вход по аккаунту


Amygdala cell loss and atrophy in Alzheimer's disease.

код для вставкиСкачать
Amygdala Cell Loss and Atrophy
in Alzheimer's Disease
Samuel A. Scott, PhD,'§ Steven T. DeKosky, MD,? D. Larry Sparks, PhD," Craig A. Knox, MD, PhD,$
and Stephen W. Scheff, PhD"
The amygdala and its subnuclei undergo severe volumetric atrophy in Alzheimer's disease (AD). To determine whether
this atrophy is due to loss of neuropil, specific neuronal populations, or both, we evaluated the number, size, and
packing density of neurons and glia in the cortical and magnocellular basal amygdaloid subregions. The neuropil
fraction did not change with AD in either region. Despite a mean 35% increase in cell packing density in the AD
amygdala, total numbers of neurons and glia within tissue sections were reduced significantly; medium and large
neurons were preferentially affected. The total number of small neurons was stable in the AD sample despite sharp
reductions in nuclear size, suggesting that AD also results in pronounced amygdaloid neuronal shrinkage. Differences
in the degree of cell loss between the two nuclei as well as changes in glial cell numbers are discussed in relation to
characteristic AD neuropathology and relevant anatomical connectivity.
Scott SA, DeKosky ST, Sparks DL, &ox CA, Scheff SW. Amygdala cell loss and atrophy
in Alzheimer's disease. Ann Neurol 1992;32:555-563
The amygdala, a primary limbic structure anatomically
interconnected with neocortex, hippocampus, entorhinal cortex, nucleus basalis of Meynert (nbM), and other
areas {l,2 ) , is severely affected in Alzheimer's disease
(AD) {3-91. One of the many proposed functions of
the amygdala involves the assignment of emotional significance to memories 1101, underscoring the clinical
relevance of amygdaloid disruption in AD.
Several studies have demonstrated volumetric atrophy in key subregions of the amygdala as a function of
AD 111, 121. It is unclear what specific changes at the
cellular level are responsible for this shrinkage. The
amygdala has been reported to lose neurons in AD
113, 141, yet only a single investigation has provided
quantitative evidence to support this conclusion 111).
Alterations in the glial cell population were not assessed in that study, and it is unknown whether amygdaloid neurons shrink in size, as has been documented
in other brain areas in AD 115, 161. Although neuronal loss and atrophy may be sufficient to result in the
observed amygdaloid involution with AD, regressive
changes restricted to the neuropil might also be involved 117). In light of these considerations, the present study assessed possible AD-related changes in the
number and size of neurons and glia as well as alterations in the neuropil fraction within amygdaloid subregions.
Materials and Methods
From the 'Departments of Anatomy and Neurobiology, and Neuroloev and Sanders-Brown Center on Aging. Universitv of Kentucky
Medical Center, Lexington, Ky; and tg <Departments ofPsychiat&
and Neurology, University of Pittsburgh School of Medicine, Pittsburgh, PA; and $Associates in Neurology, Lexington, KY.
§Present address: Department of Neurosurgery, University of Cincinnati School of Medicine, Cincinnati, OH 45267-0515.
Subjects and Tissue Preparation
Autopsy tissue from 5 men and 1 woman (mean age, 73
years) without history of neurological or cognitive abnormality and free of heart disease [IS] served as control samples.
Gross anatomical and microscopic observations failed to reveal any substantial neuropathological changes within these
samples. The AD group consisted of 6 age-equivalent subjects (4men, 2 women), assessed using standard National
institute of Neurological Disorders and Stroke-Altheimer's Disease and Related Disorders Association (NINDSADRDA) criteria [19, 201 (Table 1).
At autopsy, the right cerebral hemisphere was immersed
in 10% neutral buffered formalin for 1 to 2 months. Coronal
blocks containing the amygdala and adjacent temporal cortex
were sectioned 3 to 5 mm thick and embedded in paraffin.
Embedded blocks were then coded, sectioned 15 (*.mthick,
and stained with cresyl violet. The magnocellular basal nucleus (BNMC; basolateral nucleus in subprimates) and the
deep portion of the cortical nucleus (DCN) were selected
for analysis. A section near the rostrocaudal midpoint of the
amygdala was examined using a projection microscope to
locate the BNMC and DCN, adhering to the anatomical
boundaries previously described [2 1). Each nucleus was outlined and its cross-sectional area determined. Subsequent
quantitation of cell size and packing density was accomplished utilizing a Kontron image processing system (IPS,
Munich, Germany) coupled to an Olympus Vanox microscope (Lake Success, NY).
