f. phd j. in brain phosphorus metabolite concentrations ... · rs large arger exo- uding

R S large arger exo- uding

borne D, rance of me- iJNR 1981;

n D . Com- ases of syriw

G. R a b

a1 and s id bms& db

;;;esg,nd

ed. London:

logy 1966; 86

aZEfi2

Dominique Sappey-Marinier, PhD Raymond F. Lkickem, MD George Fein, PhD GiovaMa Calabrese, MD BMO Hubescb, PhD Craig Van Dyke, MD William P. Dillon, MD Linda Davenport, PhD Dieter J. Meyerhoff, PhD Michael W. Weiner, MD

Alterations in Brain Phosphorus Metabolite Concentrations Associated with Areas of High Signal Intensity in White Matter at HR Imaging'

reas of high signal intensity in hite matter are identified on brain mgnetic resonance (MR) imaging udies in %%-XI% of elderly sub- cts. The authors used phospho- e31 MR spectroscopy to char- *&e the metabolic status of mispheric white matter brain vol- mes in 30 elderly subjects with hite matter areas of high signal in- msity at MR imaging. Compared ith white matter volumes with no or l i n i a l areas of high intensity, hite matter volumes with extensive lpas of high intensity evidenced a 6% decrease in the adenosine iphosphate (ATP)/inorganic phos- hate (Pi) ratio (P = .03) and a 21% ecrease in the ATP concentration

hanged. A pilot P-31 spectroscopic naging study in a subject with a we, coalescing white matter area of igh signal intensity demonstrated wge reductions in metabolite con- entrations in the high-signal-inten- ity area. These results suggest that KkWiVe white matter areas of high ignal intensity indicate a ptocess ut affects white matter cellular en- rgy metabolism.

= .05), with the Pi level un-

i , du tens: Aging, 10.83 Brain, MR, 11214 Brain, white matter, 10.83 Lkmen- a. 10.83 Magnetic r e s ~ n a n ~ e (MR), phospho- 1% studie-. 10.1214 - Magnetic resonance (MR), w o p y , 10.1214

&log?. 1992; 183247-256

AGNETIC resonance (MR) imag- M ing of the brain reveals the presence of areas of high signal intensity around ventricles and in deep white matter in 2596-5096 of unse- lected patients over the age of 60 years (1-10). Comparable phenomena, when observed with less sensitivity as areas of high attenuation at computed tomography (CT), have recently been given the name leukoaraiosis. These white matter findings are concentrated in the end and bor- der zones of arterial pfrfusion and have been hypothesized to indicate an underlying ischemic process- Di- rect evidence for an association between such findings and decreased cerebral blood flow comes from two recent studies. In one, xenon CT was used to demonstrate an association bemeen leukoaraiosis and blood flow reductions in subcortical white matter (11). In the other, MR imaging and oxygen-15 positron emission tomography were used to demonstrate glo- bal gray matter reductions in both cerebral blood flow and the ratio of cerebral blood flow to cerebral blood volume in patients with extensive white matter areas of high signal intensity (12). In the positron emission tomographic study, however, low signal-to-noise ratios prevented measurement of cerebral blood flow and the ratio of cerebral blood flow to cerebral blood volume in white matter.

There is overwhelming evidence that white matter areas of high signal intensity are more prevalent in eld-

'horn theMagwticResonanceUnit(D.SM,R.F.D.,C.C,B.H.,D.J.M,MW.W~)andPsyddatry (R.F.D., G.F., C-V.D.), Veterans Adminiition Medical Center, 4150 Clement St, San Fran-

SO. CA 94121; and the Departmentsof Radiology (DSM,G.C., B H , W.P.D., D.JM,M.W.W.), Idkine (DSM, B.H., M.W.W.), and Psychiatry (-9.. C.F., C.V.D., LD.), Univesity of California,

Francisco. Received May 1,1991; revision requested July 2; revision d v e d October 16; ac- Ped November 4. Supported by Philips Medical Systems; DSM supported by a grant fnwn the m h Ministry of Foreign A f f a i ~ ~ , RF.D. supported by the Veterans Administration psychiatrist esearrh Training program, C.V.D. supported by National Institutes of Health grant MLNS22029,

-tionMedical ReseamhSewice. A W neprintmpeatsloGF. CRSNA.1992

UW.W. supported by National lnstituks of Health p t ROl-DK33293 snd the Vekians M-

erly individuals with cerebrovascular disease or dementia than in n o d elderly persons. While there is sug- gestive evidence that such areas indicate a patk@ogic process that may lead to impaired cognition; the evidence for a clear cause-and-eff ect relationship is inconclusive. For example, Steingart et al(13,14) and Gupta et d(7) reported a relationship between white matter areas of high signal intensity and impaired cognition, while Hershey et a1 (8) failed to find such an association. When extensive white matter areas of high signal intensity are observed in demented individuals and Alzheimer or multiin- farct disease is ruled out at autopsy, the high-signal-intensity areas are usually assumed to have been re- sponsiile for the dementia and a Binswanger type of dementia or subcortical arteriosclerotic encephalopa- thy is diagnosed. At present, how-- ever, there is no direct evidence that these areas of high signal intensity represent the demential process. In a recent article, we showed that the presence of extensive white matter areas of high signal intensity of many years duration is not pathognomonic of cognitive impairment (15). These areas may indicate different processes in different individuals (eg, infarction, gliosis, demyelination, ventricular diverticula, brain cysts, or Virchow- Robin spaces). It is possible that the nature of the underlying process in combination with the location and extent of the white matter areas of high signal intensity determine the likelihood, magnitude, and quality of clinical manifestations.

Neuropathologic postmortem studies (10,1649) have shown that such

ATP = adenosine hiphos- . . phak, PCr = phosphocreatine, PDE = phos- phodkkr, F'i = horgank phosphate, PME =

VOI = volume of inkrest. phosphornonoester, SI = Spertmscoprc imapin&

areas or leukoaraiosis usually mpre- sents either frank infarction or incomplete infarction with a pattern of gliosis and demyelination in association with vascular hyalinization. A proportion of these findings, how- eveF, have no detectable associated disease. Recently, Englund et a1 (18) used MR imaging and neuropathologic methods to study white matter in patients with presumptive senile dementia of the Alzheimer type. They found that white matter areas of high signal intensity in such patients have increased T1 and T2, with the increase in relaxation times proportional to the severity of the white matter histologic changes (ie, partial loss of myelin, ax- ons, and oligodendroglial cells; mild reactive astrocytic gliosis; sparsely distributed macrophages; and arterio- lar stenosis resulting from hyaline fibrosis). The water content of the white matter was only slightly increased.

Phosphorus-31 MR spectroscopy is a valuable noninvasive technique for studying cerebral metabolism in vivo under both normal and pathologic conditions (20-23). This technique allows assessment of brain high-energy phosphate metabolism by mea- suring phosphorus metabolite levels (eg, adenosine triphosphate [ATP], phosphocreatine [PCr], and inorganic phosphate [Pi]) and tissue pH. Animal studies have shown that ischemia causes a reduction in the PCr/Pi and ATP/Pi ratios and a fall in pH owing to lactic acidosis. P-31 MR spectroscopy can also allow monitoring of phospholipid metabolism by measur- ing concentrations of phosphomono- esters (PMEs) and phosphodiesters (PDEs).

The purpose of this study was fo use P-31 MR spectroscopy to characterize cerebral oxidative and phospholipid metabolism in patients with white matter areas of high signal intensity at MR imaging. We used P-31 MR spectroscopy to test both the hypothesis that these areas have an ischemic basis, which would result in reductions in energy phosphate concentrations, and the hypothesis that they are associated with changes in white matter phospholipid metabolism, which would result in changes in PME and PDE concentrations.

