“EEE
Selective Cortical and Hippocampal Volume
Correlates of Mattis Dementia Rating Scale
in Alzheimer Disease
Rosemary Fama, PhD; Edith V. Sullivan, PhD; Paula K. Shear, PhD; Laura Marsh, MD; Jerome A. Yesavage, MD;
Jared R. Tinklenberg, MD: Kelvin O. Lim, MD; Adolf Pffferbaum, MD
Objective: To examine whether each of the 5 Mattis
Dementia Rating Scale (DRS) scores related to mag-
netic resonance imaging~derived volumes of specific
cortical or limbic brain regions in patients with
Alzheimer disease (AD).
Designs Relations between DRS measures and regional
brain volume measures were tested with bivariate and
multivariate regression analyses.
Setting: The Aging Clinical Research Center ofthe Stan-
{ford (Calif) University Department of Psychiatry and Be-
hhavioral Science and the Geriatric Psychiatry Rehabili-
tation Unit ofthe Veterans Affairs Palo Alto Health Care
System, Palo Alto, Calif
Patients and Other Participants: Fifty patients with
possible or probable AD. Magnetic resonance imaging data
from 136 healthy control participants, age 20 0 years,
were used to correct brain volumes for normal variation
arising from intracranial volume and age.
Main Outcome Measures: The DRS scores and vol-
umes of regional cortical gray matter and of the hippo-
campus.
Resultes Memory scores of the patients with AD were
selectively related to hippocampal volumes, tention
and construction scores were related to several ante-
rior brain volume measures, with attention showing a
significantly greater association to right than left hemi-
sphere measures. Initiation/perseveration scores were
not significantly correlated with any measure of
regional gray matter volume, but performance was
related to prefrontal sulcal widening, with a greater
association with the left chan right sulcal volume.
Conclusions: Certain DRS subtests are predictably
correlated with selective regional brain volumes in
AD. The specific relation berween memory and hippo-
campal volumes and the nonsignificant relations
between memory and regional cortical volumes sug-
gest a dissociation between cortical and hippocampal
contributions to explicit memory performance.
Arch Neurol, 1997;54:719-728
EUROIMAGING studies of
Alzheimer disease (AD)
provide in vivo evidence
for enlargement of ven-
tricularand cortical suleal
volumes'? and complementary deficits in
cortical gray matter volume throughout
much of the brain.™* The regional pattern
of brain volume abnormalities in patients
‘with AD isdistine from that seen in normal
agingand in another neurodegenerative dis-
order, Parkinson disease,” and the most
‘marked deficits re in the medial temporal
regions *#"" These neuroradiological find
{ngsare oonsstentwithpastmortem evidence
of gray matter atrophy in the hip;
and association cortices of AD brains.'*!*
‘Whether the extent of tisste-volume defi-
citsin elective brain regionsin AD srelated
cogniivedeficitsisonly
From the Department of
Paychiatry and Behavioral
Scences, Sanford Universtty
School of Medicine, Stanford
Calif (Drs Fama, Sullivan,
‘Shear, Marsh, Yesavage,
Tinklenberg, Lim, and
Pyeferbaum); the Psychiarry
Service, Veterans Affairs Palo
‘Alto Health Care System, Palo
‘Alt, Calif (Drs Yesavage
Tinklenberg, and Lim) and the
[Neuropsychiatry Program, SRI
International, Menlo Park,
Calif (Drs Sulivan and
Pfefferbaum), Dr Shear Is now
ln the Department of
Paychology, University of
Cincinnat.Cincinatl, ON
toobservable specifi
sparsely established, 2
‘The Mattis Dementia Rating Scale
(DRS) is widely used in clinical and re
search settings as a global measure of cog-
nitive functioning inpatients with AD. The
DRS is composed of 5 subscales (memory,
initation/perseveration, attention, concep-
tualization, and construction) each of which
assesses a different cognitive domain, and
‘total score, which provides an overall in-
dex of dementia severity. The DRS has
greater sensitivity to longitudinal cogni-
tive decline in AD than either of 2 other
commonly used assessment tools, the
Blessed information-Memory-Concentra-
tion test and the Mini-Mental State Exami-
nation (MMSE). The DRS has reasonably
high power to discriminate AD from other
illnesses The indices thought
to best discriminate patients with AD from
healthy older individuals are the memory
and initiation/perseveration subscales”?METHODS AND PARTICIPANTS
METHODS
Paricipansin hist include 50 paces with AD rere
from the Geriatric Rehabilitation UnitandtheNa-
‘onal insiutof Mental Health Aging Clinical Research Cen-
ter, both cated atthe Veteran flats Pal Alto Health Care
‘System, Palo Alto, Cai. Allpaents withAD met the Natonal
Insitute of Neurlogieal and Communicative Diseases and
Sitoke-Altheimer Disease and Related Disorders Association
tena for probable AD" Normal cnolsubjecs sasing
the adult age range (20-84 years of age) were Wed aa refers
ence group or te MRl volumetric measures, These subjects,
‘orasubsetof them, have fommed the nonns fr other stidies
‘fom our laboratory "=" The normal contol group consisted
of 95 men and 41 women fr the axial protocol and 84 men
and28 women forthe coronal protocol. Sreenng forallpar-
‘iipans included a paychian interview and medical exami-
‘ation, Subjects were excludedifthey hada histor fpayehi-
sxticorneurological disorder unrelated to their diagnos (eg,
sake, ose ead injury) pastor resent alebolor drugabuse
ordependence oraserousmedical condition. iiormcscon-
sent was obtained fromall participants o their conservators
Demographic information forthe AD group sas follows
arc Men (30) ame
tony naa) star
elton y wes $2
‘Age at onset of symptoms, y 66.0(8.0) 49-83
Prana ese" mae) ee
SE sae 154) 128
Veeanday ata cot m25 an
® Nato At aig it
t Wc aa iene
DEMENTIA RATING SCALE
The DRS is composed of5 subscales: memory (recall nd
recognition of verbal an visual stall after immediate and
short delays), inidation/perseveration (semantic [supermar-
ket] ueney, motor fluency, and’ tion), attention
orward and backward digi span, abiltytofollow commands,
‘and matchingro sample timull with distractor), conceptu-
alization (abstraction from verbally and visually presented
stimall), and construction (copy of geometric figures and
signature wring).