Received J~ 24, 1992. Accepted for publication Apr 12, 1992.
Address correspondence to D r Scheff, 31 1 Sanders-Brown Center
on Aging, University of Kentucky, Lexington, KY 40536-0230.
Copyright 0 1992 by the American Neurological Association 5 5 5
Table I . Subject Information
Mean f. SEM
Mean f. SEM
73 t 4.4
77.5 f. 2.9
11.3 f 1.89
6.8 k 1.9
1,310 f. 19.87
1,090b f. 36.3
12.45 2 1.35
4.52' t 0.82
4.5 +- 0.52
2.5b 1.16
Nuclear Area (mm')
of Illness
9.5 t 2.2
"PMI indicates the time between death and immersion of tissue in fixative.
b p < 0.01.
' p < 0.005.
BNMC = magnocellular basal nucleus; DCN = deep portion of cortical nucleus; SEM = standard error of mean.
Manual Identification of Cell Type & Size
Procedure for Distinguishing Neurons from Glia
In studies using semiautomatic methods to count Nisslstained cortical cells, neurons are distinguishable from glia
based on size [Ib, 22-24}, and the automated counts correlate with manual methods 124, 251. The "cutoff" value used
to distinguish neurons from glia in the present study was
derived as follows. All cells within 12 randomly chosen fields
per subject (6 per nucleus) were studied under 100 x magnification with oil immersion. A cell containing abundant basophilic cytoplasm surrounding a euchromatic nucleus was
scored as a neuron, regardless of the presence of a nucleolus.
Large cells containing basophilic cytoplasm but lacking a distinct nucleus in the plane of section were also scored as neurons. Glia were identified primarily by the absence of stainable cytoplasm. Astrocytes could be distinguished by a patchy
appearance of the nuclear chromatin. The remaining cells,
uniformly smaller and more basophilic, were presumably oligodendroglia and microglia 1261. N o attempt was made to
differentiate between the latter two types of glia.
The cross-sectional area for each profile (n = 3,140) was
determined and scored as a neuron or glial cell. Area frequency histograms were then generated based on the entire
sample (Fig 1). Cells smaller than 10 pm2 comprised less
than 1% of the sample, whereas many fell within the range
of 10 to 15 pmZ.Therefore, 10 pmZ was employed as the
lower limit of detection, while the intersecting point of the
neuronal and glial size curves served as the cutoff value
throughout the procedure described below.
Cell Counts
Employing the same sections, an automated stepping stage
was used to analyze each amygdaloid nucleus as a series of
nonoverlapping images projected onto the computer screen
(image size = 0.09 mm2).Every image was selected provided
556 Annals of Neurology
Vol 32 N o 4
October 1992
4405 1
25 20 1510-
21-30 31-40 41-50 51-60 61-90 91-150151-250,250
Cross Sectional Area pm'
Fig 1. All cells within random samples of the magnocellular
basal nucleus and deep portion of the cortical nucleus were manually identified as either glial or neuronalfar both normal aged
control and Alzheimer (AD) subjects. Each curve represents the
percentage of the size range for these cells. Almost all of the glial
cells far both control and A D were less than or equal t o 50 &
in cross-sectional area and almost all of the neurons were found
to be larger than 50 p d . This value was used to automatically
discriminate neurons from glia during subsequent cell counting.
the nuclear border was not traversed, resulting in an average
of 38 images (30 minimum) analyzed per nucleus in each
case. Extensive image editing was performed to include: (1)
separating contiguous cells, (2) filling in cells whose staining
density lay below the threshold gray value required for automated detection, and (3) elimination of vascular structures
and artifacts. These steps are essential for accurate assessment
of cell number and size using semiautomated techniques
I27). Objects were discriminated from the background based
on agray value derived separately for each image, minimizing
the impact of any differences in staining intensity among the
slides. Packing density (cells/mm2), areas, and diameters of
approximately 149,000cells were determined in this fashion.
The Abercrombie equation was used to correct for the overcounting of cells split by the microtome E28). The total number of neurons and glia in each section was then determined
by multiplying cell packing densities by the nuclear area.
Individual patient values were handled using a Macintosh
computer (Apple Computers, Cupertino, CA) and several
statistical programs (SUPER ANOVA & Statview 512, Abacus Concepts, Berkeley, CA). Group comparisons were performed using two-tailed t tests, regression analyses, and Pearson product-moment correlation coefficients. t Test results
are reported as two-tailed unless indicated otherwise.
