Automated image analysis of disturbed cytoarchitecture in Brodmann area 10 in schizophrenia: A post-mortem study

J+og. Neum-FsydwpharmawL

(P Rid. Psych&.

2000,

Comght

Vd.

0 2000

24, pp. 1093-l

FTinted in the USA.

Au lights resavcd

0278-6646/OO/&sce

YASUHIRO

KAWASAKI’,‘,

KAI VOGELEY’, VOLKER JUNG’, RALF TEPEST’, AXEL SCHLEICHER’ and PETER FALKAI’

front matter

HELGE HUTTE’,

‘Dept. of Psychiatry, University of Bonn, Bonn, Germany ‘Department of Neuropsychiatry, Kanazawa University, Japan %Iogt Institute of Brain Research, Univ. of Duesseldorf, Duesseldorf, Germany

(Final form, September 2000)

Kawasaki, Yasuhiro, Kai Vogeley, Volker Jung, Ralf Tepest, Helge Hijtte, Axe1 Schleicher and Peter Falkai: Automated image analysis of disturbed cytoarchitecture in Brodmann area 10 in schizophrenia: A post-mortem study. Prog. Neuro-Psychopharmacol. & Biol. Psychiat.. 2OO0, 24, pp. 1093-l 104. @2000 Elawier Science Inc. 1.

2.

3.

4.

Among different etiological concepts in schizophrenia research is the disconnectron Adequate empirical research requrres hypothesis involving distributed brain regions. correlational studies of multiple brain regions. In this pilot study, the authors therefore tested the applicability of an automated image analysis device as a scanning tool to detect cytoarchitectural abnormalities in Brodmann area (BA) IO. The authors applied the gray level index (GLI) method as automated image analysis on 10 The GLI as perikarya-neuropil-ratio IS schizophrenic brains compared to 10 controls. obtained as the ratio between the area covered by cellular cross sections and the area of the total measuring field in 101 continous measuring fields from pial surface to the cortrcal depth. Resulting data provide a specific cytoarchitectonic profile curve. An analvsrs was performed separately for mean GLI and GLI values in six compartments coverrng approximately the different cortical laminae. A statistically significant reduction of the mean GLI was demonstrated in the schizophrenic group covering laminae Ill to VI, as detected by multivariate analysis and corroborated by univariate analyses and t-tests. This result clearly underlines a cytoarchitectonic disturbance with a perikarya-neuropll-ratro reduction in BA IO, that is associated with schizophrenia. This is suggestive either of an increased neuropil fraction or a decreased neuronal perikarya fraction. The latter could either be due to a volume or a total number reduction of neuronal perikarya. These data are compatible with previously published data on cell loss in schizophrenics in BA 10.

Kevwords

automated

image

analysis,

Brodmann

area 10, cytoarchitecture,

prefrontal

cortex,

schizophrenia. Abbrevlatlons: interest

Brodmann

area (BA), gray

level index

IROI).

1093

(GLI), prefrontal

cortex

104

Elsevicr Science Inc.

(PFC), region

of

1094

et al.

Y. Kawasaki Introduction Schizophrenia

temporal

research

lobes,

Identify

the limbic

a single

disturbance

syndrome

substantial pathological

changes lobe

clngulate

gyrus

pathological

changes,

1988,

Schlaepfer

1999)

and post

al., 1986,

1991,

1996).

prefrontal

Notably,

to

of a

undergo

evidence

cortex

for

(PFC),

the

gyrus,

the

The PFC undergoes 1986,

neuropsychological respect

regions

temporal

1998). et al.,

failed

to a complex

empirical

superior

(Andreasen

special

1998).

brain

is

and Falkai,

et al., 2000),

1995,

the

studies

and

the hypothesis leading

several

the

and

studies

with

however, Thus,

There

comprising

(Vogeley

the latter

et al.,

concept,

of the frontal

Shelton

et al.,

data (Vogeley

et al.,

to cytoarchitecture

Benes et al. 11986,

(Benes et

1991)

reported

cell loss in layers III and V in BA 10.

