18ournal ofNeurology, Neurosurgery, and Psychiatry 1994;57:1458-1465

White matter magnetic resonance hyperintensitiesin dementia of the Alzheimer type: morphologicaland regional cerebral blood flow correlates

Gunhild Waldemar, Pernille Christiansen, Henrik B W Larsson, Peter H0gh,Henning Laursen, Niels A Lassen, Olaf B Paulson

Department ofNeurology, UniversityHospital,Rigshospitalet,Copenhagen,DenmarkG WaldemarP Hogh0 B Paulson

Department ofClinical Physiology,Bispebjerg Hospital,Copenhagen,DenmarkN A LassenMagnetic ResonanceResearch Center,Hvidovre Hospital,Copenhagen,DenmarkP ChristiansenH BW LarssonInstitute ofNeuropathology,University ofCopenhagen,Copenhagen,DenmarkH LaursenCorrespondence to:Dr Gunhild Waldemar,Department of Neurology,Rigshospitalet, 9,Blegdamsvej, DK-2100Copenhagen, Denmark.

Received 21 February 1994and in revised form31 May 1994.Accepted 8 June 1994

AbstractIn a prospective MRI study the presence,appearance, volume, and regional cere-bral blood flow (rCBF) correlates ofperiventricular hyperintensities (PVHs)and deep white matter hyperintensities(DWMHs) were examined in 18 patientswith probable Alzheimer's disease and in10 age matched healthy control subjects,all without major cerebrovascular riskfactors. The 133Xe inhalation method andthe [9"-Tc]-d,l-hexamethyl-propylene-amine-oxime (HMPAO) technique withsingle photon emission computed tomog-raphy (SPECT) were used to measurerCBF. Rating scores for PVHs were sig-nificantly higher in the Alzheimer's dis-ease group (p < 0.01) and correlatedsignificantly with the volume ofventricles(p < 0.05) and with systolic arterial bloodpressure (p < 0.01), but not with rCBF.By contrast, there was no significant dif-ference in the rating scores or volumes ofDWMHs between the two groups,although three patients had extensiveDWMH lesions in the central white mat-ter. In the group of patients withAlzheimer's disease as a whole, thevolume of DWMHs correlated well withrCBF in the hippocampal region (r =-0-72; p < 0'001), but not with frontal,temporal, parietal, or occipital rCBF.Postmortem histopathology of extensiveDWMH lesions in one patient with defi-nite Alzheimer's disease showed a partialloss of myelin and astrocytic gliosis, butno ischaemic changes. It is concludedthat DWMH lesions may be associatedwith reduced rCBF in the hippocampalregion. The heterogenous topography ofneocortical rCBF deficits in Alzheimer'sdisease could not be explained by deaf-ferentation from underlying white matterhyperintensities and therefore mayreflect variations in the topography ofcortical abnormalities.

(y Neurol Neurosurg Psychiatry 1994;57:1458-1465)

Although the hallmarks of Alzheimer's diseaseare the plaques and tangles seen in corticalbrain tissue on neuropathological examina-tion, Alzheimer's disease is not just a corticalbrain disease. Structural involvement of thewhite matter was suggested in a neuropatho-logical study by Brun and Englund,' whoshowed symmetric deep ischaemic white mat-

ter changes in most of their patients with adefinite diagnosis of Alzheimer's disease.Clinical studies measuring the occurrence ofwhite matter abnormalities on MRI inpatients with Alzheimer's disease, however,have yielded conflicting results.2-7

Recently, we described a pronounced topo-graphical heterogeneity of regional cerebralblood flow (rCBF) patterns measured withhigh resolution single photon emission com-puted tomography (SPECT) and [99mTc]-d, l-hexamethyl-propylene-amine-oxime(HMPAO) in patients with a clinical diagno-sis of Alzheimer's disease.8 A heterogeneity ofhypometabolism patterns in Alzheimer's dis-ease was also reported in PET studies.9 10 Inthe absence of vascular pathology, focaldeficits in cortical blood flow may reflecteither abnormalities in the cortex itself ordeafferentation from the underlying whitematter. The possible contributions of whitematter hyperintensities seen on MRI to thepathogenesis of Alzheimer's disease and to theheterogenous impairments of cognitive func-tion and rCBF remain to be clarified. Also, it isnot known whether white matter hyperinten-sities seen on MRI during the course of thedisease have any relation with the histopatho-logical white matter disorder described inneuropathological studies.1The aim of this study was to examine the

correlation ofMRI white matter hyperintensi-ties with rCBF as measured with SPECT and[99mTc]-d,l-HMPAO. In particular, we aimedto investigate the hypothesis that the pres-ence, topography, or severity of white matterhyperintensities could contribute to the rCBFreduction in the hippocampal regions or tothe topographical heterogeneity of rCBFdeficits in neocortical brain regions. Thisreport also includes the postmortem neuro-pathological study in one patient with Alz-heimer's disease and with extensive deepwhite matter hyperintensities on MRI.

