinuses and sigmoid sinuses appear onthe TOF; (2) the IJVs are not stenotic; and (3) Fd/Fsd atC6/C7 is less than 4 and Fsd/tA is more than 0.14. TheFd/Fsd of the example shown in Fig 3, D is 2.1.

Except for category 4, all the other three categoriescontain some structural or flow abnormalities. Categories1, 2, and 3 are ranked on the basis of the severity of thevascular structural abnormalities,with the abnormality incategory 1 being the most severe.

Statistical analysis

The optimal cutoff values for Fd/Fsd, WMH volume,and WMH visual score to separate IPD patients and con-trols were determined by ROC analysis. An unpaired ttest was used to test the difference between the two popu-lations for age, tA, common carotid arterial flow, vertebralarterial flow, normalized total IJV flow (tIJV/tA), normal-ized left IJV flow (LIJV/tA), Fd/Fsd, normalized collateralvenous flow, and WMH visual score. A paired t test wasapplied to test the difference between left and right IJVs.The categorization distribution difference between IPDpatients and the normal controls was analyzed by c2 test.A Wilcoxon rank sum test was used to test the WMH vol-ume difference between patients and controls because ofoutliers.

RESULTS

The age range of the IPD patients was 43 to 74 years,with a mean age of 62.6 years and standard deviation of8.3 years. For the normal subjects, the age range was 56to 81 years, and the age mean was 64.1 years with a stan-dard deviation of 6.7 years. There is no statistical age differ-ence between the two populations (P ¼ .48).

Fig 2. A, Plots of the integrated flow (1), volume flow rate (2), and positive volume flow rate (3) of the vessels ofinterest over a full cardiac cycle at C6/C7 for the same case shown in Fig 1. B, The same types of plots as shown in (A)but for a case without circulatory flow. Note that in comparing the positive volume flow rate between (A and B), thepositive flow in the right internal jugular vein (IJV) with circulatory flow (arrows in A(3)) is much higher than in theIJV without circulatory flow. CCA, Common carotid artery; DCV, deep cervical vein; EJV, external jugular vein;NFA, no flow area; VA, vertebral artery; VV, vertebral vein.

JOURNAL OF VASCULAR SURGERYVolume -, Number - Liu et al 5

Fig 3. A-D, Example cases from each of four categories and different abnormal internal jugular vein (IJV) conditionsshown in the coronal projection of the two-dimensional time-of-flight (2D TOF) images. A, Two cases from category1; A(2), Fd/Fsd ¼ 13.85. B, Two cases from category 2; B(1), Fd/Fsd ¼ 21.1. C, A category 3 case, Fd/Fsd ¼ 4.39.D,A category 4 case, Fd/Fsd ¼ 2.1. A(1): missing IJV. A(2), Stenotic left IJV (arrow in A(2)). B(1), Banding artifact inthe left IJV. The right IJV is big and the signal is uniform, so the banding artifact in the left IJV is caused by abnormalflow, not by swallowing or respiratory artifact. B(2), Uneven signal contrast in the left IJV. Compared with the rightIJV, the signal in the left IJV is not uniform and has a stenosis (arrow in B(2)).

JOURNAL OF VASCULAR SURGERY6 Liu et al --- 2014

Category analysis. The categorization distribution ofthe IPD patients along with the normal subjects by useof the criteria defined before is shown in Fig 4, A. Onthe basis of the venous anatomic characteristics and asym-metric flow in the IJVs, there were clearly four categorieswith different severity of venous vasculature. However,with a small number of subjects in some categories, thedistribution comparison between the patients and controlsmay not be applicable. We summed up the subject numberin categories 1 to 3 that represent the ones with bothanatomic and flow abnormalities and then did the

comparison. For normal controls, 30% were in categories 1to 3 and 70% in category 4; for the PD patients, 57% werein categories 1 to 3 and 43% belonged to category 4.According to the c2 analysis, these two populations aresignificantly different (c2 ¼ 7.70; P < .01).