SUBJECTS AND METHODS Subjects

The subjects were recruited from a larger longitudinal study of white matter areas of hi& sirma1 intensitv at MR imae-

Table 1 Nennmsvcho~c Test Performance

Tgt

Intart _. == xi)*

Mean SD r&n-Maxscore 229 2.85 0-11

28.n 1 s 2F30 74.n 3u 64-78

lllRl3m 94-m 106.42 11.6 87-132 111.77 1288 91-146

- ~~

Mean SD K i s c m

la0 0.82 (F.2 19.00 4.76 14-24 50.00 8.00 4258

T925 18.96 66-106 8125 5.85 7588

4285 330

23.% 7.03 aD.% BM

a.69 336 5.42 3.78

5c85 7.02

37-18 10.94 97.51 36.07

3e-18

6-37 538

3-17 043

24-60

19-64 58-m

2275 11.03

275 275 0 0

425 1.26 2.00.- 1.63

23M'zZ.3!l

9.75 3721 32450 117.73

ing in demented and nondemented elderly individuals with and without hypertension. The larger study had a sample of about 100 subjects, all of whom had undergone an initial complete =weighted spin-echo MR imaging study. Subjects were selected for this pilot study on the basis of patient demographics and our neuroradiologist's (W.P.D.) grading of the

,extent of the white matter areas of high signal intensity on the initial MR imaging study. That grading was performed by using a fivepoint scale that we have p u b lished (3) and that has been used by others (4). On this scale, a score of 0 indicates no white matter areas of high signal intensity; 1, involvement limited to the tips of the frontal horns of the lateral ventrick; 2, small white matter areas of high signal intensity in the subependymal or subcortical region; 3, extensive subependymal and d m t e separate white matter areas of high signal intensity; and 4, foci that are large and coalescing.

On the initial MR imaging study, half of the sample had moderate or extensive white matter areas of high signal intensity (grade 2 or greater) and half had normal MR images with regard to the white matter (ie, grade 0-1, indicating either no highsignalintensity areas or high-signal- intensity areas limited to the tips of the frontal horns of the la ted ventricles), with the two subsamples comparable with regard to the distribution of age and sex. S i the initial MR imaging studies were performed up to 2 years before the current MR imaging and spectroscopic study, some of the s u b i i had more extensive

white matter areas of high signal intensit) than on their initial MR images. Thirty subjects (20 men and 10 women; age. 67 6 years 2 3.2 [mean -r- standard deviation!) partiapated in this study. The subjects had no documented evidence of c e r e b vascular disease; neurologic, endocrine, m respiratory disorders; or major psychiatric illness (schizophrenia, major affectil e dls- order, organic mental disorder, or alcohol or other substance abuse disorder). >even teen of the patients were hypertenqive. On the MR images obtained in coni unc- tion with the MR spectroscopic studv (see below), there was no association between the presence of hypertension and the extent of white matter areas of high signal intensity. As part of the larger study, all sublech

had undergone neurologic and neuropst- chologic assessments. The neurologic examination included assessment of < r a d nerve function, motor tone and s t n ngth. sensory systems, coordination, gait, and reflexes. None of the neurologic examina- tions revealed any significant abnonnah- ties.

Neuropsychologic testing included measurement of general intelligence, memory, language, and visuospatial and psychomotor processes (see Table 1 for a list of the specilk tests). Two neurcysy- drologic screening tests, the Mini-hlental State Examination (25) and the Neurobe- havioral Cognitive Status Examination (26), were used to classify subjects as demented or nondemented. Because of OW desire to minimize false-positive results, subiects were dassif~ed as demented if an:

lqglue1. L spheric VOI matter VOI I

only if they 3fortheN and 64 for t Status Exan dassified as patients me ad Statisticb Three of thc matter area other did nc jubjects wa them had a of greater tl

Table 1 p

best scores f demented s

16 for their who had sc one measur bad two scc h impaire ~ures, excep the Boston I

performanc mon for urn Me 1 shows mmce beh b m t e d s AU proceq

Committee University ( jent was ob hey partici consent wa inembers of mpairmeni

i k k t i O W ,

unpaired gI

MR studi MMRei

with a Gyrc b o p y t( %ms Nc Opaating a

Placec

0-4 3-51

54-145 250-500

&Mental *Adult

- ?moly scale,

‘2 years 2

1; min-max

a1 intensity

1; age, 67.6 leviation]) subjects if cerebro- idoaine, or psychiatric fective dis- or alcohol ler). Seven- tensive. conjunc- study (see I between id the ex- ;h signal

I subjects neuropsy- logic ex- of cranial strength, ;sit, and . examina- mormali-

iuded ence, #a tial and de 1 for a uropsy- ii-Mental \leurobe- nation :ts as de- se of our

* results, nted if and

.Thirty

A l l IPM

1. MR images illustrating the hemispheric VOl of a subject with exteWive white matter areas of high signal intensity (left), the hemi- m a i c VO1 of the a m and sex-matched s u m without white matter areas of high signal intensity (center),and the bilateral superior white

.lly if they had scores below the cutoffs of 5 for the Mini-Mental State Examination and 64 tor the Neurobehavioral Cognitive jtatus E <adnation. Four patients were dassified as demented, and all of these patient. met the criteria in the Diagnostic ad Stni:stiarl Manual (33) for dementia. Three nf the demented subjects had white matter areas of high signal intensity; the other did not Only one of the demented Nbprt. was hypertensive, and none of them had a Hachinski Ischemic Score (34) of p a ter than 4.

Tabk 1 presents the means, standard dmiations, and range of neuropsychologic test scores for the demented and the nondemented subjects. The subjects in the un- unpaired p u p scored within normal lim- ~h for their age, except for nine sd@cb who had scores in the impaired range on one measure each and one subject who had two scores in the impaired range. All the impaired scores were on memory measures, except for one impaired score on the Boston Naming Test. s pattern of

mon for unimpaired elderly subjects. Ta- ble l shows the minimal overlap in performance between the unimpaired and the demented subjects.

All procedures were approved by the Committee on Human Research of the University of California, and written mn- sent was obtained from all subjects before they participated in the study. Written consent was also obtained from family members of participants with cognitive impairments.

performance on a battery ? o tests is com-

MR Studies

All MR experiments were performed with a Gyroscan MR imaging/MR spec- hDscopy 20-T system (F’hilips Medical Systems North America, Shelton, CONI) operating at 34.79 M H z for P-31. Subjects were placed m a supine position with

their head inside the P-31 spectroscopy head mil, which was within a hydrogen-1 imaging head coil placed orthogonally to it.