Patients and methodsPATIENTSThis prospective study comprised 18 righthanded patients who were consecutivelyreferred to the dementia clinic for diagnosis.They fulfilled the DSM-IIIR" criteria fordementia and the National Institute ofNeurological Disorder and Stroke and theAlzheimer Disease and Related DiseaseAssociation (NINCDS-ADRDA)"2 criteria forprobable Alzheimer's disease (table 1). Thepatients were part of a prospective dementia

1458

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

White matter magnetic resonance hyperintensities in dementia of the Alzheimer type: morphological and regional cerebral bloodflow correlates

Table 1 Clinical characteristics and MRI data

Control Alzheimer's disease

Clinical data:No of subjects (female/male) 10 (4/6) 18 (11/7)Age 64 (53-80) 69 (53-78) NSDuration of symptoms (y) - 4 (0-5-10)MMSE'4 score (max 30) 30 (28-30) 17 (3-27) **GDS" score (max 7) - 4 (3-6d)

Resting arterial blood pressure:Systolic 130 (110-160) 140 (95-165) NSDiastolic 85 (70-90) 78 (60-95) NS

MRI data:PVH score 0-0 (0-1) 1-5 (0-3) **DWMH score 1-0 (0-1) 1-0 (0-3) NSTotal volume ofDWMH (cm3) 0-07 (0-4-96) 1-04 (0-47-6) NS

(mean (SD)) (0-79 (1-55)) (5-39 (11 3))Total volume of ventricles (cm3) 24-5 (8 9-56 2) 46-9 (18-6-127-2) NS

(mean (SD)) (31-1 (17-8)) (54-2 (33-1))

Values are median (range) unless stated otherwise. Control = healthy control subjects;Alzheimer's disease = patients with a clinical diagnosis of probable Alzheimer's disease; MMSE =mini mental state examination'4; GDS = global deterioration scale'5; PVH = periventricularhyperintensities; DWMH = deep white matter hyperintensities. **p < 0-01, NS = non-significant (patients with Alzheimer's disease v control subjects, Wilcoxon two sample rank sumtest).

study reported elsewhere,8 and the presentreport includes only patients in whom a cra-nial MRI study was available. In one patient(case 04-19), the diagnosis was later con-firmed by postmortem neuropathologicalexamination. All patients underwent an exten-sive study programme to rule out any otherdisorders that might be associated with demen-tia, cognitive dysfunction, or altered rCBF.8Briefly, cranial x ray CT was without focalpathology and without periventricular hypo-density in all patients. No patient had a historyof psychiatric or neurological disease, exceptfor Alzheimer's disease. All patients hadHamilton" depression scores below or at 10.Cerebrovascular risk factors were minimised:patients with diabetes mellitus, moderate tosevere impairment of cardiac or pulmonaryfunction, significant stenosis of the carotidarteries on Doppler examination, chronic arter-ial hypertension, or increased cholesterol con-centrations in the blood were excluded.Resting systolic/diastolic blood pressure wasbelow 180/100 on the study day. The severityof dementia was documented (table 1) by themini mental state examination (MMSE)'4 andthe global deterioration scale (GDS).'5

Ten age matched healthy volunteers,recruited from advertisements, served as acontrol group for the MRI data (table 1).They had no dementia symptoms or signs,and they met the same exclusion criteria asthe patients with Alzheimer's disease. Themapping of significant rCBF deficits in thepatients with Alzheimer's disease was basedon the rCBF data from a larger control groupwith 25 healthy control subjects, median age70 (range 53-83) years.8 The patients, withtheir relatives, and all control subjects gaveinformed consent to participation in thestudy, which was approved by the ethics com-mittee of the Cities of Copenhagen andFrederiksberg.

MAGNETIC RESONANCE IMAGINGThe MRI study was carried out with a wholebodyMR scanner (Siemens Magnetom H- 15)operating at 1-5 T. The brain was imaged inthe axial plane using a double spin echosequence with repetition time of 1800 ms andan echo time of 30/90 ms. The slice thicknesswas 4 mm, and the number of slices was 15with an interslice space of 4 mm. The matrixsize was 256 x 256 giving a voxel size of1-2 x 1-2 x 4 mm3. The interslice spaces leftwere also imaged.The images were analysed by two experi-

enced readers of MRI who were notacquainted with the diagnosis of the patientsor the rCBF data. Firstly, the readers wereasked to rate (in consensus) the presence,shape, and severity of any periventricularhyperintensities (PVHs) and deep whitematter hyperintensities (DWMHs) accordingto a qualitative scoring method based on theratings defined by Fazekas et at2 andZimmermann et al.'6 The PVHs were rated ona 0-4 scale (0 = absent; 1 = discontinuousPVH; 2 = continuous periventricular lining; 3= continuous periventricular halo; 4 = contin-uous irregular PVHs). Discontinuous PVHswere defined as rounded hyperintense foci atthe angles of the frontal horns, caps of hyper-intensity surrounding the occipital horns, and

Table 2 Magnetic resonance hyperintensities and rCBF deficits in patients with probable Alzheimer's disease

DWMH volume (cm3)Cortical rCBF deficits

PVH DWMH Central Left lobar Right lobarCase score score Total white matter white matter white matter Left Right

01-02 2 0 0 0 0 0 T, P F, T, P02-09 0 1 2-08 2-08 0 0 - F03-17 2 1 0-68 0-68 0 0 F, T F04-19 1 2 15-20 15-20 0 0 T F, T, P, OC05-21 1 1 10-04 9-24 0 48 (F+OC) 0-32 (F) - F, T06-23 1 1 4 52 4 40 0 0-12 (P) T, P, OC T, P07-26 1 1 3 50 3-48 0-02 (OC) 0 F, P F08-32 0 0 0 0 0 0 F, T, P, OC -09-33 2 1 4-52 1-96 1-40 (OC) 1 16 (OC) T F, T10-34 0 1 0-92 0-56 0-28 (OC) 0-08 (OC) F11-37 2 1 0-10 0 10 0 0 F, T, P F12-41 3 3 47-62 47-62 0 0 F, T,P -13-43 2 1 0-64 0-64 0 0 - T, P14-44 3 1 1-16 1-16 0 0 - OC15-47 2 1 044 028 0 0-16 (F) P16-51 1 1 5-12 5-12 0 0 T,P -17-52 2 2 0-40 0-40 0 0 T, P F, T, P18-53 0 0 0 0 0 0 T F, T, P

rCBF = Regional cerebral blood flow; PVH = periventricular hyperintensities; DWMH = deep white matter hyperintensities;F = frontal cortex; T = temporal cortex; P = parietal cortex; OC = occipital cortex.