Quantitative flow analysis. For the IPD patients andnormal subjects, the tA was 14.67 6 2.49 mL/s and16.25 6 2.1 mL/s, respectively, and there was significantdifference between them (P ¼ .04). The average flow inthe common carotid arteries was 5.98 6 1.26 mL/s inIPD patients and 6.51 6 1.12 mL/s in controls. The

Fig 4. A, Distribution of the idiopathic Parkinson’s disease (IPD) patients and normal controls according to thedefined categories. The normal population lies predominantly in category 4. Most IPD patients are distributed incategories 1 to 3, which have venous structural or flow abnormalities. PD, Parkinson’s disease. B, Scatter plot ofnormalized dominant internal jugular vein (IJV) vs subdominant IJV flow at C6/C7 neck level in IPD patients (circle)and normal controls (cross). The higher line represents the Fd/Fsd ¼ 4:1. In comparison to the normal subjects (5 of 23cases lie below the 4:1 ratio line), a higher percentage of the IPD patients (11 of 23 cases; 48%) showed Fd/Fsd greaterthan 4:1. Moreover, all the patients with stenosis in the IJV (solid circles) lie below the line, indicating that they havemore severe asymmetric flow in the IJV. The lower line represents Fd/Fsd ¼ 10:1. In this case, all but one stenotic PDcase fall below the 10:1 line, and only one normal falls below this line. ST, Stenotic;NST, nonstenotic. C, Scatterplot ofcategory vs Fd/Fsd for the 23 patients. They have good correlation (R2 ¼ 0.72), which indicates that patients withhigher Fd/Fsd tend to be in the category with more severe venous structural and flow problems.

JOURNAL OF VASCULAR SURGERYVolume -, Number - Liu et al 7

blood flow in the common carotid artery was significantlylarger (P ¼ .04) in the normal people than in the IPD pa-tients. The average flow in the vertebral arteries was 1.3960.65 mL/s and 1.57 6 0.76 mL/s in the IPD patients andnormal subjects, respectively, and there was no significantdifference between them. The tIJV and tIJV/tA for thenormal controls were statistically higher than for the pa-tients (IPD: 10.85 6 3.29 mL/s, normal controls:13.54 6 2.14 mL/s, P < .01; IPD: 0.74 6 0.17, normalcontrols: 0.84 6 0.14, P ¼ .03).

The normalized flow of the right IJV was found to besignificantly larger than that of the left IJV for both IPDpatients (0.56 6 0.16 vs 0.18 6 0.16; P < .0001) andthe normal population (0.52 6 0.2 vs 0.32 6 0.19; P ¼.01). The LIJV/tA was significantly higher in the normalcontrols (IPD, 0.18 6 0.16; normal controls, 0.32 60.19; P ¼ .01). Another measure of flow variation betweenthe left and right IJVs is the ratio of the two IJVs, Fd/Fsd.

Because Fsd can become quite small in some cases, werestricted the ratio to be 10:1 in performing the averagevalues so as not to skew the data (this ratio was set to10:1 for eight PD patients and only one normal control).This ratio is significantly larger in IPD patients than innormal controls (P ¼ .04). Fig 4, B shows the scatterplotof normalized dominant vs subdominant IJV flow for thenormal controls and IPD subjects.

According to the four vascular categories we defined,category 1 and category 2 contain the conditions of IJV ste-nosis or missing IJVs, so we refer to patients in categories 1and 2 as stenotic patients and patients in categories 3 and 4as nonstenotic patients. According to Fig 4, B, the stenoticpatients showed much higher Fd/Fsd than the nonstenoticpatients. For the tIJV and tA, there is no significant differ-ence found between the stenotic and nonstenotic patients(tIJV: 10.2 6 4.4 mL/s vs 11.3 6 2.4 mL/s, P ¼ .52;tA: 14.7 6 4.4 mL/s vs 14.6 6 1.7 mL/s, P ¼ .96).

