close

Вход

Забыли?

вход по аккаунту

?

Patterns of metabolic activity in the treatment of schizophrenia.

код для вставкиСкачать
Patterns of Metabolic Activity
in the Treatment of Schtzophrenia
Jonathan D . Brodie, MD, PhD," David R. Christman, PhD,? Juan F. Corona, PhD,f Joanna S. Fowler, PhD,I.
Francisco Gomez-Mont, MD," Judith Jaeger, MA," Peter A. Micheels, MA," John Rotrosen, MD,*6
Jerome A. Russell, PhD,t Nora D. Volkow, MD,' Alan Wikler, MA,"
Alfred P. Wolf, PhD,? and Adam Wolkin, MD*§
Six patients with chronic schizophrenia were studied w i t h positron emission tomography (PET) before and after
neuroleptic treatment, using fluorine-18-labeled fluorodeoxyglucose. After treatment, the mean whole-slice glucose
metabolic rate at the level of the basal ganglia showed a 25% increase. However, patterns of frontal hypometabolism
observed w i t h the schizophrenic patients were not altered by medication. Pattern analysis using the fast Fourier
transform was applied to a set of 422 images from a mixed g r o u p of normal, depressed, and schizophrenic subjects.
Reconstruction of the images with low-frequency coefficients was excellent, reducing considerably the number of
variables needed to characterize each image. Hierarchical cluster analysis categorized the transformed images accordi n g to anatomical level and subject g r o u p (patient versus control). The results suggest the utility of this procedure for
the classification a n d characterization of metabolic P E T images from psychiatric patients.
Brodie JD, Christman DR, Corona JF, Fowler JS, et al: Patterns of metabolic activiry in the treatment of
schizophrenia. Ann Neurol 15(suppl):S166-S169, 1984
Several laboratories, including o u r own, previously reported that patients with schizophrenia have a decreased rate of glucose utilization in the frontal cortex
[a-61. I n this study w e examined the effect o f
neuroleptic treatment o n the local cerebral metabolic
rate of glucose utilization (ICMRGlu) of a small g r o u p
of patients with chronic schizophrenia. We also report
the results of o u r initial attempts at data compression
and pattern analysis, in an effort to characterize better
t h e regional metabolic properties of psychiatric
populations.
Methods
Six hospitalized male patients with chronic schizophrenia
who were moderately to severely ill, with a mean Brief Psychiatric Rating Scale score of 44 (range, 33 to 5 2 ) , were kept
free of neurolepric drugs for a period of at least two weeks.
They all met Research Diagnostic Criteria and were rated for
psychiatric disorder and cognitive functioning with a battery
of instruments, including the Schedule for Affective Disorders, the Schedule for Assessing Negative Symptoms, and
the Clinical Global Impression scale. Cognitive testing included the Wechsler Adult Intelligence Scale as well as other
psychometric tests of attention, concentration, and memory.
None of the patients had any contributing medical or
neurological illness. The five age-matched normal controls
were screened for the absence of psychopathological dis-
From the *Department of Psychiatry, New York University Medical
Center, New York, NY, the ?Department of Chemistry, BrookhaNational
NY' the 'IBM Research Center,
Mexico city, Mexico, and the §PsychiatryService, New York Veterans Administration Medical Center, New k'ork, NY.
Sl66
orders by the Structured Clinical Interview. Prior to the PET
run, each patient had a computed tomographic (CT) scan
taken through the canthomeatal (CM) line, which was used
for both alignment and analysis of the PET images. All of the
subjects were scanned o n the P E I T VI using fluorine-18labeled fluorodeoxyglucose ("FDG) under resting conditions
in a dimly lit room with eyes open and ears plugged. No
provocative stimuli were provided at any time. The subjects
were rated for anxiety before and after the run and were
questioned for the specific occurrence of hallucinations during the uptake period. The patients then entered a treatment
phase, during which they received thiothixene, a standard
neuroleptic, in doses ranging from 20 to 100 mg per day.
Treatment was based solely on clinical response and tolerance of side effects and was continued until an end point
determined by clinical judgment. At the end of the treatment
period, which ranged from four to ten weeks, the patients
were scanned again with l8FDG under the conditions described previously. Informed consent was obtained from all
subjects, following a full explanation of all procedures.
PET scans were analyzed by the region-of-interest method,
in which the appropriate transverse C T scan of each patient is
precisely aligned with the corresponding PET image and anatomical landmarks are used to define the regions of interest.
Since prior studies C3, 51 used instruments of lower resolution, such as PETT 111, we attempted to reproduce at least
some of the previous regions, such as the frontal region 12,51
in the plane parallel to the CM line that includes the thalamic
]]Onleave from Mexican Institute of Psychiatry, Mexico City, Mexico.
Address reprint requests to Dr Brodie, Department of Psychiatry,
New y o & University Medical Cenrer, >>,, First Ave, New York,
N Y 10016.
Normals
(N=5)
Schizophrenics
Pre-Weatment (N=6)
Fig 1. Local cerebral metabolir rate of glurose utilization i n normal subjects and in drug-free schizophrenicpatients. Data are