1459

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

Waldemar, Christiansen, Larsson, H0gh, Laursen, Lassen, Paulson

Table 3 Correlation coefficients for white matter magnetic resonance hyperintensities vclinical data and rCBF in patients with probable Alzheimer's disease (n = 18)

PVH score DWMH score DWMH volume

Age 0-46 0-52t 0 29Age at onset 0-45 0 44 0 30Duration of symptoms - 007 - 022 0-23Resting arterial blood pressure:

Systolic 0-76tt 0O48t 0.23Diastolic 0-47 0 46 0 28

MRI data:PVH score NR 0-51t 0-44DWMH score 0.51t NR 0 76***DWMH volume 0 11 - 076tt NRVolume of ventricles 0 52t 0-38 0.47*

MMSE score -0 17 0-25 0.20Cerebral blood flow (a/b):

Global CBF - 034/- 035 -0-29/- 014 -052/- 0-24rCBF

Frontal cortex - 038/- 0-36 - 031/- 0-10 -0-42/-0 30Temporal cortex - 029/- 027 -0 26/-0 09 -0-46/-0 18Parietal cortex - 0 32/-0-24 - 0 24/-0-06 -0-27/- 0 24Occipital cortex -0 41/-0 34 -0 32/-0-14 -0 30/-0-26Hippocampal region -0 15/NA -0 22/NA -0-72***/NACentral white matter - 042/NA - 0-28/NA - 0-56*/NA

*p < 0 05; ***p < 0-001 (Pearson product-moment correlations).tp < 0 05; ttp < 0 01 (Spearman rank order correlations).a/b = correlation coefficients; these are shown for rCBF data obtained with both CBF methods([99-Tc]-d,l-HMPAO(a)/"'3Xe inhalation (b)); PVH = periventricular white matter hyper-intensities; DWMH = deep white matter hyperintensities; MMSE = mini mental state examinationscore; NA = not available; NR = not relevant.

streaks of hyperintensity extending along theatria of the lateral ventricles. Continuousirregular PVHs were defined as PVHs extend-ing from the ventricular lining into the deepwhite matter to the corticomedullary junctioninvolving most of the white matter, the lateralmargins of the hyperintensity being irregular.Separate white matter signal hyperintensitiesthat were not confluent with the ventricularlining were rated as DWMHs on a 0-3 scale((O = absent): 1 = punctuate (solitary) foci; 2 =beginning of confluence of foci; 3 = large con-fluent areas). Next, the readers measured thevolume of all DWMH elements in: (a) thecentral white matter in the frontal and occipitalregions and parallel to the lateral ventricles,including all white matter tissue within thecorticomedullary junction; (b) lobar whitematter separately in the left and right frontal,temporal, parietal, and occipital regions.These regions were defined on templates con-structed before the study with reference toaxial brain slices from an anatomical brainatlas.17 Each hyperintensity lesion was encir-cled with a cursor on the computer screen,and the area was calculated automatically.The slice volume of each lesion was estimatedby multiplying by the voxel volume. The totalvolume of the hyperintensities was calculatedby adding together the volume of all lesionson all slices. The total volume of the ventricleswas measured by a similar procedure.

SINGLE PHOTON EMISSION COMPUTEDTOMOGRAPHYFirstly, rCBF was measured by the '33Xeinhalation technique with a Tomomatic 64(Medimatic, Hellerup, Denmark), a rapidlyrotating and highly sensitive brain dedicatedthree slice instrument for brain SPECTdescribed elsewhere. 18 The 133Xe inhalationstudy was performed to obtain absolutemeasures for CBF. From this study globalCBF and rCBF in the cerebellum and in thefrontal, temporal, parietal, and occipital cor-tex were calculated. Methods for calculation

of CBF and regional data analysis have beendescribed elsewhere.'8-20 Subsequently, rCBFwas measured semiquantitatively by staticimaging of the distribution of an intravenousdose (1.1 GBq) of [99mTc]-d,l-HMPAO(Exametazime, Ceretec, Amersham, London,UK). The imaging protocol and the analysisof rCBF data have been described elsewhere.820Briefly, the count rates from nine contiguousimage slices covering the whole brain werecorrected to adjust for the initial preferentialback diffusion of the tracer from high flowregions by the algorithm suggested by Lassen etal,21 with a conversion to clearance ratio of1-5. The cerebellar hemisphere with the high-est count rate was used as the referenceregion. Mean rCBF was calculated in the hip-pocampal regions and in frontal, temporal,parietal, and occipital cortical regions of inter-est.82022 The side to side asymmetry index wasdefined as the difference between rCBF val-ues in the right and the left counterparts of theregion of interest, relative to the higher of thetwo values. Two anterior:posterior ratios weredefined as the ratio of mean rCBF in thefrontal cortex to mean rCBF in the temporalcortex or the parietal cortex, respectively.20Based on the rCBF data from a large controlgroup,8 the regions of interest with significantrCBF deficits were mapped for each patient.In each region of interest rCBF was consid-ered reduced if the relevant side to side asym-metry index or anterior:posterior ratio ofrCBF was significantly abnormal (deviatingmore than 2 SD from the control mean), or ifthe mean rCBF value was severely reduced,by more than 3 SD from the control mean.8