Fig 5. A, The distribution of idiopathic Parkinson’s disease (IPD) patients and normal controls with low white matterhyperintensity (WMH) volume (<500 mm3) and high WMH volume ($500 mm3) for different categories. Themajority of category 4 patients and normal controls contain low WMH volume, whereas more of the patients incategories 1 to 3 show high WMH volume. B, The distribution of IPD patients and normal controls with low WMHvisual score (#3) and high WMH visual score ($4) for different categories. All the patients in category 1 have highWMH visual scores. For patients in categories 2 and 3, more of the patients show high WMH scores.

JOURNAL OF VASCULAR SURGERY8 Liu et al --- 2014

The category type appears to correlate well with Fd/Fsd(Fig 4, C), which indicates that the higher the Fd/Fsd ratio,the more likely the patient sits in the severe abnormal struc-tural and flow categories.

Correlation between FLAIR WMH and categori-zation. Of the 23 IPD cases, 21 showed WMH. Themean, median, and standard deviation of WMH volumein IPD patients were 5827 mm3, 502 mm3, and10,755 mm3, with a range of 0 mm3 to 42,177 mm3.The mean, median, and standard deviation of the WMHvolume in the normal controls were 776 mm3, 265 mm3,and 1358 mm3, with range of 0 mm3 to 6147 mm3. Thereis a significant difference between the two populations ac-cording to the Wilcoxon rank sum test (P ¼ .04). TheIPD patients and normal controls were further dividedinto a high WMH volume group and a low WMH volumegroup. The threshold for this separation was set to500 mm3, which was determined by an ROC analysis be-tween the normal controls and PD patients. The distribu-tion of the patients and normal controls with high andlow WMH volume along with the category type is shownin Fig 5, A.

The WMH visual score for the IPD patients was 5.8 65.3, whereas the score of the age-matched normal controlswas 3.1 6 3.0. There is significant difference between thetwo populations for the WMH visual rating (P ¼ .05).Moreover, all the patients in category 1 showed a WMHvisual score $4. The distribution of patients with lowWMH score (#3) and high WMH score ($4) accordingto their categorization is shown in Fig 5, B. We chose ascore of 3 as the threshold because it could best separatethe patients’ categorization distribution.

DISCUSSION

This paper is the first to study venous abnormalities ofpatients with IPD. The categorization defined in this workcould help distinguish IPD patients from normal controls:

70% of the normal controls appeared in category 4,whereas 57% of the IPD patients belonged to categories1 to 3 (those containing venous structural or venous flowabnormalities). In addition, those patients with higherWMH scores tended to be in categories 1 to 3, which in-dicates a likely correlation between the vascular abnormal-ities and brain WMH. Moreover, three cases with Fd/Fsd ofless than 4 and a WMH score higher than 4 showed circu-latory flow in at least one of the IJVs. The combination ofstructural abnormalities, asymmetric IJV flow, andabnormal hemodynamics such as circulatory flow couldtogether be a major risk factor for development of WMHand perhaps even the disease itself.

Not all PD patients showed these flow abnormalities.However, PD is not a disease but a syndrome, and theremay be many sources for its development. In this light,one may consider CCSVI to be just a comorbidity associ-ated with PD. As for the normal controls, there is definitelyan overlap with categories 1 to 3. If one used this to deter-mine who might get PD or neurodegenerative disease,normal controls seen in these three categories would beincorrectly assessed; but we have no way of telling todaywhere they will sit in the future.

Our finding of the asymmetric pattern of the IJVflow is consistent with many other research publicationsbecause a majority of individuals are right dominant.29

However, the flow ratio between the two IJVs (Fd/Fsd)of IPD patients is significantly higher than that of thenormal controls. Moreover, patients with vascular stenosishave significantly higher Fd/Fsd than the nonstenoticpatients.