MRI SCANNING AND QUANTIFICATION
‘Axial Protocol
Acquisition Parameters. Subjects were scanned with 15-7
MBI scanners (Signa, General Elecre, Milwaukee, Wis)
lage acqtstion prsedures and parameters have ben
described in deta °° Axial MR seans were Sram thick
(2.5-mm skip) snd were acquired in an oblique plane us-
inga spin-echo sequence (20- nd 80-ailisevond echoes)
whe 2m ldo wew and 2560256 mar Aen
‘Son was gated to every other cardiac eyce for an ellee-
tive repetidon time of re than 2400 alleconds with
1 Cxctation foreach of 256 phase encodes,
‘Allimages were stored on magnetic tape and trans-
fered to optical disks for analysis. For each subject the
Index slice was identified as the most inferior slice above
the level ofthe orbits, where the anterior horns ofthe lt
ceral ventricles could be seen bilaterally. Seven consecu-
{vee began wih he sone ota
and proceeding superiorly, were analyzed foreach sub-
Jeet. The index slice or the slice below it was used for quan-
‘ification ofthe third ventricle
Regional Divisions and Segmentation of images Each
MI ce was segmented inf certrospinalfukd (CSP,
‘pay mate and whe mae comparanents, sing asem
Somated nage analy technique "To separate the ce
‘ebrl hemispheres amine wat denwn mantaly on exch
Slice Each hier ao was dvded nts inner 39% region
(tacit qunaleaton of cenzal SP, which arse pr
taal bat not exasvely rom the ler venice) and
router 49% (oui quantiication ofthe corte
‘Se volumes and sla CSE)
Theimages were dived acorn to anatomic and
mais tnd a prod geomenie ero achive sundae
‘egal dso ofthe nin images The corteal taverns
du ihr eachinage was dedi 6 rome
cal defined egonsofierest which rough contponccd
telbarsntomy The eons inet ecompa ce
{revalumecf the col labes fe which they were named
bat provided eb baste fo dding coral scone
Tolorm hese dvsonscachMRlsice ws dvdedinio4
Bensty coronal panes which pase rough theron
fer cei tego cerca on
postsiorexreme of splenism ofthe corpscalons a
Fd betwen them. The fis plane erable x bound
yr prefolregon, From tse quadransand sis,
wedevsed coral eon, defined sll prfontal—
‘Gemostanteroguadnfall sles, which inde nox
cle preronal one ronal~—the sbterr mide quad
tantol aces 30 tempore aneioc
idle qushatoflees Land whch nluded cane
‘pero temporal yrassnd te mos posteroretensofthe
froma lobes bee ofthe sper emporsgrspor
‘conser tempor she poner mile uci ostes
‘and whichinaded he pontcroniperor tempor gs
sndasnallporconol theater exerts ofthe preaches
stare sopeiortenpora gyro anerior sia
Posterior mide undress end posteriori
Secipial~thematpoterior quads faces wi
shotncuded mach ofthe occptal lobes (Figure)
Coronal Protocol
{A detailed description of the acquisition methods, ana-
tomlcal borders, and reliability used in our laboratory 10
‘examine the temporal lobes and hippocampus has been re-
ported! Summaries are presented below.
‘Acquisition Parameters. With this protocol, 2 conlguous
‘3-mrn-thick coronal images were acquired usingamuldecho,
‘Dow-compensatd, cardiac gad pulse sequence (echotime=40
‘and 0 milliseconds; effective repetition time, 2800 milisec-
‘onds) witha 24-cm field of view, l excitation, anda 256x256
‘matrix. The plane of mage acquisition was oriented perpen-