CORRELATIONS BETWEEN MRIHYPERINTENSITIES, CLINICAL DATA, AND rCBFThe rating scores for PVHs and DWMHs andthe quantitative volume measurements ofDWMHs were compared in the two groups.Within the group of patients with Alzheimer'sdisease a correlation analysis was carried outfor the PVH and DWMH rating scores andfor the volume of DWMHs v age, MMSEscore, age of disease onset, duration of symp-toms, resting systolic and diastolic arterialblood pressure, volume of ventricles, andglobal as well as rCBF. A correlation analysiswas also performed for the side to side asym-metry of the global DWMH volume v that ofglobal CBF. Finally, the MRI variables werecompared in the groups of patients with andwithout frontal hypoperfusion and, when rele-vant, the rCBF patterns were examined forrCBF deficits in the cortical regions of interestadjacent to any lobar DWMH lesions.

STATISTICAL METHODSThe Wilcoxon two sample rank sum test wasused for intergroup comparisons of data. Thepossible relations between the quantitativeDMWH volume measurements and othervariables were analysed by Pearson productmoment correlations. Correlations concern-ing the qualitative rating scores for PVHs andDWMHs were analysed by the Spearmanrank order analysis. A Bonferroni correction

1460

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

White matter magnetic resonance hyperintensities in dementia of the Alzheimer type: morphological and regional cerebral bloodflow correlates

was applied to the p values for differences inrCBF data.8 For all statistical tests theminimal level of significance was p < 0-05.

ResultsMAGNETIC RESONANCE IMAGINGThe rating scores for PVHs were significantly(p < 0-01) higher in the Alzheimer's diseasegroup than in the controls (table 1). By con-trast, there was no significant difference in therating scores or volume measurements ofDWMHs between the two groups (table 1).In all control subjects and in most patients

Figure1 MRI images (top) and [99mTc]-d,l-HMPAO SPECTimages (bottom) from a72 year oldfemale patient with a diagnosis of definite Alzheimer's disease verified byneuropathological examination at necropsy (case 04-19). The left hemisphere is shown tothe left. In the SPECT images the colour scale indicates rCBF (normalised to thecerebellum). This patient presented with presenile onset ofprogressive dementia, fulfillingthe clinical criteria for probable Alzheimer's disease. The MMSE score was 15, and theGDS (global deterioration scale) score was 5. There was no history of cerebrovascular riskfactors. In the neuropsychological examination memory, language, visual perception, andvisuoconstruction were significantly impaired. On MRI (top) a moderate and diffusecortical atrophy was seen, the PVH score was 1, and the DWMH score was 2. ExtensiveDWMH lesions, with a total volume of 15 2 cm3, were located symmetrically in the centralwhite matter. The SPECT study (bottom) showed asymmetric focal rCBF deficits in thetemporoparietal regions and in the rightfrontal cortex. The symptoms and SPECTabnormalities continued to progress, still with no clinical signs of vascular brain disorder.Fig 2 shows the postmortem neuropathological examination obtained three years later.

1461

with Alzheimer's disease the total volume ofDWMHs was below 5 cm3, and more than95% of the total DWMH volume was locatedin the central white matter (tables 1 and 2). Inthree patients with Alzheimer's disease (casesNo 04-19, 05-21, and 12-41), however, thetotal volume of DWMHs was very high,exceeding by far the maximal DWMH volumein the control group.

SINGLE PROTON EMISSION COMPUTEDTOMOGRAPHYGlobal CBF measured with the [99mTc]-d,l-HMPAO method (normalised to the cerebel-lum) was 85-2 (SD 7 9)% in the controlgroup and 75 2 (9 6)% in the Alzheimer's dis-ease group (p < 0-01). There was a significant(p < 0-01) reduction of rCBF in the frontal,temporal, and parietal cortexes, and in thehippocampal regions, whereas the occipitalcortex and the subcortical grey matter regionswere spared.8 The individual rCBF patternshave been described elsewhere8 and table 2summarises these. All patients had an abnor-mal rCBF pattern. A pronounced topographi-cal heterogeneity of rCBF patterns for thenumber of affected regions, laterality, andanterior posterior asymmetry, was found.There was no significant difference betweenthe two groups in cerebellar blood flow asmeasured with the "'3Xe inhalation technique.

CORRELATIONS BETWEEN MRI

HYPERINTENSITIE S AND CLINICAL DATAWithin the group of patients with Alzheimer'sdisease the DWMH but not the PVH ratingscores correlated significantly with age. ThePVH scores and the DWMH volumes corre-lated significantly with the volume of the ven-tricles (table 3). There was no significantdifference in any of the MRI variablesbetween patients with presenile and senileonset of symptoms. The PVH scores corre-lated significantly with resting systolic arterialblood pressure (r = 0-76; p < 0-01). Likewise,there was a slight, but significant, correlationbetween the DMWH scores and systolic arter-ial blood pressure (r = 0-48; p < 0-05). Noneof the MRI variables correlated with age atonset, duration of symptoms, diastolic arterialblood pressure, or cognitive performance asmeasured with the MMSE score.