Recently, more attention has been paid to venous col-laterals.30,31 By development of a model to calculate thevenous collateral flow, which is normalized to the tA, acollateral flow index has been compared between normalcontrols and MS patients with CCSVI. In the supine posi-tion, the patient’s collateral flow index was reported to be

JOURNAL OF VASCULAR SURGERYVolume -, Number - Liu et al 9

significantly higher than that of the normal controls (61%;P < .0002).30 As our data were also collected in the supineposition with quantitative flow in the common carotids,vertebral arteries, IJVs, and vertebral veins, a similar typeof comparison can be made between the IPD patientsand the normal controls. In fact, we too found that thenormalized collateral flow in the IPD patients was statisti-cally higher than in the normal controls (0.26 6 0.17 vs0.16 6 0.14; P ¼ .03). One might well ask, if the flowin the stenotic IJV or transverse sinus can be recovered,could PTA be a potential therapy for some of thesepatients?32,33

The original criteria established by Zamboni et al34 arenot directly observable by the methods that we haveemployed. Whereas the original five CCSVI criteria mainlyconcern flow and stenosis patterns at the lower neck level inthe IJV and vertebral veins as well as in the azygos vein, theobservations that we have made constitute new criteria thathave been shown only in MRI thus far.29 Whether similarfindings could be made on ultrasound is unclear and needsfurther study; however, given that a lack of venous outflowand structural abnormalities have been observed with MRIin the transverse sinus and IJV in our group of PD subjects,the authors expect that a similar observation would bepossible with ultrasound.

There are several different MRI visual rating scales forWMHs.27 We chose Scheltens et al for this study because itis commonly used.35 In addition, according to the study ofPantoni et al,36 the four MRI rating scales referenced intheir paper are well correlated. Although having WMH isnot a standard clinical diagnostic criterion for IPD, in thisstudy cohort, 21 patients showed WMH; five had veryhigh WMH volume (>10,000 mm3), and of these, fourhad a high visual score (>9). A study by Piccini et al6 founda positive relationship between the WMH volume and PDsymptom severity.

According to many in vivo imaging studies, 30% to55% of PD patients show WMHs,37 and some studieshave suggested that WMHs are more frequently presentin patients with PD than in normal elderly individualsand hypertension patients.6,38 Our finding that IPD pa-tients have higher WMH volume and visual score thanthe age-matched normal controls is consistent with thosestudies. WMHs may represent either lacunar infarcts orother tissue damage in the thalamocortical motor system,a key area in the pathogenesis of parkinsonian symptoms.2

WMHs are thought to relate to cardiovascular disease.8

Young et al9 reported the association between WMH andloss of vascular integrity in a neuropathologic study, whichsuggested a vascular origin for these lesions. An MRI-pathologic correlation study by Gao et al7 demonstratedthat focal and periventricular hyperintensities often relateto venules and can increase or decrease over time. Theyproposed that venous collagenosis dilates the veins, makingthem macroscopic and causing venous insufficiency withconsequent vessel leakage and vasogenic edema.7 Thesedata support our findings, whereby patients with venousstructural or venous flow problems are more likely to

have more WMHs. Other factors could affect the WMHscores besides venous abnormalities. However, it isbelieved that WMH, especially in aging, can be caused byreductions in perfusion to the brain.39

Limitations and future directions. There are a num-ber of limitations to this study. First, the sample size issmall. Nevertheless, even with this small number, wefound clear abnormal venous vascular characteristics inIPD patients with significant differences from the normalsubjects. This suggests that there is a prevalence of vascularabnormalities in patients with IPD. With a larger numberof patients, it would be possible to study the correlationbetween the vascular problems with IPD subtypes. Thesecond limitation is the lack of three-dimensional time-resolved contrast-enhanced arteriovenography scans,which could offer the advantage of acquiring separatearterial and venous phases and better delineate the duralsinuses. Even though 2D TOF causes reduced signal fromin-plane flow, for patients with fast enough flow in thetransverse sinuses (such as is seen in many cases on theright-hand side), we still expect some signal representingthe anatomy of the vessels. Third, collecting flow in thetransverse sinuses would also be useful to understand thefluid dynamics of the evidently abnormal venous flow onthe left side.