The mean weight of the AD brain prior to fixation
was significantly reduced by 18% ( t { l O ] = 5.32, p <
0.001) (see Table 1). Additional t tests revealed no
group differences in subject age or postmortem delay,
nor was there a correlation between age and brain
weight. The size of the two regions of the AD amygdala, however, showed a highly significant decrease
(see Table 1). Cross-sectional areas of the BNMC and
DCN were reduced to 36% (t(l0)= 5.02, p < 0.001)
and 55% of control values (t(l0)= 2.85, p < 0.01),
respectively. There were no correlations among age,
postmortem delay, brain weight, or duration of illness
and any of the cell count measures.
Manual Cell Ia'entijEcation
The cross-sectional area of 50 pm2 (approximately 8
km in diameter) was selected as the break point to
discriminate glia from neurons (see Fig 1). Almost all
morphologically identified glia fell within the range of
10 to 50 km2, leaving only 2% of the control and 6%
of the A D glia outside of this range. Oligodendrogiia
and microglia fell primarily within the 10- to 25-pm2
range and were rarely observed to exceed 30 pm2,
whereas the majority of cells resembling astrocytes
were at least 30 pm2 in area and often extended into
the 41- to 50-pm2 range. Cells in the 41- to 50-pm2
range comprised more than 17% of total glia in AD
compared with only 6% in the control group (Fig 1).
The percentage of morphologically identified neurons
having an area equal to or below 50 pm2 was less than
1% in the control and less than 4% in the AD sections.
Total Cell Packing Density
A two-way repeated measures analysis of variance
(ANOVA) was used to test for overall effects of AD,
cell type, and amygdaloid nucleus (cell type nested
within nucleus; nucleus as the repeated measure) on
cell packing density [29f. AD resulted in significantly
increased cell densities ( F ( 1 , l O ) = 5.28,p < 0.05) and
altered the proportions of cell types (neuronal versus
glial) ( F ( 1 , l O ) = 130.99, p < 0.0001). The region of
the amygdala also had a significant effect (F(1,lO) =
4.41, p < 0.06) that was independent of AD, while
region and cell type effects were significantly interactive (F(1,lO) = 7.15, p < 0.02) (Table 2).
These data permitted further analyses of the separate
effects of AD and cell size on the packing density
within each nucleus. There was a significant AD-related increase of 35% in overall cell density (t(10) =
2.23, p < 0.05) within both amygdaloid regions. After
partitioning cells into the general categories of neurons
and glia, t tests revealed AD-related increases in the
densities of each cell type within each nucleus (see
Table 2).
Total Cell Number
The total number of each cell type within the section
was analyzed as above. A two-way repeated measures
ANOVA revealed significant overall effects of AD
(F(1,lO) = 13.7,p < 0.005), nucleus (F(1,lO) = 64.9,
p < O.OOOl), and cell type (F(1,lO) = 87.9, p <
0.0001) on total cell number. Significant interactions
were noted between nucleus and AD ( p < 0.005), cell
type and AD ( p < 0.05), and nucleus and cell type
( p < O.OOOl), and a three-way interaction was noted
between AD, nucleus, and cell type ( p < 0.005). The
effects of AD on cell number were thus dependent, in
part, on the nucleus and type of cell being measured.
Subsequent t tests disclosed significant AD-related declines in both total glia and total neurons within each
subnucleus, with the exception of glial cells in the
DCN (Fig 2).
While the BNMC appeared to have greater overall
cell loss ( 5 1 and 54% reductions in glia and neurons,
respectively), the DC N showed a preferential loss of
neurons (41%) versus glia (24%). In order to substantiate these comparisons, total cell number for each AD
subject was evaluated as a percentage of the respective
control mean. t Tests confirmed that the BNMC lost
the most cells overall (t(10) = 2.92, p < 0.01). While
the extent of neuron loss did not differ significantly
between the two nuclei ( p > O . l ) , the reduction in
glial numbers was greater in the BNMC (t(10) = 3.9,
p < 0.005).
Glial Changes
PACKING DENSITY. Glia were subsequently divided
into small (10-30 pm2) and large (31-50 km2) cell
Scott et al: Amygdala Cell Loss in AD
Table 2. Cell Packing Density (cehlmn?)"