IS necessary

this disconnection

to examine

pathological

multiregional

correlational

image

(Schleicher

and

independent,

fast,

differences

of neuronal

is the first

study

abnormalities

scanning

In this pllot

study

device

the

for

reproducible,

perikarya,

et

applies

tools

gray

1999).

al.,

perikarya

The

related

cell nuclei

it

with

of

such

a

to detect

the potentially

morphological

studtes.

level index the GLI

automated

of GLI are detectable.

the GLI method

of

sample,

the results

brain samples

are needed

the so-called

glial and endthelial

and to correlate

to more detailed

and objective

of stained

In a post mortem

To examine

characterization

Schleicher

in the density

which

fast

forwarded

the area percentage

lnterlaminar

brain sites,

are then

1990,

highly

of schizophrenia

brain sites in one brain sample

approach,

analysis Zilles,

hypothesis

of different

which

we employed

automated

measures

multlple

changes

brain regions,

this purpose,

volume

Vogeley

sites,

brain structures

formation

by imaging

studies,

Selemon

this

regions

structures

et al., 1994, mortem

In order to study

relevant

brain

hippocampal

as shown

different

Following

pathology

schizophrenia.

(Welnberger,

in multiple

and subcortical

subcortical

underlying

involving

changes

the

for distributed

and multiple

is proposed.

including

evidence

pathology

network

pathological

temporal

putative

system

site of brain

in a distributed

disconnection

pyramidal

has generated

(GLI) method

cytoarchitectural method

image

For

profile

is an

analysis

as

observer-

device,

which

to the area of the measuring

field.

The GLI is an estimate

of the total

(Schleicher

1990).

as scanning

tool

and Zilles,

to detect

This

cytoarchitectural

in schizophrenia.

Methods Brain Collection BA 10 was studied

post

age- and sex-matched given

inTable

controls,

1. Before

from the responsible

mortem

all collected

autopsy,

authority

in brains

informed

from

10 cases of schizophrenia

between consent

1985

and 1995.

was obtained

In cases under legal care. These

compared

Demographic

from the nearest procedures

were

to 10

data are relative,

approved

or by

Qtoarchitecture the local ethics obtained.

committee

Patients

myocardial

satisfied

both

any neurologic relative.

were

illness

excluded

to DSM-III-R

fixation,

a reference

the

criteria,

removed

callosum.

The tissue

cases

the

All cases

had no history

of

from the nearest

hemispheres

occipital

wrth

of schizophrenia review.

and interview

from

and

the brains were

arteriosclerosis

on chart

Control

to chart review

temporal

where

The diagnoses

based

for schizophrenia.

the

1095

cancer,

from the study.

PFC was

line joining

Dusseldorf,

fmetastatic

and ICD9

disease according

of the genu of the corpus

Stainina

poles

wrth

was then embedded

usrng the

a cut

posterror

In paraffin.

and lmaae Acauisition

Serial level

systemic

and ICD9 criteria

formalin

to

boundary

or stroke)

or psychiatric

After

orthogonal

severe

according DSM-III-R

area 10 in Schizophrenia

of the Heinrich-Heine-Universitlt

with

infarction

was established

of Brodmann

sections,

of genu

(Merker,

20 urn in thickness,

of the

1983)

microscoprc pole with

corpus

clearly

distinct

which

is not

as densely

poles

were

were examined

covering

6 different

overall

structural

uniformity

a CCD-video

a Marzhluser

each field of view

of Interest

GLI measurements

the selected

(Wetzlar,

defined,

attached

was 25x. Germany).

Images

Using

measuring

fields

with

were

a video

to a square microscopical

into square

areas.

stamrng

uniform

lamma

III,

sections

from

the

3 serial

original

sulcus.

each

based

were

microscope acquired frame

Cases

case numbers.

In ROI

on the

to the surface, Images

In

of the frontal

of ROls was

to a light

the

inspection

3 per hemisphere,

The selection

ROI and adjacent

from

silver

by visual

the frontomarginal

of cells perpendicular

Japan),

Lens magnification table

Gallyas

Ill and V, rather

and substituting

(ROls) were

to anterior,

in the rostra1 portion

covering

25 x 25 mm.