MR imaging and single-wlume image-se- k t e d in uiw MR spech.oscopy.-At the be- ginning of the MR session, all patients underwent nweighted multisection axial MR imaging of the brain (repetition time msec/echo time msec = 2,000/30,60; 5-mm section thickness; 2 5 m m intersec- tion gap) to determine the presence, location, and extent of white matter areas of high signal intensity at the time of the MR spectroscopic study. This MR image was used by the spectroscopist (DSM., R.F.D., G.C., B.H.) in selecting the volume of in- terest (VOI) for the MR spectroscopic study. A standard VOI (8 x 3 x 3 an) was placed by the spectroscopist in the hemispheric white matter adjacent to the ventricles (see Fig 1). In subjects with extensive white matter areas of high signal intensity apparent to the spectroscopist on the MR images (which indudes all subjects who were later rated by the neuroradiologist as having grade 3 or 4 white matter areas and some who were rated as having grade 2 areas), the VOl was placed in the hemisphere wherein the spectmsm pist identified the most extensive white matter areas of high signal intensity (Fig 1, left). In subjects in whom extensive white matter areas of high signal intensity were not apparent to the spectmsmpist (which includes all subjects who were later rated by the neuroradiologist as having grade 0 or 1 white matter areas and some who were rated as having grade 2 areas), the VOl was placed in one or the other hemisphere such that the dishibution of the hemisphere studied would be the same for subjects with extensive and with absent white matter areas of high signal intensity (see Fig 1, center, for a representative VOI from a subject without white matter areas of high signal intensity). Eight of the VOIs

. .

were in the left hemisphere, and 22 were in the right hemisphere. After the MR studies were complete, our neuroradiologist read the current MR images and graded the extent of the white matter areas of high signal intensity on the same five-point scale as was used for initial subject selection. This grading was performed before the MR spectra were processed and was therefore blind with respect to the spectroscopic results.

For the final 13 subjects studied, MR spectroscopy was performed on a second, bilateral, primarily white matter VOI (8 x 5 x 25 an) in addition to the hemispheric white matter VO1 described above. This second VOI was chosen to be in an area in which white matter areas of high signal intensity are seldom evident. None of these 13 subjects evidenced white matter areas of high signal intensity in this VOI, so these data can be used to represent a normal-appearing white matter control VOI. As displayed in Figure 1 (right side), this VOI encompasses the cen- tra semiovale bilaterally, bordered inferi- orly by the inferior margin of the corpus callosum and laterally and superiorly by the inner margin of the cortical mantle. The cingulate gyri bilaterally were included as-contaminating gray matter structures, and the cerebrospinal fluid of the interhemispheric fissures occupied a small proportion of this VOI as well.

The correspondence between the spectroscopic VOI and the white matter areas of high signal intensity was complicated by the “chemical shift offset” phenomenon (35). Because of the magnetic field gradients during radio-frequency pulsing in MR spectroscopic volume selection, the actual VO1 studied diffendslightly for each metabolite, depending on the difference between the resonance frequency of the metabolite and the carrier frequency used in the MR spectroscopic study. The spatial displacement between the PME

Rdioloev 249

and B-ATP resonances at the outer bor- ders of the spectroscopy VOI was 17 mm in all directions The nominal VOI, which the spedmzopist set for the study, was defined for the MR spectroscopic carrier frequency, which was very dose to the mwmnce of PCr. For extensive white matter areas of high signal intensity (ie, grade M), even with this metabolite-spe- cific spatial displacement of the VOI, sub s- white matter areas of high signal inteiwiity remained in the hemispheric spectroscopic VOI for each metabolite. For grade 2 white matter areas of high signal infensity, in which category there were a &ble number of small lesions, this VOI dkphcement made it difficult if not im- *le to establish whether the hemi- sp+eric VOIs for the various metabolites contained any white matter areas of high signal intensity.

For MR spechoxopy, the magnet was shimmed on the whole-head H-1 signal of water, with use of the H-1 imaging head &I, to a linewidth of less than 15 Hz. For P-31 spectroscopy, a 16.5un-diameter Helmholtz head coil (ie, closely coupled),

separation) orthogonal to the & field, was tuned and a W-exatation one-pulse spec- truinof the calibration standard (seebe- low) was recorded. Finally, P-31 spectra of the VOI were acquired for 640 data acqui- sitions with a 90” exatation and a repeti- -time of 1.75 seconds by usinganim- pxwed image-guided threedimensionally localized pulse sequence (ie, image-selected in vivo spectroscopy in combination with a postacquisition saturating pulse 136n. To obtain molar concentrations, we

used a p’pnriously published quantitative n\ethod (37). For calibration, a 3-mL external reference of hexamethylphosphorous hiamide ( S i p Chemical, St Louis) was positioned in the center of one of the loops of the P-31 coil. It was used to con- firm coil position at MR imaging and to determine the 90” pulse length at the center of the VOI by using a computer-gener- ated B, field plot (35). This standard w q also used to take into account differences in coil loading between each patient and the phantom The P-31 MR spectra from the VOI and the external standard (hexamethylphosphorous triamide) were compared with one P-31 spectrum of a VOI obtained in a large Pi phantom of known concentration with the same external standard.

The VOI, Pi phantom, and hexamethylphosphorous triamide spectra were all processed by using NMRl data processing software (New Methods Research, Syra- cuse, NY) on a Sun 3/60 computer (Sun Microsystem, Mountain View, Calif). The broad signal from the less mobile metabolites was first removed by convolution difference (exponential broadening, 150 Hz; convolution factor, 0.9). and the resulting free induction decay was apodized with an exponential l i e broadening factor of 7 Hz. After Fourier transformation, a base- line flattening procedure was applied. In-

cOmpOSed of two circular loops (16-cm

J. .I -1 -r---1: _.--- 1 _-._^-- rL-- A

tained from least-squares-mwe fitted spectra and were used for calculation of metabolite concentrations and their ratios.

Computer simulations were used to adjust for variations in VOI position relative to the P-31 head coil and Tl saturation and frequency offset effects (38). such that calibration with a large Pi phantom was not required with each experiment. Tl relaxation times, which we had previously measured in typical subjects (B), were used to adjust for partial Tl signal saturation (Tl[PME] = 1.7 seconds, Tl[Pi] = 1.4 seconds, Tl[PDE] = 1 3 seconds, Tl[PCr] = 27 seconds, Tl[y-ATpl= 0.6 seconds, Tl[a-ATP] = 1.0 seconds, and Tl[&ATPJ = 0.7 seconds). A normal white matter water content of 72% (40) was a s sumed in calculating absolute metabolite concentrations. Ratios of ATP to Pi and PCr were computed by using the B-ATP resonance. The &ATP resonance best rep resents the ATP concentration in tissue because, unlike the 7 and a reso~11ces, it is free from signal conhibution from other phosphate-containingmetabolites such as adenosine diphosphate and reduced nico- tinamide adenine dinucleotide phos phates (41). The pH was calculated from the chemical shift of Pi referenced to the chemical shift of PCr (42).

The comparison groups for statistical analysis were constructed on the basis of the extent of white matter areas of high signal intensity in the spectroscopic VOL This was accomplished by examining the neuroradiologist‘s grading of the extent of the white matter areas of high signal intensity together with hard copies of the MR image on which the hemispheric spectroscopic VOI was displayed. We constructed three comparison groups consisting of subjects with no, minimal, or extensive white matter areas of high signal intensity in the spectroscopic VOL Twelve patients, all of whom had whole-brain grades of 0-1, clearly had no white matter areas of high & p a l intensity in the spectroscopic VOI. Twelve other patients, all of whom had wholeham . gradesof3-4, had extensive white matter areas of high signal intensity encompassing at least 20% of the spectroscopic VOI. The combination of the chemical shift offset phenomenon and the variability in the distribution of white matter areas of high signal intensity throughout the white matter made it im- possiMe to subdivide this group into finer gradations of the proportion of the spec- h.oscopic VOI containing white matter

subsamples ended up having the same sex distributions and were closely matched in terms of age. In the remaining six patients (all of whom had grade 2 white matter areas of high signal intensity consisting d small areas of high signal intensity in the subependymal or subcortical region), the degree to which the hemispheric spectroscopic VOI encompassed white matter areas of high signal intensity was highly variable- In four of these patients, there were a few relatively small white matter areas of high signal intensity in the nomi-

areas of high signal intensity. These two

--I -&-4r V n I 1- she nthor h n

there were no white matter areas of high signal intensity in the nominal spectre -pic VOI, but there were white matter areas of high signal intensity adjacent to the spectroscopic VOI. As noted above, the chemical shift offset phenomenon made it likely that, in subjects with grade 2 white matter areas of high signal intensity, there were variable numbers (iclud- ing zero) of white matter areas of high signal intensity in the hemispheric VOI for the different metabolites.