expressed aJ regionlwhole slice ratios in the basal ganglionic
plane. Cortical regions are depicted with the frontal region at the
top and the temporal and occipital regions at the bottom, with the
left regions on the left. Ovals represent the regions of the thalamic
nuclei. and wedges are the regions of the lentifom nuclei.
nuclei and transects the basal ganglia, for comparison with
our earlier work.
In addition, 422 PET images were obtained from 12 controls, 13 schizophrenic patients, and 6 depressed subjects ( 5
schizophrenic patients and 3 depressed subjects had images
taken before and after treatment). Each image was standardized by z- transformation [7f to a mean of zero and a standard
deviation of 1, to enable US to study the patterns of metabolic
activity independently of the absolute metabolic values. The
fast Fourier transform { 11 was applied to the standardized
images. The low-frequency 11 x 11 matrix of Fourier
coefficients was used as the feature vector for each image.
This matrix was selected because the 12 1 coefficients gave a
good reconstruction of the original image. The reduction in
the number of variables is considerable, allowing the application of multivariate techniques for classification purposes. A
cluster analysis using the farthest neighbor method was done
on the 422 Fourier images, using the 121 lower frequency
coefficients as feature vectors. The number of clusters in the
dendrogram was determined at the level at which discrimination between noncontiguous anatomical slices was obtained.
The clusters were examined to see if this grouping corresponded to clinical and anatomical designations.
Results
Region-of-interest analysis of the basal ganglionic plane
is shown in Figure 1. The data are consistent with our
earlier results, which showed decreased ICMRGlu in
the frontal region of a mixed sample of medicated and
unmedicated patients with chronic schizophrenia when
compared with an age-matched population screened
for the absence of psychopathological disorders [3, 51.
The lateral frontal symmetry observed in the normal
population is replaced by frontal asymmetry in the
medication-free patient population. The mean frontal/
whole slice metabolic relationship, which is equal to or
greater than 1 in several reported studies involving
normal populations 13, 5}, is slightly less than 1 in the
Pre-Treatment
Post-Treatment
Schizophrenics (N = 6)
Fig 2. Comparison of local rerebral metabolic rate of gluco.re utilization in schizophrenir patients before and after neuroleptir
treatment. Data are expressed as in Fig 1.
Fig 3. Effect o f neuroleptir treatment on regional cerebral metabolic rate of glucose utilization. Quantitatioe data are expressed
as the percentage increase in the local cerebral metabolic rate of
glurose utilization ooer the regional means in the pretreatment
studies.
schizophrenic group and significantly less than 1 when
only the left frontal region is considered. Increasing the
size of both groups to a larger sample confirms that this
finding of left “hypofrontality” is statistically significant
(A. Wolkin and colleagues, unpublished data, 1984).
Inspection of mean regional ICMRGlu for schizophrenic patients after treatment shows a distribution of
values that look remarkably similar to those obtained
before treatment was instituted. This pattern is
schematically represented in the comparison shown in
Figure 2. Both the lateral asymmetry and the hypofrontality are preserved in the group after neuroleptic treatment.
Treatment to the clinical end point resulted in a
mean group decrease in Brief Psychiatric Rating Scale
scores from 44 to 34, accompanied by a mean increase
in whole-slice ICMRGlu of about 25%, as shown in
Figure 3, left. The statistical significance of the individual regional changes shown on the right in Figure 3
remains to be established.
The adequacy of the data compression was evaluated
by reconstructing the PETT VI metabolic images using
progressively bigger matrices of the low-frequency
Fourier coefficients. The reconstruction of the 11 x
Brodie et al: PET Studies in Schizophrenia
S167
Anatomical and Clinical Cowekztes of Cluster AnalyJiJ
No. of
Cluster
Images
Percentage of Images by Level
Upper
Middle
Lower
Controls/Patients(%Y
0
0
5
0
0
26
20
54
92
0
0
10
21
0
100
100
100
100
0
0
7 512 5
0/100
39/61
5 1/49
41/59
36/64
~
A
R
(
D
E
F
G
H
I
J
K
L
M
N
0
P
Q
57
41
21
54
29
34
20
28
13
26
4
10
11
10
25
3
11
R
9
S
T
11
5
5
0
95
100
90
5
30
76
70
24
15
0
7
8
100
50
50
59
80
39
0
0
50
40
64
30
0
0
0
0
100
100
9
70
0
0
0
0
0
0
41/59
3 1/69
76/24
67/33
0/100
o/ 100
61/39
100/0
50150
100/0
36/64
27/73
01100
01100
”Percentages adjusted to correct for the different proportion of control and patient images
11 matrix was almost identical to the original image.
Most major features in the origmal image are easily
discerned, even in the 7 x 7 low-frequency coefficient
matrix. The Table shows the results of the cluster analysis. Twenty clusters were obtained. Three clusters (A,
B, and C ) include 36% of the images. Cluster B is of
particular interest, for it includes 41 images from patients and no images from controls. All of the images
are at the level of the basal ganglia. The results suggest
that the method is suitable for the characterization and
classification of PET images.
Discussion
The results of this preliminary study on the effect of