XC75CE,

Modified

cells as BA 46.

by covering

the orientation

corresponded

frame was subdivided

pyramidal

for the diagnosis

(Sony

scanning

regrons

with

through

6.3 x 1.25, Germany).

cell-sparse

for the measurement,

section,

camera

pole.

posterior

50”’ slice. BA 10 was deftned

packed

regions

of histological

frontal

from

II and IV, wade laminae

an area of approximately

quality

starting

laminae

selected

blinded

cut,

to the

on every

identifying

frontal

each case,

callosum

was performed

images

were

and the

acquired

with

(Zeiss Planpro”

automatically

of 512

x 512

field of approximately

with ptxels,

165 pm. Each

25 x 25 urn per hemisphere,

in which

were performed.

lmaae Processing Image analysis Images applied sectron. threshold, (Fig.

were

was performed

obtained

resulting Adaptive

as digitized

in an image

the

with

was

was

level

Image Analyzer images

(Fig.

for inhomogeneities

used

the gray

threshold

IBAS 2000 gray

adjusted

threshold

established

1 b). Thus,

with

for

segmentation

level histogram determined

la).

with

Eching,

A shading

in staining

appearance

on each

(Kontron,

intensity

correction

was

throughout

the

a ROI-dependent of pyramidal

ROI separately

Germany).

gray

value

cells in layer Ill

and was

observer-

Y. Kawasaki

1096 independent. performed

To correct

lndrvidual to

the whole

profiles

300

according

due to local generated profile

structural

from

a total

comprised

with

laminae

measuring

(lamrna

IV),

69-83

fields

(7 profiles following

V),

84

101

and

fields

in width

ending

in the

profiles

by identifying

(correspondrng white

matter

per case

profiles was

to the cortical GLI profiles across

isolated.

thus

depth,

were then

all laminae

was

For this purpose,

the

local maxima

appearance.

were generated

ROls). The resulting

(Fig. Ic).

the mean GLI value were

border.

in individual

The GLI profile

the pial surface

laminae

in measuring

to the same relative

Thus, distortions

depth

was

(Bit Pad One, Summagraphics,

were standardized

from

operation

then acquired

per ROI in 6 different

by their histological I), 14

filling

to the gray-white-matter

tablet

out.

the cortical

defined

1 - 13 (lamina

(lamina

ruled

the six different were

were

1). 42 individual

thickness

From the GLI profile,

In addition,

surface

a digitizing

were

in GLI throughout

processing.

pial

and averaged.

points,

(Fig. 2), corroborated

covered

varying

et al. (1976)

of 7 x 6 profiles

of the different

first derivative

with

inhomogeneities

variations

(Fig. Id).

at the

a binary

13 measuring

(SUN Spare Station

101 measuring

for further

calculated borders

along a line with

starting

The profiles

GLI values

from the pial surface

was performed

to Hudspeth

thus describrng stored

depth

USA) and a workstation

in each hemisphere. width

‘pm),

This procedure

Fairfield,

(Wree et al., 1982). cortical

were extracted

approximately

boundary.

for holes in the area of the perikaryon,

after thresholding

fields spanning

et al.

and minima

The six different

in the lamrnae

- 23 (lamina II), 24 - 55 (lamina Ill), 56 - 68

flamina

VI).

Cortex

thickness

was

not

measured

systematically.

Fig. 1. This sketch illustrates the automated image analysis procedure. The cortex is completely scanned (left to right = pial surface to white matter border zone), covering several mm over the cortical surface (al. Images are then binarized (b) and measured in the individual measuring fields, different densrty values are obtained as mean gray level values, depicted as graytones (c). Parallel to the laminar cortex organisation average values are calculated as depicted for illustration in Id).

Cytoarchitecture

1097

of Brodmann area 10 in Schizophrenia

Fig. 2. These curves demonstrate the average cytoarchitectonic profile covering the whole cortical depth. The profiles of normals (closed line) and schizophrenics (dotted line) do not differ in shape, however, the GLI values are significantly reduced in laminae III - VI. Different iaminae are segmented according to microscopic inspection and local maxima in the first derivative (gray line, bottom).