The statistical analysis used onewed t tests to compare metabolite ratios between the subsamples with no white matter areas of high signal intensity in the spectroscopic VOI and the subsample with ex tensive white matter areas of high signal intensity in the spectroscopic VOL The data from the subsample with minimal white matter areas of high signal inten-ity in the spectroscopic VOI was not included in the analysis because it contained no iri- formation to allow distinctionbetween the null hypothesis of no difference among the subsamples and the experimental hypothesis of a monotonic change in metab olite measures with inaeasing extent of white matter areas of high signal intensit). in the specbroscopic VOI (43).

Pilot P-31 MR spchmmpic imaging (51) study.-We performed a P-31 SI study on two of the patients from the above shdy, one with grade 4 and one with grade i) white matter areas of high signal intemity The SI procedures and experimental na- rameters are described in detail elsew here (44). In brief, the subjects underwent a P-31 SI examination immediately after standard H-1 MR imaging. P-31 SI with phase encoding in all three dimensions 6 conceptionally a combination of P-31 MR. imaging and localized P-31 MR S e i +

scopic experiments. It provides both coarse maps of phosphorus metabolite dishiiutions and spectra from localized volumes anywhere in the brain. Exprri- mental parameters included 922 phabe- encoding steps over a 270-mm3 field of view, an echo time of 35 msec, a repetition time of 350 msec, and eight data ac- quisitions averaged to give an acquisition time of 43 minutes and a nominal vovel volume of 11.4 an’.

RESULTS Figure 1 illustrates the MR spectro-

sconic volumes studied. On the left,

PDE

-2. P-31 mnances fror IDES and PME

TabIe 2 P-31 Mehbo MatterAneu

PME Pi PDE Pcr

PH

studied in tk 2 displays tt. md-from tt

a n &a1 =weighted image displays the lateralized VOI studied in a subject with extensive white matter areas of high signal intensity. The axial n- weighted image in the center displays the lateralized VOI studied in the subject without white matter areas high signal intensity who was age- and sex-matched to the subject whw image is displayed in Figure 1, left. The TI-weighted sagittal image on the right illustrates the biiteral s u p rinr white matter vnliime that was

* 1. The sped ldentificatio

* from ATP (1 ppm, and p h e is &

f ence at 0 pp ppm repres phdipid mt two large re mE and Pb

, at 27 and 6. One subject lpade 2 whi

IS of high .peCtro- e matt- jacent to above,

lenon ith grade \al inten- 3 (includ. If high sig- VOI for

le-tailed t I between atter ar- 2 spec&& ith exten- ignal 11. The inimal intensity : included ed no in- tween the among -tal hy- n metab k n t of intensity

ing (SI) tudy on 'e study, ade U intensity. ital pa- lsewhere ent a after I with lsions ia '-31 MR ectro- lth mlite alized xperi- * h a s Id of epeti- ~ t a ac- uisition voxel

3ectro- I left,

lsub !r areas ialm

he reas of w- whose left. *on

;PbYS

lisplays

VaS

Pcr Pcr pa

ATP

-r:r r:r

PDE b

m.a

PDE

. ATP

WM CRI ml i@tx 2. P-31 MR spectra from the VOIs displayed in Figure 1. The spectra show the three resonances from ATP (y-, a-, and &ATP) and the wmanct's from PCr and Pi. The phospholipid re cursors and/or breakdown produ& are identified by two large resonances representing - . - - PMS and PMEs.

hti0.. 1.9 0.2 . ui a2- 1.4 ai m

03 33 0.1 25 02 25 PCr'pAl'F' 15 02 2 2 0 3 2 . 0 0 3 m PME 3.1 03 4.8 0.7 3.2 0.4 A7 Pi 1.7 02 15 a2 l.6 02 .40 PDE 95 0.6 .11.1 05 88 07 23 PCr u) 0.4 c I a 6 3 . 8 0 3 -40 B A T 2.8 03 2 2 0 2 2 2 Q 2 m

PH 7m -om 7m am 780 O M m

BAPm 2, PCr Pi

Concentrafbm

Note -SE = standad emr. * O K ~ l g f O V ~ i t h f h Y d - q s W . n d H a r r s r t n t ~ l k . r=5&

rbespc- 'DiffeEncesbtiptir;rlty~

studied in the final 13 subjects. Figure 2 displays the P-31 MR spectra gath- ered from the three volumes in Figure 1. The spectra in Figure 2 allow ready identification of the three resonances from ATP (y = -27 ppm, a = -7.1 ppm, and B = -16.1 ppm). The PCr line is used as a chemical shift reference at 0 ppm, and the Pi peak at 4.76 pprn represents a pH of 6.97. Phos- pholipid metabolites are identified by two large resonan- representing PDE and PME compounds centered at 27 and 6.3 ppm, respectively. For one subject (a demented subject with grade 2 white matter areas of high

signal intensity), the PME peak was very s m d and the curve-fitting software could not reliably estimate the PME and Pi areas; therefore, concentrations or ratios dependent on PME or Pi measurements could not be calculated for that subject.

Imagdected in Vivo SJeCtmmpc Stl?dieSof

ermsphenc Wte Matter Table 2 presents metabolite ratios

and concentratioqs from hemispheric white matter VOIS that were free of white matter areas of high signal in-

tensity, contained minimal white matter areas of high signal intensity (ie, a small number of discrete white matter areas of high signal intensity), or contained extensive, coalescing white matter areas of high signal intensity. The ratio measures were used to test the a priori hypotheses of decreased metabolism as set forth in the intro- duction. The primary finding was a 26% decrease in the AW/R ratio and an associated 21% decrease in the - ATP concentration in subjects whose spectroscopic VOIs contained extensive white matter areas of high signal intensity compared with subjects whose spectroscopic VOIS were free of white matter areas of high signal intensity (P = .03 and .OS, respectively). As shown in Figure 3, the distributions of the ATPIPi ratios dif- fered depending on the extent of white matter areas of high signal intensity in the Spectroscopic VOI. The median ATPIPi ratio for the no white matter areas of high signal intensity group is in the upper quartile of the distribution of ATPIPi for the extensive white matter areas of high signal intensity group, and the median ATPIR ratio for the extensive wh;te matter areas of high signal intensity group is in the lower quartile of the distribution of AW/R for the no white matter areas of high signal intensity group. Associated with the presence of extensive white matter areas of high signal intensity were statistical trends toward an increase in the pcr/ATp ratio (33% increase,

2.5 0

0

.- e

2.0 - .- Q (-1

1.5 a t- 6

- 8

0 0

0 0 8

0 0

0

0

e A -

0.5 I No Small Extensive

WMSHs Discrete WMSHs WMSHs

Figure 3. Dot plot illustrating the distribution of the ATP/R ratios measured from the hemispheric white matter VOI (eg, Fig 1, left and center) for patients with extensive (n = 12). small and discrete (n = S), and no (n = 12) white matter areas of high signal intensity (WSMHs). The category of patients with no white matter areas of high signal intensity includes patients with white matter areas of high signal intensity limited tu the tips of the frontal horns of the lateral ventricles; this is thought to be a nonpathologic phenomenon in the n o d aging brain.

P = -07) and a decrease from 7.08 to 7.00 in intracellular pH (P = .On.