CORRELATIONS BETWEEN MRIHYPERINTENSITIES AND rCBFThe DWMH volumes correlated significantlywith rCBF in the central white matter (r =-0-56; p < 0 05), and remarkably well withrCBF in the hippocampal region (r = -0-72; p< 0-001), but not with rCBF in any of thefour major cortical regions (table 3).Accordingly, there was no significant correla-tion between the side to side asymmetry of theDWMH volume and that of cortical rCBF,and there was no significant difference in anyof -the MRI variables between patients withand without frontal hypoperfusion. Only twoof the six patients with small DWMH lesions

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

Waldemar, Christiansen, Larsson, H0gh, Laursen, Lassen, Paulson

Figure 2 Histological tissue preparations showing white matter lesions in patient 04-19 (for case history seefig 1).(A) Macroscopically identified white matter lesion. Myelin stain shows extensive demyelination. Khwver-Barrera stain(bar 180 pm). (B) Same lesion as in A. Tissue heavily infiltrated by macrophages. Haematoxylin-eosin stain (bar 70pm). (C) Macroscopically unidentified white matter lesion. Myelin stain demonstrates pallor and oedematous dispersion ofmyelinated tracts, but no demyelination. Kiuver-Barrera stain (bar 70 pm). (D) Same lesion as in C. Astrocytes areswollen with distorted outlines. Glialfibrillaty acid protein immunostain (bar 30 pm).

in the lobar white matter had rCBF deficits inone of the relevant adjacent cortical regions(table 2). No significant correlations were

found between the qualitative rating scores forPVHs and DWMHs and rCBF. The topogra-phy of cortical rCBF deficits in the rCBF pat-terns could not be subtyped according to thepresence or severity of white matter lesions(table 2). In particular, the cortical rCBF pat-terns of the three patients with extensiveDWMH lesions did not differ from those ofthe other patients.

NEUROPATHOLOGYIn one of the patients (case 04-19) with

extensive DWMH lesions a postmortem neu-

ropathological examination was obtained (figs1 and 2). Abundant plaques and tangles were

seen in all cortical brain regions. Only very

few of the deep white matter lesions on MRIcould be identified macroscopically. Tissueblocks taken from these areas and fromperiventricular areas with well defined MRIlesions were examined histologically (fig 2).

Some of the lesions were easily identified assmall areas of demyelination infiltrated withmacrophages and with astrocytosis. In otherlesions, the only obvious change was astrocy-tosis identified by increased immunoreactivityin glial fibrillary acid protein immunohisto-stains. In some of these lesions a slight pallor inthe myelin stain suggested oedematous dis-persion of fibre tracts, but obvious demyelina-tion was not present. In all lesions hyalinisedblood vessels were present, but completeobstruction or thrombosis was not found.

DiscussionThe highly significant correlation between thevolume ofDWMHs and hypoperfusion in thehippocampal regions suggested that DWMHsmay be a marker for the severity of dysfunc-tion or atrophy in medial temporal lobe struc-tures. This study also showed that whitematter hyperintensities could not explain thecharacteristic topographical heterogeneity offocal cortical rCBF deficits in dementia of the

1462

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

White matter magnetic resonance hyperintensities in dementia of the Alzheimer type: morphological and regional cerebral bloodflow correlates

Alzheimer type. Therefore, the heterogeneitymust reflect variations in the topography ofcortical pathology. We used an MRI scoringmethod that distinguished between PVHs andDWMHs, because histological and epidemio-logical studies have suggested a different aeti-ology for these categories of white matterlesions."3 23 24 The inter-rater reliability of thescoring method has been shown to be veryhigh,5 and in our study the qualitative estima-tion of DWMHs was refined by quantitativevolumetric measurements.The PVH foci just anterior to the frontal

horns probably do not reflect pathology. Theyare thought to be caused by an age related lossof ependyma leading to increased water con-tent.23 Thus if a PVH score greater than 1 isabnormal, then abnormal PVHs (continuousPVHs along the side of the ventricles) wereseen in 50% of our patients with Alzheimer'sdisease but not in any of the healthy subjects.The PVH scores correlated significantly withsystolic arterial blood pressure, but differ-ences in blood pressure could not explain thedifferences in PVHs between the two groups.The low frequency of PVHs in our group ofhealthy subjects without major cerebrovascu-lar risk factors corresponds to that reportedby Fazekas et al.25 The increased frequencyand extension of PVHs in patients withAlzheimer's disease, which suggested thatPVHs may be associated with the Alzheimer'sdisease process, is in agreement with the find-ings of McDonald et al.26 By contrast, Leys etal,6 Kozachuk et al,5 and Fazekas et al27 failedto find any difference in PVHs betweenpatients with Alzheimer's disease and healthysubjects. Scheltens et al7 found increasedseverity of PVHs in his group of patients withAlzheimer's disease and with senile onset ofdisease, but not in presenile onset Alzheimer'sdisease. We were unable to show any associa-tion between the severity of PVHs and age ofonset or duration of symptoms. Thus whereasour data on the presence of PVHs inAlzheimer's disease confirmed the results ofsome, but not all, previous reports, the impor-tant finding was that although PVHs were sig-nificantly associated with the diagnosis ofAlzheimer's disease PVHs were not a deter-mining factor for the cortical rCBF deficits orfor global cognitive performance.By contrast with PVHs the rating scores