Statistical

Analvsis

Demographic usrng t-tests cortical

analysrs

values

of

investigate dependent

was

covering

results,

to

with

diagnosis,

from the multrvariate unpaired

potentrally

samples,

analysts

where

fixation

confounding

were followed

as fixed

by univariate

GLI value

two-sided)

six

For statlstrcal were

Secondly,

laminae

used

significant

and subsequent

to

on the

multivariate

as dependent

Thirdly,

analyses

over total of the

and brain weight

variables.

factors.

groups

These compartmental

per lamina.

time,

control

the curve)

laminae.

volume

of the six different

sex and hemisphere

under

(Pearson,

time,

and the

mean total

(area

neuronal

calculations

for the mean GLI and GLI values

used

were

the six different

of age, post-mortem detect

schizophrenic

GLI values

of the total

correlative

effects

the

variables

of the

approximately

estimates

firstly,

variables

between

Dependent

calculations

provide

possible

compared

samples.

and integral

compartments

GLI integral

were

for unpaired

depth

different

analysis

variables

variables findings t-tests

for

applicable

Results Demoaraphrc Descriptive reported

Data statistics

In Table

of

variables

that

1. At first hospitalization

male schizophrenics

It

significantly

than

heavier

= -2.803, brains

df

=

of female

might

possibly

female

schizophrenics

15.063, patients

influence

p = 0.013). (t = 4.087,

the

were Brains df

=

measurements

significantly

are

older than

of male patients 18, p = 0.001).

were The

1098

Y. Kawasaki et al.

mean duration was

not

of illness

in male patients

statistically

significant

between

the groups

drfferences

(t

=

was

longer

1.328,

df

than

=

in female

16.107,

for mean age of death,

p

=

post mortem

patients,

however,

0.203).

No significant

delay or formalin

this

fixatron

trme were observed. Table 1 Demographic

Data Males

Controls

T Age of Death (yrs) Disease duration

5

sdev

n

54.60

6.54

5

51.80

5.54

5

22.00

4.95

5

29.80

4.44

(yrs)

Frrst hospitalization

(vrs)

Brarn Weight

(g)

Post Mortem

Delay (hrs)

Formalin

Schizophrenics

M

Time (days)

m

sdev

1408.00

113.67

5

1475.00

56.86

40.80

20.07

5

43.20

20.08

906.40

1557.68

5

1045.40

1380.78

Females Controls N Age of Death (yrs) Drsease duration

5

Brain Weight

(g) Delay (hrs)

of Potentiallv

variables mortem

of

age (r = delay

5

58.80

11.26

5

23.75

12.23

5

33.75

10.63

95.13

5

1217.90

83.03

24.00

0.00

5

36.00

24.00

466.00

359.65

5

1199.00

1 101.63

1200.00

n: cases, m: mean, sdev: standard

In the

13.24

60.60

Formalin Time (days)

Correlations

n

(vrs)

Post Mortem

separately.

sdev

(yrs)

Frrst hospitalization

Correlations

the

Confoundina dependent

control 0.559,

(r = -0.523,

with

Deoendent

variables

group,

the

p =

were

dependent

O.OlO),

p = 0.018).

brain

these vanables

as confounding

were not included analysis

sdev

Vary&&

calculated variables

weight

In contrast

wrth

for multrvariate

m

deviation

mean GLI value did not correlate

as required

Schizophrenics

M

(r =

variables,

both

-0.459,

to this,

any of the potentially

for

correlated

diagnostic

significantly p =

0.042),

in the schizophrenia

confounding

variables.

as the preconditions

groups, with

the

and post group

the

Therefore,

for covariates

were not fulfilled.

GLI Analvsis Descnptive multivariate

statistics

of GLI measurements

and univariate

analyses,

where

are listed applicable,

in Table

are given

2. Significant

results

from

in Table 3. For the mean GLI

Cytoarchitecture analysis,

a statistically

significant

df = 6, 27, p = 0.025). any interaction

term.

covering

six

the

dragnostic curve

groups

integrals

the significant

of Brodmann

impact

whereas

different

coverrng

performed

demonstrated

Ill to laminae

lamina

differences

analysis,

laminae,

for laminae

was only found

for the factor

there was no significant

The univariate

assessed

influence

a

of gender,

significant t-tests

VI. separately.

by univarrate

diagnosis

for the different

VI. Subsequent

III to lamina

1099

area 10 in Schizophrenia

analysis

(F = 5.564, hemisphere

areas under the curve

difference were

Comparing

or

between

performed

the

for the GLI

the diagnostic

groups,

were corroborated.