Image-selected in Vivo Spectrmco ic Studies of Bilateral .

Table 3 presents metabolite concen- Superior rf;l ite Matter

trations and ratios and pH from the bilateral superior white matter VOIs, in which no white matter areas of high signal intensity were evident for any subject. Although the number of subjects studied was small (n = 6 for grades 0-1, n = 3 for grade 2, and n = 4 for grades 3 4 ) and none of the comparisons met the P < -05 criterion for statistical significance, there were trends in the MR spectroscopic data that were consistent with the data from the hemispheric VOIs. For this VOI, subjects with a 3 4 grade at MR imaging evidenced (a) a 24% lower ATP/Pi ratio (P = .15), (b) a 14% lower ATP concentration (P = -24). and (c) a decrease from 7.19 to 7.08 in intracellular pH (P = .07) when compared with subjects with a grade of 0-1.

P-31 SI Pilot Studies Figures 4 and 5 display the results

from the subject with grade 4 white matter areas of high signal intensity. A T2-weighted axial H-1 MR image obtained through the large left hemispheric white matter area of high sig-

Mean SE Mean SE Mean SE High-IntensityAm~

1 3 0.3 3.7 0.3 1 3 0.1 2.8 0.3 3.7 03 2.4 02 1.8 02 w) 03 1.8 0.1

3.7 02 3.6 0.7 2.7 0.4 12 0.1 0.9 a0 13 0.1 8.0 0.3 9.1 0.9 7.1 0.6 33 0.2 3.4 0.2 3.1 0.3 22 0.1 2.0 . 0.2 1.9 0.1 7.19 0.S 7.04 0.02 7.011 0.03

.lS

.15

.a

.19 23 .L? 33 24 .w

Nate.-sE=stmdaFdam.

of Figure 4. The right side of Figure 4 displays the P-31 spectroscopic image from the axial 2 . m - t h i c k plane centered around the MR imaging section. The spectroscopic image, which was interpolated from 12 x 12 x 12 to 64 x 64 x 64 resolution, displays the integrated signal intensities of all phosphorus metabolites. The spatial correlation of the H-1 and P-31 images is facilitated by the superimposition of the edge-detected H-1 h4R image on the P-31 spectroscopic image. On the P-31 spectroscopic image, red represents the highest P-31 signal intensity while blue represents the lowest intensity, with green as an inter-

matter area of high signal intensity is characterized by a metabolite deficit relative to the contralateral side. This is also evident in the image profile plotted at the bottom of Figure 4. This profile displays the total P-31 m e t a b lite signal integral along the straight line perpendicular to the midline, through the middle of the white matter area of high signal intensity. This degree of bilateral asymmetry was absent in the subject with grade 0 white matter areas of high signal intensity and in normal (though younger) subjects studied in this l a b ratory. Figure 5 shows two localized P-31 MR spectra obtained from the same SI data set as used in Figure 4. Both spectra were obtained from tissue volumes of about 27 m3, one centered on the white matter area of high signal intensity and the other from the corresponding area on the contralateral side. While both spectra show the characteristic resonances for cerebral phosphorus metabolites, the

. mediate-intensity level. The white

the white matter area of high s i p 1 intensity (bottom) has less signal strength than the spectrum obtained from the contralateral region. There was a 30% decrease in ATP signal intensity in the white matter area of high signal intensity relative to the contralateral region. This is consisten with the finding of a somewhat smaller reduction in ATP concentration (directly p r o p o r t i ~ ~ l to intensity) when larger, more heteroge- neous tissue volumes are sampled, a: in the image-selected in vivo sprctro- scopic experiments described above.

DISCUSSION In vivo P-31 MR spectroscop~ re-

vealed a reduced white matter ATP/Pi metabolite ratio associated with extensive white matter areas ot high signal intensity in the elderly in this study. A 26% reduction in the ATP/R ratio and a 21% reduction in the ATP concentration were found ir h4R spectroscopic volumes that contained substantial white matter area- of high signal intensity compared with comparable volumes from sub jects with minimal or no white mattc areas of high signal intensity at MR imaging. These results suggest the possibility of altered energy phosphate metabolism associated with e\- tensive white matter areas of high signal intensity. To our knowledge, these results are the first to demonstrate in vivo brain metabolic alterations in individuals with white matter areas of high signal intensity. We did not find any statistically si@ cant daerences or trends for the Con centrations of the phospholipid me

r- I_ --

Figure4. Di matter areas c benuspheric i xned from tl 'he left. The s

3 f the edge-d displays the L! :he highest in ?laying the tc mdline, throi

!o be associ aatter area ilso, the co the spectro changed in white matte tensity corn none.

Numerot have show] causes a fa1 rise in the 11 ploduct Pi, p'obably d1 mefhanism me rnetabo ATP/PI anc nificant cha clear. One 1 d metaboli *thin the 1 Pled with A mpk, it is population glia), one w .-

-

MI

ned ere 1 in- If

le stent

ha- n-

i, as *o- ve.

e-

1 of

I in

in i in n- ?as

b tter 7

ex-

,

at- 7e

In- c

0

---_LI--_ - - - _ --- .

F w 4. Display of data from the P-31 SI study of a patient with a large, coalescing white matter areas of high signal intensity. Left, H-1 MR image obtained at the level of a large left hemispheric white matter area of high signal intensity. Right, P-31 spectroscopic image obtained from the axial 225cm-thick plane centered around the MR imaging section depicted at the left The spatial melation of the H-1 and P-31 images is facilitated by the superimposition of thc rdgedetected H-1 MR image on the P-31 spectroscopic image. The spectroscopic image displs 1 s the integrated signal intensities of all phosphorus metabolites, with red representing the highest intensities while blue represents the lowest intensities. Bottom, image profile dis- phyinc the total P-31 metabolite signal integral along the straight line perpendicular to the mid1ir.e. through the middle of the white matter area of high signal intensity.

to be associated with extent of white matter areas of high signal intensity. Also, the concentrations of PCr from the spectroscopic VOI were un- changed in patients with extensive white matter areas of high signal intensity compared with patients with none

Numerous studies in animal brain have shown that acute ischemia causes a fall in PCr and ATP levels, a rise in the level of the ATP hydrolysis product Pi, and a fall in pH (a,&), probably due to lactic acidosis. The mechanism by which changes in tissue metabolism might decrease ATPlPi and ATP level without a significant change in PCr level is not clear. One possibility is heterogeneity of metabolic change in different cells within the large volume of tissue sampled with MR specmscopy. For example, it is possible that there are two populations of cells (eg, neurons and glia), one with high levels of PCr and ATP and another with low levels of

PCr and high levels of ATP. Selective loss of the second population of cells would tend to cause a fall in the ATP level without a change in the PCr level. A second possibility is that chronic ischemia produces a different type of metabolic change than acute ischemia; chronic ischemia might cause depletion of the total adenine nucleotide pool (for example, by acti- vation of adenosine monophosphate deaminase [4q) while the creatine kinase reaction remains in equilib- rium, maintaining the PCr leveL A third possr’ibity is that changes in the ATPIPi ratio do nof represent changes in metabolic concentrations but are due to alterations in T1. How- ever, since the Tls of ATP and Pi are in the same range, it is unlikely that the T1 of ATP would be selectively altered without changes in the T1 of Pi. Unfortunately, selective alterations in the Tls of PG; ATP, or pi might mask metabolic changes in the PCr/ ATP and PCrlPi ratios. Fw, it is

10 0 -1 0 -2 0 PPY

F i 5. P-31 MR spectra obtained from the same SI data set used in Figure 4. Both spectra were obtained from tissue volumes of about 27 an’, one centered on the white matter area of high signal intensity (DWML) and the other from.?he corresponding area on the contralateral side (Contml.).