10-30 pm
31-50 pm
51-90 pm
L,lOOb f 98
205 t 7.9
296 t 28
6 4 5 t 98
803 t 84
181 2 11
2 9 7 ' ~26
157b f 19
49 5
80 t 17
691 f 78
l,OOOc t 116
253 18
280 t 43
507 t 81
619 f 75
184 2 15
38Ib 2 49
95 f 11
194' f 37
135 14
78d t 12
* 93
91-200 pm
201-300 p m
>300 pm
48 t 5
42 t 6
44 t 10
17' t 5
22 t 4
7d t 2
* 0.6
* 0.4
"Data are means 5 standard error of the mean. The two-tailed t test was used to determine level of significance.
bp < 0.005.
'p < 0.05.
dp < 0.01.
BNMC = magnocellular basal nucleus; DCN = deep portion of cortical nucleus.
size data on the glial populations confirmed this finding
by manifesting a rightward shift of the glial size curve
in AD (see Fig 1).
Assessment of glial cell loss according
to size by ANOVA revealed significant effects of AD
(F(1,lO) = 11.7, p < O.OOl), amygdaloid nucleus
(F(1,lO) = 63.9, p < O.OOOl), and glial size (F(1,lO)
= 40.5, p < 0.0001). t Tests revealed AD-related declines for each size category in the BNMC ( p < 0.05)
whereas the D C N showed loss of only small glia. It
should be noted that despite substantial AD-related
atrophy of the DCN (see Table l ) ,a moderate increase
in the total number of large glia was observed in sections of this nucleus (Fig 3A).
Cell Classification
Fig 2. The mean number (* standard error of mean) of glia
and neurons per section, as determined by automated counting,
plotted for each subregion of the amygdala used in this study.
'p < 0.01 (t test).
size groups. A repeated measures ANOVA again revealed a significant effect of AD (F(I, 10) = 5.15, p <
0.05). There was also an effect of gIial size (F( 1,lO) =
49.4, p < 0.0001) and an interaction between region
and glial size (F(1,lO) = 25.02, p < 0.0005). Post
hoc comparisons revealed no AD-related change in the
density of small glia in either nucleus (t(10) = 1.23,
p > 0.1). In contrast, the density of large glia was elevated in AD by an average of 86% between the
BNMC (t(l0)= 4.06, p < 0.005) and DCN (t(10) =
3.87, p < 0.005) (see Table 2). The manually obtained
558 Annals of Neurology Vol 32
No 4 October 1992
Neuronal Changes
PACKING DENSITY. Neurons were also classified by
size: 5 1 to 90, 91 to 200, 201 to 300, and larger than
300 pm2 (see Table 2). A repeated measures ANOVA
failed to detect an effect of AD on cell packing density
after such partitioning, although there was a neuronal
size effect (F(3,lO) = 46.33, p < 0.001) as well as
size-by-group (F(3,lO) = 12.76, p < 0.001) and sizeby-region interactions ( F ( 3 , l O ) = 14.2, p < 0.001).
Post hoc comparisons revealed AD-related increases in
the density of small neurons (51-90 pm') for both the
< 0.005) and DCN (t(10)==
BNMC (t(10) = 4 . 4 1 , ~
2.56, p < 0.05). Conversely, the packing density of
larger neurons decreased in AD. The DCN showed
AD-related density declines in the 91- to 200-pm2
(t(l0) = 3.06, p < 0.01) and 201- to 300-pm2 categories (t(10) = 3.41, p < 0.01). Packing density in the
BNMC was reduced significantly only in the largest
size category (> 300 pm2) (t(10) = 2.44, p < 0.05).
pi T
Cell size pm2
’“I T
50-90 90-200 200-300 ,300
50-90 90-200 200400 ,300
Cell Size pm2
Fig 3. Mean cell numbers per section in the magnocellular basal
nucleus (BNMC) and deep portion of cortical nucleus (DCN)
of the amygdala plotted as a function of cell size. (A) Glia.
(B) Neurons. Each bar represents the mean (+- standard ewor
of mean). *p < 0.05; **p < 0.01 (t test).
A two-way repeated measures ANOVA applied to total neuron number revealed significant effects of AD (F(1,lO) = 16.8, p < 0.005), nucleus (F(1,lO) = 30.0, p < O.OOl), and neuron size
(F(3,lO) = 39.6, p < 0.0001). Significant interactions
were noted between AD and neuron size ( p < 0.05)
as well as between nucleus and neuron size ( p < 0.05).