and volumes of DWMHs were not higher inthe group of patients with Alzheimer's diseaseas a whole. Solitary foci of DWMHs wereseen in 70% of our control subjects, a highfrequency that corresponds with that reportedby Fazekas et al." The lack of difference inDWMH scores between patients withAlzheimer's disease and normal subjects wasin agreement with the findings of some previ-ous reports,'s628 but contradicted others.' 29-33It is well known that the presence of cere-brovascular risk factors increases the risk ofDWMHs,26'6 and Scheltens et al7 suggestedthat DWMHs may be more prevalent inpatients with senile onset of Alzheimer's dis-*ease than in patients with presenile onset,even when cerebrovascular risk factors have

been controlled. In previous studies only sub-jective rating methods have been applied. Byinclusion of volumetric measurements ofDWMHs, our study confirmed that whencerebrovascular risk factors have been con-trolled, the clinical diagnosis of probableAlzheimer's disease is not likely associatedwith increased DWMHs. Within the group ofpatients with Alzheimer's disease, however,there was a high variation in the volumetricdata, and three patients had extensiveDWMH lesions, suggesting a heterogeneity inthe presence and severity of DWMHs. A veryhigh variation in DWMH volume measure-ments was also reported by others.35Furthermore, the highly significant negativecorrelation with blood flow in the hippocam-pal regions points to a possible pathophysio-logical link between DWMHs and thedegenerative process of Alzheimer's disease,at least in some patients.

Three questions remain unanswered.Firstly, could DWMHs be part of theAlzheimer's disease process in selectedpatients? Secondly, if DWMHs on MRI arenot associated with Alzheimer's disease, thenwhat are the clinical correlates of the histolog-ical white matter lesions that have beendescribed in 60% of patients with Alzheimer'sdisease at necropsy? Brun and Englund sug-gested that these abnormalities were due tohypoperfusion of the concerned white matterterritories, but no information from MRI wasavailable in their study.' Thirdly, what are themorphological correlates of the DWMHlesions seen on MRI in healthy subjects aswell as in patients with Alzheimer's disease?The literature on histopathological examina-tion of MRI white matter lesions inAlzheimer's disease is sparse: in a preliminarypostmortem report on three patients with def-inite Alzheimer's disease Morris et al36 foundincreased water content, but no histologicallesions, in regions with "periventricular highsignals". Whether these high signals repre-sented deep white matter hyperintensities orperiventricular hyperintensities was notdefined in their report, and the vessels werenot described. We were able to perform apostmortem neuropathological examinationof the brain from one of our patients with def-inite Alzheimer's disease and extensiveDWMH lesions on MRI. The partial loss ofmyelin and astrocytic gliosis, which was seenin the white matter regions corresponding tothose containing DWMHs on MRI, resem-bled some of the abnormalities described inthe necropsy studies by Brun and Englund inpatients with Alzheimer's disease.' Similarhistopathological changes were found, how-ever, by Fazekas et al24 in healthy subjectswith DWMHs on MRI. In our patienthyalinised blood vessels were found in thecentral white matter, but they were not associ-ated with vessel occlusion or ischaemicchanges. Thus perivascular damage could notbe proved to be the substrate of the DWMHs.We suggest that neither the DWMHs on MRInor the histological white matter findings inour patient were related to the Alzheimer's

1463

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

Waldemar, Christiansen, Larsson, Hogh, Laursen, Lassen, Paulson

disease process. Most likely DWMHs seen onMRI are related to normal aging, cerebrovas-cular risk factors, and chronic cerebrovasculardisease,37 rather than to the Alzheimer's dis-ease process. The prevalent histological whitematter disorder described by Brun andEnglund' in Alzheimer's disease as well as ourfinding of a significant correlation between thevolume of DWMHs and hypoperfusion in thehippocampal regions, however, support thehypothesis of a pathophysiological linkbetween DWMHs and Alzheimer's disease.The DWMHs did not contribute to the

characteristic cortical rCBF deficits inAlzheimer's disease. The finding was con-firmed with the '33Xe inhalation method aswell as with the [99mTc]-d,l-HMPAO tech-nique for measuring rCBF. In accordancewith this, measurements of glucose metabo-lism with PET have failed to show hypome-tabolism of cortical areas adjacent to whitematter hyperintensities.? Although none ofour patients had chronic arterial hypertension,the significant correlation between the severityof PVHs (and to some extent also DWMHs)and systolic arterial blood pressure supportsthe hypothesis that arterial blood pressuremay be an aetiological factor for white matterhyperintensities.

In conclusion, clinically diagnosedAlzheimer's disease without major cerebro-vascular risk factors was associated with anincreased frequency and severity of PVHs.There was no relation, however, between thepresence of PVHs and global cognitive perfor-mance or the.rCBF patterns. By contrast,DWMHs were not significantly associatedwith the diagnosis of Alzheimer's disease. Thehistopathology of extensive DWMH lesions inone patient with Alzheimer's disease includedpartial loss of myelin and astrocytosis,changes that were not different from thosedescribed in normal subjects with DWMHs.Within the group of patients with Alzheimer'sdisease, however, the variation in DWMHvolumes was high, and the total volume ofDWMHs correlated significantly with thereduction of rCBF in the hippocampalregions, suggesting a pathophysiological linkbetween DWMHs and Alzheimer's disease.The characteristic and heterogeneous topo-graphy of focal cortical rCBF deficits in prob-able Alzheimer's disease could not beexplained by deafferentation from whitematter hyperintensities and therefore mustreflect variations in the topography of corticalabnormalities.This study was supported by grants from the Danish MedicalResearch Council, the Lundbeck Foundation, and the DanishHeart Foundation. Dorthe Givard, Glenna Skouboe, PiaTejmer, and Gerda Thomsen are gratefully thanked for theirexcellent secretarial and technical assistance.