possible that the alteration in the ATPlPi ratio is not due to ischemia or incomplete infarction but rather rep resents changes in response to other pathologic processes associated with the white matter disease. Our current work in patients with

chronic stroke (48) reveals both much larger effects and a different pattern of effects compared with those reported here for white matter areas of high signal intensity. Chronic stroke is characterized by an increased brain pH, an increased PCrIATP ratio, and a decrease in PME, PDE, and-ATP - concentrations. The reduction in the ATP/R ratio associated with white matter areas of high signal intensity occurs in the absence of reductions in the PCrlPi and PCr/ATP ratios and thus is not consistent with the tradi- tional metabolic changes that are known to characterize ischemia and infarction. The difference in the pattern of P-31 metabolite concentration and ratio effects in acute and chronic stroke compared with those in white matter areasof hi& signal intensity may imply a different neuropathologic process underlying these phenomena. It is also possible, however, that the white matter areas of high signal intensity may represent either a benign or a morbid process in different subjects (eg, Virchow-Robin spaces vs ischemia) (49); if so, such variability in the underlying phenomena would lessen the sensitivity for detecting all the sequelae of the more morbid phenomena.

Differences in vasculature between gray and white matter may underlie a selective vulnerability of white matter to ischemia (10). The deep white mat-

ter is supplied by long, minimall Y branched penetrating arteries arisiig from the large arteries at the base of the brain. Because of this, the deep white matter is less well vascularized than other regions of the brain and is without luxuriant capillary networks with anastomotic potential. In ba- boons, it has been shown that the white matter possesses a less sensitive and effective regulatory blood flow mechanism than does the gray matter. In these studies, reduction of per- fusion pressure secondary to exsan- guination or via an increase in intracranial pressure resulted in zero blood flow in white matter while per- fusion continued in gray matter (5021). Lindenberg (52) suggested that when right-side cardiac failure results in venous stasis, the white matter is preferentially affected.

Given that white matter areas of high signal intensity are assodated with changes in energy phosphate metabolism in the white matter VOI with extensive white matter areas of high signal intensity, it is possible that the metabolic changes are due to metabolic changes in the foci of the hy- perintense areas within the VOIs and/or that the metabolic changes are due to a process that affects all the white matter within the VOI and even extends to white matter outside the VOI. In this regard, the results of. the positron emission tomographic study by Meguro and colleagues (12) suggest that metabolic changes may possibly extend even to the gray matter. As our first look at this auestion. -

essarily contains a mixture of gray matter, white matter, and a variety of cell types The second limitation is the chemical shift offset phenomenon (35), which occurs with all localization techniques in which radio-frequency pulses are used in the presence of section-select gradients, resulting in sig- nals of different metabolites arising from different, partiaUy overlapping tissue regiom thus, the tissue pmvid- ing the ATP signal is not exactly the same as the tissue providing the Pi si+ This phenomenon can lead to spurious data, especially if signal contamination from tissues outside the brain (eg, skeletal muscle) is allowed to occur. SI with phase encoding after signal excitation does not have this problem. However, because the SI technique uses a point-spread function to select the VOI, the outer VOI contamination is increased by the side lobes of the point-spread function. To reduce this effect, a Hammm gspatial filter was used, which however increased the VOI size. A third problem concerns the effects of relaxation times. Under conditions of rapid repetition for effiaent signal acquisition; partial saturation modulates the signal intensity as a function of Tl. Alter- ation of T1 would change signal intensity and the calculated molar concentrations and metabolite ratios. Finany, intracellular tissue pH was calculated on the assumption that the dissociation constant of phosphoric add determined in vitro applies to in vivo conditions. Alterations in concentrations of other ions, including . Na+, K*, and M2+, may ais0 cause we performed MR spectr&opy in a

bilateral suwrior white matter VOI in the final lfsubjects studied. These data, although preliminary, suggest that the metabolic changes are due to a process that affects all the white matter (ie, both inside and outside the VOI). Moreover, our pilot SI study illustrates the potential of that method for addressing questions such as this. The results from the SI study suggest that large, coalescing white matter areas of high signal intensity may be associated with a substantial focal reduction in the ATP/Pi ratio.

There are several limitations to the interpretation of these and other MR spedroscopic studies. First, P-31 MR spectroscopy has relatively low sensitivity, such that only substances in the millimolar range can be detected, and the spatial resolution is relatively poor. This is in comparison to H-1 h4R imaging, which allows detection of water and fat, both of which are in great abundance. Because of poor crutiinl mlirtinn. the sinele VOI nec-

shifts in the Pi peak, leiding to a spurious measurement of pH (53). How- ever, all of these problems (and other unstated assumptions discussed below and elsewhere [%]) are common to in vivo h4R spechmcopic studies.

A better understanding of the phenomena underlying white matter areas of high signal intensity is ex- tremely important. Although Alzheimer-type dementia and vascular dementia have been cast as etio- logically distinct, the picture has become more complicated in recent years as the possible contributing role of white matter ischemia to demential illness has become more fully appreci- ated. Even in patients with presumptive senile dementia of the Alzheimer type with no clinical evidence of cerebrovascular or cardiac disease, substantial proportions of patients are found to have white matter areas of high signal intensity. In postmortem studies, Englund et a1 (18) consis- tenth found white matter ischemia

with and without complete infarction in about 60% of patients with con- firmed Alzheimer disease, including sizable numbers of patients who had no clinical evidence of cerebrovascular disease. The most striking in vivo results suggesting the importance of white matter areas of high signal intensity in senile dementia of the Alzheimer type occured in the study by Wallin and coUeagues (19). After excluding patients with any- clinical evidence of a vascular contribution to their dementia, they found that 24 of 30 patients (80%) with presumptive senile dementia of the Alzheimer typt had ledcoaraiosis. There have also been reports that the m-occurrence a Alzheimer disease and leukoariaosis results in a more severe and more rapidly progressing dementia than is seen in patients with Alzheimer Xis- ease but without leukoariaosis.

In summary, there is overwhelming evidence in population studies that white matter areas of high signal intensity often represent a pathologic process, with increasing prevalence, severity, and extent of white matter areas of high signal intensity assodated with increasingly morbid out- comes. However, it is also clear that, in individual cases, the ochrrence OI white matter areas of high signal intensity, even when they are extmsivt is not pathognomonic of a morbid demential process. We need to increase our understanding of white matter areas of high signal intensky so that we can know when their p r s ence indicates a process that may con tribute to the development and 'or morbidity of dementia. Enterveiltion in such a process may have potential for lessening that morbidity.

This study also illustrates the im portance of quantitation methodol- ogy, including computer simulations as described in the Subjects and Methods section, and of image- guided, three-dimensional, localized P-31 h4R spectroscopy of human brain. The improved image-selected in vivo spectroscopic technique, which has been carefully studied in computer-simulated analyses and phantom studies (38), provides accu- rate localization by means of a well- defied delineation of the VOI. In addition, computer simulations a d F for sensitivity variations as a fundion of VOI position, T1 saturation, and chemical shift offset (38). These variations would not be taken into amoW if metabolite ratios were reported from peak integrals only.

There are a number of s i m p l i g assumptions that, as we have noted

xeviously :he c a l d a tions. Fint, within the mogeneou centration concentrat ences beh tion of tiss scopic VO: metabolite Second, th centration subjects fo which we] the VOIs i was neces study timt

study for I involve 8- magnet. T tissue wat the water jeds, conc fected; ho accountec metabolitc equally. T assumptic tion in thc suits. Ne\ clearly in1 in aspeck are assod of white I intensity.