neuroleptic treatment on lCMRGlu in a group of patients with chronic schizophrenia are consistent with
the notion that, once manifested, this illness is never
cured, even though the expression of symptoms and
the degree of impairment may vary. Relative left hypofrontal glucose metabolism may well be one of
the physiological derangements characteristic of the
chronic presentation of this illness, since it clearly appears to be a consistent finding [3). Although the few
reported schizophrenic patients with acute psychosis
IS, S} seemed not to show this “hypofrontal” picture, it
should be noted that none of the patients in our population reported active hallucinations during the uptake
period of either PET run.
The failure of drug treatment to alter the hypofrontal pattern of ICMRGlu, despite the clinical im-
provement, with each subject serving as his or her own
control, helps to explain our earlier findings with a
larger group of patients with chronic schizophrenia IS}.
In that study, when the PET data were analyzed as in
the present study, the group behaved as a single statistical entity upon probit analysis, even though roughly
half of the patients were receiving neuroleptic medication at the time, while the others were drug-free.
The present study also suggests some of the limitations of this approach to the study of the metabolic
concomitants of psychiatric disorders that are largely
defined by behaviors. We are reporting the partial analysis of one of fourteen tomographic slices obtained in a
simple test-retest paradigm. It is clear that any volume
element that can be resolved by PET technology is
undoubtedly functionally as well as neuroanatomically
heterogeneous. In addition, the many clinical descriptors required to characterize a psychiatric trait are further complicated by the variety of psychological states
for which there are few objective measures and even
less experimental controls. Although the use of the
subject as his or her own control simplifies the obtaining of meaningful data, the extraction of a signal that
represents an aberrancy of mentation or mood remains
a formidable problem, which can best be approached
by a stimulus challenge to the putative disturbance.
The enormous quantities of clinical and metabolic
data dictate the need for appropriate data compression
and faster image analysis. The determination of metabolic patterns in images of the whole brain involves the
S168 Annals of Neurology Supplement to Volume 15, 1984
analysis of much larger matrices of data than when
studying selected regions of interest. The data have to
be compressed in a way that adequately reflects the
original image. Dividing the image into a given number
of regions and obtaining the average for each region is
one way to achieve this compression. However, in order to get a good reconstruction of the original image,
we would need to draw regions of interest that would
approximate the spatial resolution of the tomograph.
We have reported a procedure to compress the original
128 x 128 image and to present preliminary data on
the analysis of these images and their application to
psychiatric populations, using multivariate statistical
techniques. Our preliminary efforts at applying the fast
Fourier transform and cluster analysis to PET images
from psychiatric patients (F. Gomez-Mont, N. Volkow
and colleagues; unpublished data, 1984) suggest that
these and similar techniques will be of great value in
the study of the “functional” psychiatric disorders.
~
Supported by Grant NS 15638 from the NINCDS, by the Department of Energy, and by the Veterans Administration. The authors
gratefully acknowledge the generous support of Roerig Division of
Pfizer, Inc, IBM de Mexico, and the Mexican Institute of Psychiatry.
References
1. Brigham EO: The Fast Fourier Transform. Englewood Cliffs, NJ,
Prentice-Hall, 1974
2. Brodie JD, Wolf AP, Volkow N , et ak Evaluation of regional
glucose metabolism with positron emission tomography in normal
and psychiatric populations. In Heiss WD, Phelps ME (eds). Positron Emission Tomography of the Brain. Berlin, Springer Verlag,
1983, pp 201-206
3 . Buchsbaum MS, Ingvar D, Kessler R, et al: Cerebral glucography
with positron emission tomography. Arch Gen Psychiatry
39:251-259, 1982
4. Farkas T, Reivich M, Alavi A, et al: The application of “F-2deoxy-2-fluoro-d-glucose and positron emission tomography in
the study of psychiatric conditions. In Passonneau JV, Hawkins
RA, Lust WD,et al (eds): Cerebral Metabolism and Neural Function. Baltimore, Williams & Wilkins Company, 1980, pp 403408
5. Farkas T, Wolf A, Jaeger J, et al: Regional brain glucose metabolism in the study of chronic schizophrenia. Arch Gen Psychiatry,
1984 (in press)
6. Ingvar D H , Franzer G: Abnormalities of cerebral blood flow distribution in patients with chronic schizophrenia. Acta Psychiatr
Scand 50:425-462, 1974
7. John ER, Karmel BZ, Corning WC, et al: Neurometrics. Science
196:1393- 1410, 1977
8. Widen L, Bergstrom M, Blomqvist G, et al: Positron emission
tomography studies of brain energy metabolism in schizophrenia.
In Heiss WD, Phelps ME (eds): Positron Emission Tomography
of the Brain. Berlin, Springer Verlag, 1983, pp 192-195
Brodie et al: PET Studies in Schizophrenia
S169
Документ
Категория
Без категории
Просмотров
5
Размер файла
356 Кб
Теги
treatment, patterns, metabolico, activity, schizophrenia
1/--страниц
Пожаловаться на содержимое документа