Table 2 Descriptive

Statistics Males

Controls

Schizophrenics

GLI Mean GLI Lam I

Sdev

n

14.94 107.72

24.23

18.00

2.69

5

156.33

46.16

5

sdev

GLI

1.55

Lam II

185.44

30.23

5

154.69

23.03

lam Ill

597.55

79.81

5

502.83

54.41

lam IV

263.38

36.42

5

222.39

24.58

lam V

288.81

41.47

5

247.16

21.69

lam VI

326.66

47.09

5

275.14

24.27

Females Controls

sdev

sdev

N

GLI

18.57

2.60

5

17.54

3.35

163.39

40.90

5

159.29

61.39

mean GLI lam I

Schizophrenics

GLI

lam II

186.93

21.71

5

181.06

37.95

lam Ill

611.94

86.70

5

591.86

105.95

lam IV

269.97

34.14

5

251.85

41.09

lam V

302.30

43.84

5

279.51

46.68

lam VI

340.98

38.88

5

308.31

59.42

n: cases, GLI: GLI integral (area under the curve) covering fields of one lamina, sdev: standard deviation of one

measuring

Table 3 ANOVA

and T-test .-.

ANOVA F

(factor df ”

diagnosis)

t-test P

t

(factor df ”

diagnosis) P

Mean GLI

5.564

1, 32

0.025

2.362

38

0.023

Lam I

3.093

1, 32

0.088

1.742

38

0.090

Lam II

3.725

1, 32

0.063

1.945

38

0.059

Lam III

4.432

1, 32

0.043

2.078

38

0.045

Lam IV

6.714

7, 32

0.014

2.641

38

0.012

Lam V

6.027

1, 32

0.020

2.507

38

0.017

Lam VI

8.245

1, 32

0.007

2.954

38

0.005

” Hypothesis

degree of freedom,

Error degree of freedom.

1100

Y. Kawasaki et al.

Thrs study

demonstrates

agarnst controls GLI method comparison

significant

changes

in BA 10 as assessed

developed

by

to previous

in the cytoarchitectural

profile

of schizophrenrcs

by the observer-independent

automated

image analysis

and Zilles

to discuss

the

Schleicher

cytoarchitectural

(19901.

studies,

In order

the method

must be critically

results

in

evaluated.

Methodoloav The GLI is defined measurrng

field.

compartment

The stained

represents

factors

comprrsrng

starnrng

rntenslty,

brnanzatron The

as the area percentage

and

mainly

the

size

absolute

of

compartments neuronal

value

starned

of structures

contributrng,

structures

density

in relation

perikarya,

on different

thickness

to the threshold

to the

the unstained

of GLI depends

structures,

according

and different

includes

species

in different

of

sectrons,

used for image

1973,

stated,

In Nrssl-stained

that

clear distinction a potentially

sections,

using

schrzophrenra, neurodegeneratrve

is

no

drsease

by Harrison

on which

many

along

on neuronal respect

support

for

schizophrenics

the

of studies

and controls,

(Nurnberger,

impossible.

the

that

in glial

cell

would

Tower

be clearly

were

done,

of for

condition

of

density,

there

a

also apply

schizophrenia

on glial cells revealed that

constant

1958,

So the problem

pathological

view,

suggesting

non-neuronal

it must

studies

and

in different

of as an additive,

cell counts

to

an increase

The vast majority

of these

of stereological

With

with

density

In addition,

is virtually

used as an

Schleicher

As studied

curves

et al., 1986).

1986,

cells.

be thought

Schleicher

convincing

going

(1999).

between

the volume

and can thus

sections.

et al.,

glial and endothelial

glial cell populations

Nissl-stained

there

Schleicher

glial cells and small neurons

contaminating

studies

1982,

the shape of the GLI profile

Wree et al., 1980,

between

by the GLI and can be reliably

demonstrated,

laminae

but not influencing

and Young,

is measured

(Wree et al.,

Thus measurement

regions

drfferences

density

of stained

for the volume

reviewed

represents

Basically,

and discrimrnation

cells do not differ

all

by stained

(Wree et al., 198.2).

Z~lles, 1990). cortical

neuropil.

packing

areal densrty

estimate

compartment

covered

IS a

as recently no significant

is no contamination

wrth glta cells (Falkar et al., 1999). The GLI value does show (Wree et al., 1982). the gurnea

In this

study,

pig in Nissl stained

62.5%.