In endi that in vi. in its infa can be ex hardwan and P-31 developn which ca lactate cc many otl vances w quantital through( voxels. V troxopy elucidati ohm ass cesses.

P-31 Tls. \

Referenct 1. Awad

U L C SiOnS

aging and E 1986;

I Brad1 Y&€ bicul a a r n ing. P 41.

ction I-

:! scu- vivo E of I in-

tUdY fter ical ion to 24 of tive r type Is0 ince of aosis

ian is - &-

elming that al in-

ence, atter &- out- . that, nce of la1 in- tensive, .bid in-

,bite mity !ii pres- ray con- dlor 2ntion ltential

.e im- d o l - llations \d

calked

! I d le, lied in and es a m - a well- 11. In ns adjusf function I, and se varia- I account htted

plifying 9 noted

're

logic

e-

ran

previously (3537-39), are involved in the calculation of molar concentrations. First, for quantitation, the tissue within the VOI is assumed to be homogeneous, with the measured concentrations representing the mean concentration within the VOI. Differ- ences between groups in the distnibu- tion of tissue types in the spectroscopic VOls could result in different metabolite concentrations and ratios. Second, the Tls used to calculate concentrations were derived from normal subjects for large brain volumes, which were less homogeneous than the VOIs in the current study. This wa- necessary because of the lengthy study times necessary to measure P-31 Tls. We estimated that a T1 study for the white matter VOI would involve 8-10 hours per subject in the magnet. Third, we assumed a uniform tissue water content across subjects. If the water content vaned among subjects, concentrations could be affected; however, this could not have accounted for our results, since all metabolites would have been affected equally. The use of these simphfymg assumptions argues for further cau- tion in the interpretation of our result>. Nevertheless, our results do dearly indicate that there are changes in aqxcts of the P-31 MR spectra that are associated with increasing extent of white matter areas of high signal intensity.

In ending, it is important to note that in vivo P-31 MR spectroscopy is in its infancy. Continuing advances can be expected in magnet desip, hardware, radio-frequency pulses, and P-31 SI (553). Together wifh the development of H-1 s-py, which can allow quantitation of brain lactate concentrations and those of many other metabolites, Fese advances will result in the ability to quantitate metabolite concentrations throughout the brain in eversmaller voxels. We believe in vivo MR spectroscopy shows great promise for the elucidation of changes in brain metabolism associated with disease pro- ce5ses. m

References 1. AwadIA,SpetzlerRF,HodakJA,A~ad

CA, Carey R Incidental subcoriical lesions i- on magnetir resona~eim- aging in the elderly. I. correhtian with age andrerebnrvascubrridcfactorr Stmke 1986; 171084-1089. Bradlq. WC, Wahlch V, Brant-zawaddd M, Yadley RA, Wycoff RR. Patchy, perken- t r i c u k r w h i t e m a t t e r ~ in the ammmonobsemationdudng NMRimag- i n g . N a r i n v P f i v e M e d ~ l % l % 41.

3. Brant-zlwadzki M, Fein G, Van Dyke C, KiemanR,DavenpatL,DeCrootJ. MR imagingoftheagingbrain:pakhywhite matter lesions and dementik AJNR 1- 63675682

4. Fazekas F, ChawhrkJB, Ahvi A, Hurtig HI,

at IS T in A M e i m e f s dementia and normal aging. AJNR 1987; WlZ126.

5. m e AE, DeLeon MJ, Gentes a, et al.

ZimmermanRA. MRsignalabnormah 'ties

Lehencephbpathy in normal and pathologic aging. I. a of blah lucendes. AJNR 1986; 7561-566.

6. Gerard G, Weisberg LA. MRI periventric- ukrlessonsin adults. Neurdogy 1986; 36: 996-1001.

7. Cupta !3t, Naheedy MH, Young JC, Cho- Lnial M, Rubino FA, Hind0 w. Periven- tricuhr white matter changes and demen- tkdinical r t i o n . Ardr Neurd 1988; 45.W-641.

8. Hershey LA, Modir MT, G m o u g h PG, Jaffe DF. Magnetic T~SOMIPX imaging in vascular dementia. Neurdogy 1967; 3729- 36.

laions in MR imaging of clinically healthy brains of elderly Slbpcts: posible p a h

10. Marshall VC, Bradley WG. Marshall CE, Bhoopat T, Rhodes RH. Deep white mat-

and histopathdogk findings. Radiology

11. KobariM,MeyerJ,khijoM. Leuko-araie sis, cerebd atrophy, and c?erebral perfu- sion in d a g i n g . Arch Neurd 1990;

12 Meguro K, Hatazawa J, Yamaguchi T, et al. Cerebraldrcubtionandoxygenmetabo-

ularhyperintensityasshmbymagnetk reso~llce imaging. Ann N d 1990; 28: 378-383.

13. steingart A, Hachin& VC, Lau C, et al. Cognitive and neumlogic findings in subjects with diffuse white matter lucenaies on computed tomographicxan e). Arch Neurol 1987; 4432-3!5-

14. Steingprt A, Hadrinslri VC, Lau C,etaL Cognitive and neuroiOgir findings in demented patients with diffuse white matter hrencies on computed tomographic scan (leukrwaiosis)). Arch Neuroll987; 44:s 39.

15. Fein G, Van Dyke C, Davenport C, et al. Pmmvation of normal cognitive function- ing in elderly subjects with extensive white-matterlesionsoflongduration.kch Gen psydriahy 1990; 47-223.

16. B r a h n B, Z%mmman R, Tmjanowski J, Gonatas N, Hickey W, Schhepkr W.

white-matter fod in the elderly. AJR 1988; 151559-566.

k i n dementia of the Atzheuner . type:a pathoanatomical study. Ann Neurol1986; 1 9 2 5 3 - ~

9, KirkpatrickJB,Haymanu whitematter

b&icbaSis. Jtadiology 1587; 162509-511.

terinf~.correlationofMRimaging

1988; 167517-522

47161-165.

~ ~ t e d w i t h ~ periVt?llhic-

&linMR:pathobgiccodationwith gXl6Sandhistopathdogy.lLHJp3ilIteIlSe

17. R U I A , E I I & ~ E . A ~ h i t e ~ ~ ~ ~ t t f f d i s o r -

18. EnglunaE,bA,persOnB. Correb- tiOnSbetWeenWhiteIMitterChangeSand protonMRrcbxationtimesindementia. Alzheima Dis Assoc Dhrd 1987; l(3):lSfs- 170.

19. WallinA,BlennowK.UhkmannC,Lang- stromG,GonfriesCC. Whitematterlow attenuation on amputed tomogmphy in Awlehnersdiseaseandvanulmdemen-

Actr Neurol Sand 1W, 80518-523. PA, Draw BP, smith IS.

QvonicodultEerebrslinfarrtionstudRd

t l : d . i a ~ a n d p ~ a S p e c &

20.

by phosphorus NMR spectroscopy. Radioi- ogy 1986; 1m.763-766.

21. segebarth C, Baleriaux D, Amold DL, Luyten FX, Den Hollander JA. MR image-guided P-31 MR spectmxopy in the evaluation of brain tumor treatment. Radi- ology 1987; 165.215-219.

22 Hubesch B, Sam-Marinia D, Hethering- ton HP, Twieg DB, Weiner MW. Clinical MRS studies of the brain. Invest Radid 1989; x1039-1042.

23. Weiner MW, Hetherington HP, Hubesch B, et aL Clinical MRS of brain, heart, liver, kidney. and cancec a quantitative a p proah. NMR Biomed 1989; 2290-297.