Brun A, Englund E. A white matter disorder in dementiaof the Alzheimer type: a pathoanatomical study. AnnNeurol 1986;19:253-62.

2 Fazekas F, Chawluk JB, Alavi A, Hurtig HI, ZimmermanRA. MR signal abnormalities at 1-5 T in Alzheimer'sdementia and normal aging. AJNR Am J Neuroradiol1987;8:421-6.

3 Kertesz A, Polk M, Carr T. Cognition and white matter

changes on magnetic resonance imaging in dementia.Arch Neurol 1990;47:387-91.

4 Christie JE, Kean DM, Douglas RHB, Engleman HM,Clair DS, Blackburn IM. Magnetic resonance imagingin pre-senile dementia of the Alzheimer type, multi-infarct dementia and Korsakoff's syndrome. Psychol Med1988;18:319-29.

5 Kozachuk WE, DeCarli C, Schapiro MB, Wagner EE,Rapoport SI, Horwitz B. White matter hyperintensitiesin dementia of Alzheimer' type and in healthy subjectswithout cerebrovascular risk factors. A magnetic reson-ance imaging study. Arch Neurol 1990;47:1306-10.

6 Leys D, Soetaert G, Petit H, Fauquette A, Pruvo J-P,Steinling M. Periventricular and white matter magneticresonance imaging hyperintensities do not differbetween Alzheimer's disease and normal aging. ArchNeurol 1990;47:524-7.

7 Scheltens P, Barkhof F, Valk J, et al. White matter lesionson magnetic resonance imaging in clinically diagnosedAlzheimer's disease. Evidence for heterogeneity. Brain1992;115:735-48.

8 Waldemar G, Bruhn P, Kristensen M, Johnsen A, PaulsonOB, Lassen NA. Heterogeneity of neocortical cerebralblood flow deficits in dementia of the Alzheimer type: a[99-Tc]-d,l-HMPAO SPECT study. J Neurol NeurosurgPsychiatry 1994;57:285-95.

9 Haxby JV, Grady CL, Koss E, et al. Heterogenousanterior-posterior metabolic patterns in dementia of theAlzheimer type. Neurology 1988;38:1853-63.

10 Rapoport SI. Positron emission tomography inAlzheimer's disease in relation to disease pathogenesis: acritical review. Cerebrovasc Brain Metab Rev 1991;3:297-335.

11 American Psychiatric Association. Diagnostic and statisticalmanual of mental disorders. 3rd ed revised. DSM-III-R.Washington, DC: American Psychiatric Association,1987.

12 McKhann G, Drachman D, Folstein M, Katzman R, PriceD, Stadlan EM. Clinical diagnosis of Alzheimer'sdisease: report of the NINCDS-ADRDA Work Groupunder the auspices of Department of Health and HumanServices Task Force on Alzheimer's Disease. Neurology1984;34:939-44.

13 Hamilton M. Development of a rating scale for primarydepression illness. British Journal of Social and ClinicalPsychology 1967;6:278-96.

14 Folstein MF, Folstein SE, McHugh PR. "Mini-MentalState". A practical method for grading the cognitive stateof patients for the clinician. Jf Psychiatr Res 1975;12:189-98.

15 Reisberg B, Ferris SH, de Leon MJ, Crook T. The globaldeterioration scale for assessment of primary degenera-tive dementia. AmJPsychiatry 1982;139:1136-9.

16 Zimmerman RD, Fleming CA, Lee BCP, Saint-Louis LA,DeckMDF. Periventricular hyperintensity as seen bymagnetic resonance: prevalence and significance. AJNRAmJ7 Neuroradiol 1986;7: 13-20.

17 Eyclesheimer AC, Schoemaker DM. A cross-sectionanatomy. London: Butterworths, 1970.

18 Stokely EM, Sveinsdottir E, Lassen NA, Rommer P. Asingle photon dynamic computer assisted tomograph(DCAT) for imaging brain function in multiple crosssections. J Comput Assist Tomogr 1980;4:230-40.

19 Celsis P, Goldman T, Henriksen L, Lassen NA. A methodfor calculating regional cerebral blood flow from emis-sion computed tomography of inert gas concentrations.JComputAssist Tomogr 1881;5:641-5.

20 Waldemar G, Hasselbalch SG, Andersen AR, et al.[9"Tc]-d,l-HMPAO and SPECT of the brain in normalaging. J Cereb Blood Flow Metab 199 1;l1:508-2 1.

21 Lassen NA, Andersen AR, Friberg L, Paulson OB.Theretention of [99-Tc]-d,l-HMPAO in the human brainafter intracarotid bolus injection. A kinetic analysis.J Cereb Blood Flow Metab 1988;8:13-22.

22 Aquilonius S-M, Eckemras S-A. A colour adas of the humanbrain. Adapted to computed tomography. Stockholm:Esselte Studium, 1980.

23 Sze G, De Armond SJ, Brant-Zawadzki M, Davis RL,Norman D, Newton TH. Foci of MRI signal (pseudolesions) anterior to the frontal horns: histologic correla-tions of a normal finding. American Journal ofRoentgenology, Radium Therapy, and Nuclear Medicine1986;147:331-7.

24 Fazekas F, Kleinert R, Offenbacher H, et al. The morpho-logic correlate of incidental punctate white matterhyperintensities on MR images. AJNR AmJf Neuroradiol1991;12:915-21.

25 Fazekas F, Schmidt R, Offenbacher H, et al. Prevalence ofwhite matter and periventricular magnetic resonancehyperintensities in asymptomatic volunteers. Journal ofNeuroimaging 1991;1:27-30.