Variations

Induced

by an incomplete

focus,

a nonlinear,

but detected

Irl kirn are eliminated

of

?lpm

monotonic

the GLI value

sections.

resulted

increase was

In sections

in fluctuations

segmentation

in thin

during

thresholding

in thicker

during

averaging

In greater

with

measured of 20um

between

sections sections.

increasing

thickness

in the regio praepiriformis thickness,

60.8%

the GLI value

and 63.9%.

and projections Differences

series (Schleicher

section

et al., 1986).

was

This effect

of structures in section

of

is

out of

thickness

of

Cytoarchitecture The potentially the

confounding

application

sections

and secondly,

hrstogram

appearance

between

density. precisely

known.

methods.

stated,

Therefore this

which

reference

reporting

as studies

and an overall Rajkowska changes

increase

inhomogeneities

by

within

determined

according

for staining

inhomogeneities

to the

neuronal

was not

is not

conclusive

of the ratio can in turn be of the average

neuronal

of the GLI value.

other

region

reports

failed

of schizophrenics

of about

25%

density

and

smaller

of the

10-47).

Also

of reductions GLI method

a decrease Beasley

as a global

of small and

in accordance

of neuronal

interneurons

Reynolds

(1997)

resultrng

found

could

also reflect

a reduction

of the neuronal

in a compensation

demonstrated

the PFC of schizophrenics.

by Rajkowska cell

though

in the

30%. of

rostra1

matching

are of course

For instance,

II by

cases.

bodies

is exactly

et al. (1999)

not exactly

which

a reduction

22 control

cytoarchitectural

Benes

Looking

our

not under

for

et al.

another

parvalbumin-positive

Parvalbumin-positive

neurons

by 35%.

GLI measurements et al. (19981

against

1998,

cases. This report

procedure.

found

1995,

only the left

neuronal

in lamina

1998)

of BA IO studying

subpopulations,

estimation

et al.,

et al. (19861

reported

with,

as

Benes

to 9 control

neuronal

et al.,

quantitative 1993).

there

et al. 1999)

(Rajkowska

(Selemon

to detect

in the deep laminae

cases compared

Rajkowska

decrease

(Pakkenberg

reduced (BA

1991,

in BA 9 and 46

were

cortex

in the PFC of schrzophrenics,

volume

results

and an increase

as obtained

to the decrease

volume

any of the stereological

volume

a decrease

disturbances

density

in BA 10 in 18 schizophrenic

interestingly

with

neuronal

and/or

similar

were found to be reduced

Rajkowska

number

of neuronal

but its magnitude

The disturbance

Recently,

subpopulatron,

However,

of

loss in BA IO (Benes et al., 1986,

whereas

density

are findings

found

not compete

our study.

demonstratrng orbitofrontal

a biased estimate

in our sample,

change.

both contribute

in 10 male schizophrenic

in lrne with

the

neuronal

of neuronal

prefrontal

a decreased

scope

fraction

of

lamina-specific

et al., 1998), in the

hemisphere

does

change

to cytoarchrtectural

well

140%,

firstly,

Disturbance

on neuron

neurons

was diminished

staining

threshold

provides

the GLI method

of total

are both reports

(1991)

for

III, correcting

across the specimen

could potentially

special

results,

cells in lamina

cytoarchitectural

a decrease

Cvtoarchitectural

found

intensity

corrects

that the GLI method

the underlying

due to either

With

which

by the use of a section-specific of pyramidal

In addition,

concerning

40%,

thresholding

This bias was constant

volume,

of the local staining

sections.