24. Williams JB. A strucQ~red interview guide for the Hamilton Depression Rating Scale. Arch Gen Psychiatry l-, 45.742-747.

25. Foktein MF, Folstein SE, Md-Iugh PR. Mini-mental state. J Psychiatr Res 1975; 111m-198.

26. Kernan RJ, Mueller J, Langston JW, Van Dyke C. The Neurobehavioral Cognitive Status Examination. Ann Intern Med 1987; 107:481-485.

27. Wechder D. WAIS-R manual. New Yo& Psychological Corp, 1981. -

28. Fuld PA. Fuld Ob~ect-Memory Evalua- tion. New Yo* Saul R. Korey Department of Neurdogy, Albert Einstein College of Medicine, 1977.

29. Denman S. Denman Neuropsychology Memory !kale. Charleston, 92 Denman, 1984.

xate for dmical use. J Psycho1 1945; 1987- 95.

31. Kaplan EF, coodgtass H, Weintraub S. The Boston Naming Test. Boston: E. Kap Ian and H. Goodglass, 1978.

32 Army Individual Test Battery. Manual of directions and scoring Washington, Dc: War Department, Adjutant General's Of- fice. 1944.

33. Diagnostic and statistical manual of mental disorders. 3rd revised ed. Washington, DC American Psychiattic h, 1987.

Alzheimer's dementia: multi-infarct dementia. In: Reisberg 8, ed. Alzheimefs disease: the standard reference. New Yo& Free Ress, 1983; 188-192.

35. Lawry TJ, Karamar GS, Weiner MW, Mat- son GB. Computer simulation of MRS loral*;ation techniques: an analysis of ISIS. Magn Reson Med 1909; 9299-314.

36. Matson GB, Twieg DB, Karczmar GS, et al. Application of image-guided surface coil

heart, and kidney. Radiology 1W; 169:

37. Roth K, H U M B, Meyerhoff DJ, et al. Noninvasive quantitation of phosphorus metabolites in human tissue by NMR spec- bmcopy. J Magn Reson 1989; 81299-311.

38. Matson GB, Hubesch 8, Lawry TJ, Meyer- hoff DJ, Sappey-Marinier D, Weiner WM. Quantificatim of ISIS iocalized in vivo MRS resulk (abstr). In: Book of abshacts: Society of Magnetic Resonance in Medicine 1989. Berkeley, calif: society of Magnetic Resonance in Medicine, 1989; 596.

Meyerhoff DJ, Matson GB, Weiner MW. P-31 MR spectroscopy of normal human brain and brain tumors. Radiology 1990; 1 7 4 : 4 0 1 ~ .

30. Wechsler D. A standardized memory

34. Hachinski VC. Differential diagnosis of

P-31 MR S-Y to h- liver,

541-547.

39. Hubesdr B, Sappey-Marinier D, Roth K,

40. Sappy-Marinier D. High resolution NMR ~pechmcopy Of cerrbral White mat- terin multiple sckrosis. Magn Reson Med 1990; 15229-239.

41. Clone$ T, Kopp SJ, Kot E, Pettegrew JW, HanisonWH,C~henMM. P-31NMR r e s o M n c e u u l y s i s o f b r a i n : t h e ~ C

acid extract spectrum J Neurochem 1982; Wl210-1219.

42 petrolf OAC, Richard JW, Behar KL, Alger JR, Den Hollander JA, Shulman RG. Cere- bral intracellular pH by "P NMR spectres cop^. Neurology 3.5781-788.

43. Miller R Beyond ANOVA. New Yo&

4. Hugg JW, Matson GB, Twieg DB, Mauds- ley AA, Sappey-Marinier D, Weiner MW. "P spectroscopic imaging of normal and pathological human brains. Magn Reson

45. Honkawa Y, Naruse S, Hirakawa K, Tanaka C, Nishikawa H, Watari H. In vivo studies of energy metabotim in experimental cerebral ischemia using topical magnetic resonance: changes in "P-nu- dear magnetic resonance spectra compared with electroencephalograms and ~ ~ @ o M I cerebral blood flow. J Cereb Blood Flow Metab 1985; 5235240.

46. Cermano IM, Pi% LH, Berry 1. De Armond S. High energy phosphate metabolism in experimental permanent focal cerebral ischemia: an in vivo "P magnetic resonance spPctroscopy study. J Cereb Blood Flow Metab 1988; 8-34-31.

wiley, 1986.

I-pg (in press).

47.

48.

49.

50.

51.

Monis RC Jr, Nigon K, Reed RB. Evi- dence that the severity of depletion of inorganic phosphate determines the severity ofdisturbance of adenine nudeotide me- tabolim in liver and renal cortex of the fructose-loaded rat. J Clin Invest 1978; 61: 209-220. Sappey-Marinier 0, Hubesch B, kicken R, Fein G, Matson CB, Weiner MW. Altered phosphateconcenhationsandpHinpa- tients with chmnic slmke and DWMLs (abstr). In: Book of abstrak: society of Mag- netic Resonance in Medicine 1989. Berkeley. Calif: Society of Magnetic Reso- name in Medicine, 1w; 1070. Braffman BH, Zimmerman RA, Trojan- owski JQ, Gonatas NK, Hickey WF, Schhepfer WW. &ai MR: pathologic correlation with gross and histopathology. I. Lacunar infarction and Virchow-Robin spaces. AJNR 1988; 9621-628. Romanul FCA. Selective infarction of gray or white matter caused by occlusion of large arteries. Clin Dev Med 1971; 39-40: 41-46. Symon I., Pasztor E, Dorsch NWC, Bran- ston Nu Physiological responses of local areas of the cerebral circulation in experimental primates determined by the method of hydmgen clearance. Stroke 1973; 4632442

52 LindenbergR PatternsofCNSvulnera- bility in acute hypoxanemia, including an. esthesia accidents. In: Shade JP, McMene mey WH, eds Selective vulnerability of % braii in hypoxanemia. oxford, Englanri: Blackwell !jcientific, 1963; 189-209.

53. Roberk JKM, Wade-Jardetzky N, Jardet+ 0. Inhacellular pH measurements by ''P nuclear magnetic resonance: influence of haors other than pH on "P chemical shifts. Biochemistry 1981; 205389-5394.

54. Matson GB, Weiner MW. Magnetic r e m MINX Spectroscopy. h Bradley w, stark D, eds. Magmtic resonance imaging. St Louis: M&y, 1% 438466.-

chemical shift imaging in 3 dimension>. Proc Natl Acad Sci U S A 1982; 79:3523- 3526.

56. Maudsley AA, Twieg DB, Sappey-Marinh D, et al. Spin echo "P speamscopic imag ing in the human brain Magn Reson 41ed 1990; 14:415422

55. &own TR, Kincaid BM, Ugu&ii K. K.w(

..

!

- Bruce H

New SPi

Lxial and 1

eSOnance :orresponc he freezin Javers 'H

ue the rad iumbar ne which hav Ute to nerv CT compo ~l foramir studied; in neural fori ligaments terolateral bra1 disk a pedicle, st transverse R a m . 01 as linear s nation coe adjacent f; MR image lower sign the adjace the neural that the lil men can b CT or MR

aograpfi

Index terms: -ompara tive :3992 Spit 39 1211 s

IlldiDlogy 15

_c_

From the M e g e of H Lutheran Hc waukee, WI

request , h24;acceF

Cn by gran h t u t e s of '- .HI+

'RSNA 1

f. phd j. in brain phosphorus metabolite concentrations ... · rs large arger exo- uding

Documents