26 McDonald WM, Krishnan KRR, Doraiswamy PM, et al.Magnetic resonance findings in patients with early-onsetAlzheimer's disease. Biol Psychiatry 1991;29:799-8 10.

27 Fazekas F, Schmidt R, Offenbacher H, Chawluk JB, AlaviA. Magnetic resonance imaging hyperintensities inAlzheimer's disease. [letter]. Arch Neurol 1991;48:468-9.

28 Fazekas F, Alavi A, ChawlukJB, et al. Comparison of CT,MR, and PET in Alzheimer' dementia and normalaging. JNucl Med 1989;30:1607-15.

29 Bondareff W, Raval J, Coletti PM, Hauser DL.Quantitative magnetic resonance imaging and the severityof dementia in Alzheimer's disease. Am7 Psychiatry1988;145:853-6.

1464

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

White matter magnetic resonance hyperintensities in dementia of the Alzheimer type: morphological and regional cerebral bloodflow correlates

30 Bondareff W, Raval J, Woo B, Hauser DL, Colletti PM.Magnetic resonance imaging and the severity of dementiain older adults. Arch Gen Psychiatry 1990;47:47-5 1.

31 Johnson KA, Davis KR, Bounanno FS, Brady TJ, RosenTJ, Growdon JH. Comparison of magnetic resonanceand roentgen ray computed tomography in dementia.Arch Neurol 1987;44:1075-80.

32 Besson JAO, Corrigan FM, Foreman EI, Eastwood LM,Smith FW, Ashcroft GW. Nuclear magnetic resonance(NMR) II. Imaging in dementia. Br 7 Psychiatry 1985;146:31-5.

33 George AE, de Leon MJ, Kalnin A, Rosner L, GoodgoldA, Chase N. Leukoencephalopathy in normal andpathologic aging: 2. MRI of brain lucencies. AJNRAmJNeurobiol 1986;7:567-70.

34 Fazekas F, NiederkornmK, Schmidt R, Offenbacher H,Homer S, Bertha G, Lechner H. White matter signal

abnormalities in normal individuals: Correlation withcarotid ultrasonography, cerebral blood flow measure-ments, and cerebrovascular risk factors. Stroke 1988;19:1285-8.

35 Almkvist 0, Wahlund L-O, Andersson-Lundman G,Basun H, Backman L. White matter hyperintensity andneuropsychological function in dementia and healthyaging. Arch Neurol 1992;49:626-32.

36 Morris JC, Gado M, Torack RM, McKeel DW.Binswanger's disease or artifact: a clinical andpathological study of periventricular white matterchanges in Alzheimers' disease. Adv Neurol 1990;51:47-52.

37 Awad IA, Johnson PC, Spetzler RF, Hodak JA. Incidentalsubcortical lesions identified on magnetic resonanceimaging in the elderly. II. Postmortem pathologicalcorrelations. Stroke 1986;17:1090-7.

NEUROLOGICAL STAMP

Dorothea Lynde Dix (1802-87)

Dorothea Dix came from New England. Here she noticedthe horrible conditions in an asylum. Some warders madea small charge for visitors to come and look at the madand prod them into raving antics. In 1841 she laid beforethe Massachusetts General Assembly a description of theinsane in jails and private institutions. She spent the restof her life fighting to improve conditions of the insane.Anton Ashley Cooper, afterwards Lord Shaftesbury,

had discovered similar problems in England 10 years ear-lier while sitting on a Royal Commission to investigatesuch conditions. The lunatics at Bedlam were longregarded as one of the sights of London.

Miss Dix travelled thousands of miles by stagecoach,steamboat, and on horseback and quietly collected facts,which she recorded in sober restrained prose. She pos-sessed great charm and a commanding voice. With theseattributes, facts collected from a comprehensive study ofasylums, and a selection of public speakers to sustainprotest she harassed officials and influenced legislatures.

In 1843, in the face of her continuing investigation andincreasing community agitation, the MassachusettsLegislature passed a Bill providing for adequate hospitali-sation of the insane. This was her first triumph. Shebecame an effective lobbyist in many of the States and anational figure. Her work and influence also led toimprovement of the conditions of the insane in GreatBritain-and also in Italy after an audience with the Pope.She travelled in Europe and crusaded for human rights ofpatients until her death at the age of 85.The centennial of her death, 1987, went largely un-

noticed but in 1980 a United States postage stamp hadcommemorated the first woman who brought the plightof the mentally ill into such prominence (Stanley Gibbons18188, Scott 1844).

L F HAAS

1465

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from

doi: 10.1136/jnnp.57.12.1458 1994 57: 1458-1465J Neurol Neurosurg Psychiatry

 G Waldemar, P Christiansen, H B Larsson, et al. cerebral blood flow correlates.Alzheimer type: morphological and regionalhyperintensities in dementia of the White matter magnetic resonance

http://jnnp.bmj.com/content/57/12/1458Updated information and services can be found at:

These include:

References http://jnnp.bmj.com/content/57/12/1458#related-urls

Article cited in:

serviceEmail alerting

the box at the top right corner of the online article.Receive free email alerts when new articles cite this article. Sign up in

Notes

http://group.bmj.com/group/rights-licensing/permissionsTo request permissions go to:

http://journals.bmj.com/cgi/reprintformTo order reprints go to:

http://group.bmj.com/subscribe/To subscribe to BMJ go to:

group.bmj.com on July 14, 2011 - Published by jnnp.bmj.comDownloaded from