CONCLUSIONS

In this paper, we have shown several key findings.First, WMH correlates with venous jugular flow. Second,there are venous vascular abnormalities in patients withIPD that tend not to be present in the normal popula-tion. More specifically, we found that 90% of the patientsshowed WMHs and that 57% of the patients had struc-tural or venous flow abnormalities in the transverse sinus,sigmoid sinus, and IJVs. Nevertheless, this is a prelimi-nary work, and further studies are required to confirmthese conclusions. These findings may prove to be animportant imaging means to subclassify IPD patientsand to enhance our understanding of the etiology ofIPD and perhaps even lead to the development of newtreatment regimens.

We would like to thank Jing Jiang, Ying Wang, andWei Feng for their software development supports; YanweiMiao and Jie Yang for the data analysis discussion;Khashayar Dashtipour for his valuable comments on themanuscript; and Hong Li, Chenjun Zhu, and Xin Sun forhelping collect MRI data and data processing.

AUTHOR CONTRIBUTIONS

Conception and design: ML, HX, MHAnalysis and interpretation: ML, YZ, SX, DU, MHData collection: HX, YW, TW, MHWriting the article: ML, MHCritical revision of the article: ML, HX, YW, YZ, SX, DU,

TW, MHFinal approval of the article: ML, HX, YW, YZ, SX, DU,

TW, MH

JOURNAL OF VASCULAR SURGERY10 Liu et al --- 2014

Statistical analysis: ML, MHObtained funding: HX, MHOverall responsibility: ML, MHML and HX contributed equally to this article and share

co-first authorship.

REFERENCES

1. Shulman JM, De Jager PL. Evidence for a common pathway linkingneurodegenerative diseases. Nat Genet 2009;41:1261-2.

2. Balash Y, Korczyn AD. Vascular parkinsonism. Handb Clin Neurol2007;84:417-25.

3. Calabresi P, Castrioto A, Di Filippo M, Picconi B. New experimentaland clinical links between the hippocampus and the dopaminergicsystem in Parkinson’s disease. Lancet Neurol 2013;12:811-21.

4. Poletti M, Emre M, Bonuccelli U. Mild cognitive impairment andcognitive reserve in Parkinson’s disease. Parkinsonism Relat Disord2011;17:579-86.

5. Pavese N, Metta V, Bose SK, Chaudhuri KR, Brooks DJ. Fatigue inParkinson’s disease is linked to striatal and limbic serotonergicdysfunction. Brain 2010;133:3434-43.

6. Piccini P, Pavese N, Canapicchi R, Paoli C, Del Dotto P, Puglioli M,et al. White matter hyperintensities in Parkinson’s disease. Clinicalcorrelations. Arch Neurol 1995;52:191-4.

7. Gao F, van Gaal S, Levy-Cooperman N, Ramirez J, Scott CJM. Doesvariable progression of incidental white matter hyperintensities inAlzheimer’s disease relate to venous insufficiency? Alzheimers De-mentia 2008;4:T368-9.

8. Pantoni L, Garcia JH. Pathogenesis of leukoaraiosis: a review. Stroke1997;28:652-9.

9. Young VG, Halliday GM, Kril JJ. Neuropathologic correlates of whitematter hyperintensities. Neurology 2008;71:804-11.

10. Bohnen NI, Albin RL. White matter lesions in Parkinson disease. NatRev Neurol 2011;7:229-36.

11. Thanvi B, Lo N, Robinson T. Vascular parkinsonismdan importantcause of parkinsonism in older people. Age Ageing 2005;34:114-9.

12. Papapetropoulos S, Ellul J, Argyriou AA, Talelli P, Chroni E,Papapetropoulos T. The effect of vascular disease on late onset Par-kinson’s disease. Eur J Neurol 2004;11:231-5.

13. Burton EJ, McKeith IG, Burn DJ, Firbank MJ, O’Brien JT. Progres-sion of white matter hyperintensities in Alzheimer disease, dementiawith Lewy bodies, and Parkinson disease dementia: a comparison withnormal aging. Am J Geriatr Psychiatry 2006;14:842-9.

14. Zamboni P, Galeotti R. The chronic cerebrospinal venous insufficiencysyndrome. Phlebology 2010;25:269-79.

15. Haacke EM. Chronic cerebral spinal venous insufficiency in multiplesclerosis. Expert Rev Neurother 2011;11:5-9.

16. Haacke EM, Beggs CB, Habib C. The role of venous abnormalities inneurological disease. Rev Recent Clin Trials 2012;7:100-16.