It must be clearly

the

influence

of adaptive

1101

of Brodmann area 10 in Schizophrenia

that These

by GLI measurements.

of the neuronal

density of the

there might

may

disturbances

in the mean

density

of extra

of small- and medium-sized reduction

occur

of extra

lamina-

well account

large

neurons.

global

volume.

large neurons

neurons

and cell-type-specific

for different

neuronal

by 70%

These

to

results

alterations

measurement

by

in

results

1102

Y. Kawasaki et al.

In BA 9, Selemon IS in contrast different

et al. (1995)

with

our finding.

cytoarchitectural

reduction

of the

increase,

whereas

et al. (1995)

ratio

demonstrated However,

changes. between

lending

we studied

It might

cell body

BA 9 demonstrates support

an overall

to the

and neuropil

“reduced

in neuronal

BA 10, which

be possible

an increase

increase

that

could

density,

well

demonstrate

BA 10 demonstrates

volumes

in neuronal

indicating

density

hypothesis”

changes

in the prefrontal

an overall

a relative

as reported

neuropil

whrch

(Selemon

neuropil

by Selemon

and Goldman-

Rakic 19991. Another from

line of evidence

studies

Bizzozero

on the

density

of synaptic

found

a decreased

et al. (3996)

and 20. These

results

and could for instance synaptic

protein

for cytoarchitectural

proteins

immunoblotting

are also compatible reflect

a reduced

in the measuring

as assessed

with

by immunoblotting.

signal for synaptophysin

a cytoarchitectural

neuron

number

lobe may be drawn

going

change

along

with

Perronein BA 9, IO

in schizophrenics a reduction

in total

field.

Conclusion The

results

with

cytoarchitectural in principle 1995,

1998,

scanning

an

studies

automated

Rajkowska

analysis

system

in BA 10 (Benes et al., 1986,

also compatible

tool for future

image

with

reports

et al., 1998).

in other It thus

cytoarchitectural

studies

are

1991,

prefrontal

recommends

in

good

Rajkowska Brodmann the

concordance

with

et al. 19991 and are areas

(Selemon

use of the GLI method

et al., as a

on schizophrenia.

Acknowledaements Supported Program

by the

Theodore

of the European

and Wada

Stanley

Foundation,

the

DFG,

and the

Biomed

2

Union.

References ANDREASEN NC, NASRALLAH HA, DUNN Y, OLSON SC, GROVE WM EHRHARDT JC, COFFMAN JA and CROSSETT JHW (1986) Structural abnormalities in the frontal system in schizophrenia. Arch Gen Psychiatry a, 136-144 BEASLEY CL and REYNOLDS GP (1997) Parvalbumin-immunoreactive the prefrontal cortex of schizophrenics. Schizophr Res 2% 349-55

neurons

BENES FM, DAVIDSON J and BIRD ED (1986) Quantitative cytoarchitecotnical cerebral cortex of schizophrenics. Arch Gen Psychiatry $3, 31-35

are reduced

studies

in

of the

BENES FM, MCSPARREN J, BIRD ED, SANGIOVANNI JP and VINCENT SL (1991) Deficits in small interneurons in prefrontal and cingulate cortices of schizophrenic and schizoaffective patients. Arch Gen Psychiatry 44, 996-1001 FALKAI P, HONER WG, DAVID S, BOGERTS B, MAJTENYI C and BAYER TA (1999) No evidence for astrogliosis in brains of schizophrenic patients. A post-mortem study. Neuropathol Appl Neurobiol 25, 48-53

Cytoarchitecture of Brodmann area 10 in Schizophrenia HARRISON

PJ (1999)

their interpretation. HUDSPETH

AJ,

microdensitometry.

The neuropathology Brain m,

RUARK

of schizophrenia.

review

of the data and

593-624

JE

and

KELLY

Proc Natl Acad Sci USA I&

MERKER B (1983) Silver Methods 9, 235-241

A critical

1103

staining

of cell bodies

JP

(1976)

Cytoarchitectonic

mapping

by

2928-2931 by means of physical

NURNBERGER JI (1958) Direct enumeration of cells of the brain. Windle (Ed.), pp 193-202, Thompson, Springfield.

development.

In: Biology

PAKKENBERG B (1993) Total nerve cell number in neocortex in chronic controls estimated using optical disectors. Biol Psychiatry 3% 768-772

J Neuroscr

of Neuroglia,

W.F.

schizophrenics

and

PERRONE-BIZZOZERO NI, SOWER AC, BIRD ED, BENOWITZ LI, IVINS KJ and NEVE RL (1996) Levels of the growth-assocrated protein GAP-43 are selectively increased in association cortices In schizophrenia. Proc Natl Acad Sci USA $3, 14182-I 4187 RAJKOWSKA G, SELEMON LD and GOLDMAN-RAKIC in the prefrontal cortex. A postmortem morphometric drsease. Arch Gen Psychiatry s, 215-224

PS (19981 Neuronal and glial somal size study of schizophrenia and Huntington

RAJKOWSKA G, WIE J, MIGUEL-HIDALGO JJ and STOCKMEIER CA (1999) pathology In rostra1 orbitofrontal cortex in schizophrenic postmortem Schizophrenia Res 36, 84

Glial and neuronal brain (abstract).