17. Zamboni P, Menegatti E, Weinstock-Guttman B, Dwyer MG,Schirda CV, Malagoni AM, et al. Hypoperfusion of brain parenchyma isassociated with the severity of chronic cerebrospinal venous insuffi-ciency in patients with multiple sclerosis: a cross-sectional preliminaryreport. BMC Med 2011;9:22.

18. Utriainen D, Trifan G, Sethi S, Elias S, Hewett J, Feng W, et al.Magnetic resonance imaging signatures of vascular pathology in mul-tiple sclerosis. Neurol Res 2012;34:780-92.

19. Zamboni P, Menegatti E, Weinstock-Guttman B, Schirda C, Cox JL,Malagoni AM, et al. The severity of chronic cerebrospinal venousinsufficiency in patients with multiple sclerosis is related to altered ce-rebrospinal fluid dynamics. Funct Neurol 2009;24:133-8.

20. Beggs CB. Venous hemodynamics in neurological disorders: ananalytical review with hydrodynamic analysis. BMC Med 2013;11:142.

21. Zivadinov R, Magnano C, Galeotti R, Schirda C, Menegatti E,Weinstock-Guttman B, et al. Changes of cine cerebrospinal fluid dy-namics in patients with multiple sclerosis treated with percutaneoustransluminal angioplasty: a case-control study. J Vasc Interv Radiol2013;24:829-38.

22. Zamboni P, Galeotti R, Menegatti E, Malagoni AM, Gianesini S,Bartolomei I, et al. A prospective open-label study of endovasculartreatment of chronic cerebrospinal venous insufficiency. J Vasc Surg2009;50:1348-58. e1-3.

23. Kipshidze N, Rukhadze I, Archvadze A, Kipiani V, Kipshidze N,Lapiashvili E, et al. Endovascular treatment of patients with chroniccerebrospinal venous insufficiency and multiple sclerosis. GeorgianMed News 2011;199:29-34.

24. Kostecki J, Zaniewski M, Ziaja K, Urbanek T, Kuczmik W,Krzystanek E, et al. An endovascular treatment of chronic cerebro-spinal venous insufficiency in multiple sclerosis patientsd6 monthfollow-up results. Neuro Endocrinol Lett 2011;32:557-62.

25. Haacke EM, Feng W, Utriainen D, Trifan G, Wu Z, Latif Z, et al.Patients with multiple sclerosis with structural venous abnormalities onMR imaging exhibit an abnormal flow distribution of the internal ju-gular veins. J Vasc Interv Radiol 2012;23:60-8. e1-3.

26. Haacke EM, Brown RW, Tompson MR, Venkatesan R. MR angiog-raphy and flow quantification. In: Haacke EM, Brown RW,Tompson MR, Venkatesan R, editors. Magnetic resonance imagingdphysical principles and sequence design. New York: Wiley-Liss; 1999.

27. Scheltens P, Barkhof F, Leys D, Pruvo JP, Nauta JJ, Vermersch P, et al.A semiquantative rating scale for the assessment of signal hyper-intensities on magnetic resonance imaging. J Neurol Sci 1993;114:7-12.

28. Seoane E, Rhoton AL Jr. Compression of the internal jugular vein bythe transverse process of the atlas as the cause of cerebellar hemorrhageafter supratentorial craniotomy. Surg Neurol 1999;51:500-5.

29. Feng W, Utriainen D, Trifan G, Sethi S, Hubbard D, Haacke EM.Quantitative flow measurements in the internal jugular veins of mul-tiple sclerosis patients using magnetic resonance imaging. Rev RecentClin Trials 2012;7:117-26.

30. Zamboni P, Sisini F, Menegatti E, Taibi A, Malagoni AM, Morovic S,et al. An ultrasound model to calculate the brain blood outflowthrough collateral vessels: a pilot study. BMC Neurol 2013;13:81.

31. Zamboni P, Consorti G, Galeotti R, Gianesini S, Menegatti E,Tacconi G, et al. Venous collateral circulation of the extracranial ce-rebrospinal outflow routes. Curr Neurovasc Res 2009;6:204-12.

32. Huang P, Yang YH, Lin WC, Chen CH, Lin RT. Successful treatmentof cerebral venous thrombosis associated with bilateral internal jugularvein stenosis using direct thrombolysis and stenting: a case report.Kaohsiung J Med Sci 2005;21:527-31.

33. Formaglio M, Catenoix H, Tahon F, Mauguiere F, Vighetto A,Turjman F. Stenting of a cerebral venous thrombosis. J Neuroradiol2010;37:182-4.

34. Zamboni P, Galeotti R, Menegatti E, Malagoni AM, Tacconi G,Dall’Ara S, et al. Chronic cerebrospinal venous insufficiency in patientswith multiple sclerosis. J Neurol Neurosurg Psychiatry 2009;80:392-9.

35. Beyer MK, Aarsland D, Greve OJ, Larsen JP. Visual rating of whitematter hyperintensities in Parkinson’s disease. Mov Disord 2006;21:223-9.

36. Pantoni L, Simoni M, Pracucci G, Schmidt R, Barkhof F, Inzitari D.Visual rating scales for age-related white matter changes (leukoaraiosis):can the heterogeneity be reduced? Stroke 2002;33:2827-33.

37. Lee SJ, Kim JS, Lee KS, An JY, Kim W, Kim YI, et al. The severity ofleukoaraiosis correlates with the clinical phenotype of Parkinson’s dis-ease. Arch Gerontol Geriatr 2009;49:255-9.

38. Stern MB, Braffman BH, Skolnick BE, Hurtig HI, Grossman RI.Magnetic resonance imaging in Parkinson’s disease and parkinsoniansyndromes. Neurology 1989;39:1524-6.

39. Alosco ML, Brickman AM, Spitznagel MB, Garcia SL, Narkhede A,Griffith EY, et al. Cerebral perfusion is associated with white matterhyperintensities in older adults with heart failure. Congest Heart Fail2013;19:E29-34.

Submitted Aug 12, 2013; accepted Feb 13, 2014.

Additional material for this article may be found onlineat www.jvascsurg.org.

JOURNAL OF VASCULAR SURGERYVolume -, Number - Liu et al 10.e1

APPENDIX (online only).

Modified visual rating scale for white matter hyper-intensities proposed by Scheltens et al27

White matter hyperintensities in periventricular and deepwhitematter are scored separately, as are the scores fordifferentlobes (the basal ganglia are also included in this assessment).The maximum score is 6 for periventricular hyperintensitiesand 30 for deep white matter hyperintensities as outlinedhere. Note the five separate lobe regions: frontal lobe, parietallobe, occipital lobe, temporal lobe, and basal ganglia.

Periventricular hyperintensities (minimum, 0;maximum 6)

d No caps in the occipital lobe d 0d Caps in the occipital lobe #5 mm d 1d Caps in the occipital lobe $6 mm and <10 mm d 2d No caps in the frontal lobe d 0d Caps in the frontal lobe #5 mm d 1d Caps in the frontal lobe $6 mm and <10 mm d 2d No bands at the lateral ventricles d 0

d Bands at the lateral ventricles #5 mm d 1d Bands at the lateral ventricles $6 mm and<10 mm d 2

The same criteria are applied for the frontal lobe,parietal lobe, occipital lobe, temporal lobe, and basalganglia.

Deep white matter hyperintensities (minimum, 0;maximum, 30)

d No lesion in the region being studied d 0d Number of the lesions #5, size of the lesions#3 mm d 1

d Number of the lesions $6, size of the lesions#3 mm d 2

d Number of the lesions #5, size of the lesions $4 mmand #10 mm d 3

d Number of the lesions $6, size of the lesions $4 mmand #10 mm d 4

d At least one lesion >11 mm d 5d Lesions are confluent d 6