SCHLAEPFER TE, HARRIS GJ, TIEN AY, PENG LW, LEE S, FEDERMAN BARTA PE and PEARLSON GD (1994) Decreased regional cortical gray schtzophrenia. Am J Psychiatry 151, 842-848

EB, CHASE GA, matter volume In

SCHLEICHER A, ZILLES K and WREE A (1986) A quantitative apporach to cytoarchitectonics: software and hardware aspects of a system for the evaluation and analysis of structural rnhomogeneities in nervous tissue. J Neurosci Meth 18, 221-235 SCHLEICHER of structural 367-381

A and ZILLES K (1990) A quantitative approach to cytoarchitectonics: analysts Inhomogeneities in nervous tissue usrng an image analyser. J Microscopy m,

SCHLEICHER A, AMUNTS K, GEYER S, MOROSAN P and ZILLES K (1999) ObserverIndependent method for microstructural parcellation of cerebral cortex: a quantitative approach to cytoarchitectonics. Neurolmage 9, 165-177 SELEMON LD and GOLDMAN-RAKIC PS (1999) The reduced based model of schizophrenia. Biol Psychiatry 45, 17-25 SELEMON LD, RAJKOWSKA G and GOLDMAN-RAKIC density in the schizophrenic cortex: a morphometric occiprtal area 17. Arch Gen Psychiatry z, 805-818

neuropil

PS (1995) analysis

hypothesis:

a circuit

Abnormally hrgh neuronal of prefrontal area 9 and

SELEMON LD, RAJKOWSKA G and GOLDMAN-RAKIC PS (1998) Elevated neuronal density in prefrontal area 46 in brams from schizophrenic patients: application of a three-dimensional, sterological counting method. J Comp Neurol 392, 402-412 SHELTON RC, DORAN AR, pathology in schizophrenia. Psychiatry 145, 154-I 63

PICKAR D and WEINBERGER DR (1988) Cerebral structural Evidence for a selective prefrontal cortical deficit, Am J

TOWER DB and YOUNG OM (1973) The activities of butyryl cholinesterase and carbonrc anhydrase, the rate of anaerobic glycolysis, and the question of a constant density of glial cells in cerebral cortices of various mammalian species from mouse to whale. J Neurochem 251, 269-278 VOGELEY K and FALKAI P (1998) Psychiatry Brain Res 6, 1 13-I 22

The dysconnectivity

hypothesis

of schizophrenia.

Neurol

Y. Kawasaki et al.

1104

VOGELEY K, KURTHEN M, FALKAI P and MAIER W (19991 Essential Functions Self Model are Implemented in the Prefrontal Cortex. Consc Cogn 8, 343-363

of the Human

VOGELEY K, SCHNEIDER-AXMANN T, PFEIFFER U, TEPEST R, BAYER TA, BOGERTS B, HONER WG and FALKAI P (2000) Disturbed gyrification of the prefrontal region in male schizophrenic patients - a morphometric postmortem study. Am J Psychiatry l.5.Z, 34-39 WEINBERGER DR (1996) On the plausibility schizophrenia. Neuropsychopharmacology 14,

of “the l-l 1

neurodevelopmental

WREE A, SCHLEICHER A and ZILLES K (1982) Estimation tissue with an image analyzer. J Neurosci Meth 6, 29-43

of volume

fractions

WREE A, ZILLES K and SCHLEICHER A (1980) Analyse der laminlren Struktur mit verschiedenen sterologischen MeRmethoden. Verh Anat Ges 14, 727-728

Inquiries

and reprint

requests

Dr. Kai Vogeley Department of Psychiatry University of Bonn Sigmund-Freud-Str. 25 53105 Bonn Germany

should

be addressed

to:

hypotheses”

of

in nervous

der Area striata