Subsequent t tests indicated severe AD-related reductions in neurons larger than 90 pm2, in particular those
larger than 200 pm2 (Fig 3B). N o significant differ-
ences were detected for the smallest neuron category
in either nucleus. The most significant effects were thus
attributed to loss of the largest neurons and/or shrinkage of various neuronal populations.
In order to determine whether neurons or giia were
preferentially affected in AD, the percentage of total
cells that were glia in each case was statistically evaluated. This parameter is an alternative expression of the
“ratio of neurons to &a” [16]. Glial cells tended to
comprise 70 to 80% of the total cell number in both
nuclei, and the percentages did not significantly change
with AD.
To assess whether AD-related nuclear atrophy was
due primarily to cell or to neuropil loss, the summed
area of all cells per field was divided by the image
size and evaluated. This parameter is analogous to the
“neuropil ratio” calculated by Terry and coworkers
El61 and represents the fraction of brain tissue occupied by Nissl-stained neuronal and glial somata. Cells
comprised approximately 8% of the field area in DCN
and 10% in BNMC, with no change observed in AD
despite the increased number of cells per image. Multiplying these percentages by the area of the nucleus
resulted in equal AD-related declines in the somal and
neuropil compartments within the section.
The present study demonstrates that AD results in a
markedly increased cell packing density in the amygdala, an increase that includes both neurons and glia.
When the data are corrected for gross structural atrophy, tissue sections of the amygdala actually contain far
fewer neurons and glia overall in AD compared with
age-matched control subjects (see Fig 2). Assessment
of this loss according to cell size (see Fig 3) shows that
the declines are greatest for medium and large neurons, a finding consistent with studies of neocortical
cell loss in AD [30f. Further support for large neuron
loss in the amygdala derives from the cell packing density data (see Table 2); the density of the largest neurons is either stable or reduced in a severely atrophied
structure. The functional consequences of such severe
damage to the arnygdala in AD are unknown, although
it is reasonable to conclude that they relate to the emotional, motivational, and behavioral as well as cognitive
and memory impairments observed in AD patients.
Neuronal Shrinkage in AD
The present findings of AD-related cell loss despite an
increased cell packing density are similar to those of
Hubbard and Anderson {31) in their morphometric
study of the subicular, basofrontal, and superior temporal cortices in dementia. They showed that when
considering only the numerical density of areal fraction
of cells per image, small neurons and glia appeared
to be increased in AD samples while large neurons
Scott et al: Amygdala Cell Loss in A D 559
showed no change compared with control samples.
Correction of the data for cortical atrophy revealed
that the glial cell fraction was only moderately increased and numbers of small neurons remained stable,
whereas the fraction of large neurons was reduced by
26 to 33%. Terry and colleagues 1161, in a study of
superior temporal and midfrontal cortices in dementia,
similarly showed that changes in cell packing density
assessed per area do not necessarily reflect changes in
cell number. Increased cell packing density can thus be
very misleading without knowledge of alterations in
either the area or volume of the structure of interest.
Several mechanisms could account for the AD-related increase in cell packing density despite overall
cell loss. A recent study showed that volumes of the
amygdaloid BNMC and DCN were reduced in AD by
69% and 4796, respectively {12l. A similar analysis
of nuclear area using the present material resulted in
AD-related declines of similar magnitude (BNMC,
64%; DCN, 45%). Studies from other laboratories
also demonstrated significant AD-related involution of
the amygdala {8, 11, 321, often in conjunction with
lateral ventricular dilation 132, 331. Therefore, one explanation for the increased amygdaloid cell packing
density is that the cells are condensed into a smaller
volume, implying a disproportionate loss or collapse of
the neuropil. This effect of AD was not seen in one
study of superior temporal and midfrontal cortices
1161, where true total volume is very difficult to assess.
Another study suggested that the neuropil fraction in
AD temporal cortex does decline sharply (311. In the
present study, the fraction of tissue occupied by amygdaloid cell bodies did not change with AD. That is, 90
to 92% of the field was neuropil both in dementia
and in normal aging. The most reasonable explanation
for this fractional stability is that both amygdaloid cells
and their surrounding neuropil are degenerating proportionally. However, this fraction fails to change with
AD despite a 35% increase in cell packing density. The
quantity of neuropil corresponding to one cell must
therefore decrease. The disproportional loss of the
largest neurons, which have the largest axodendritic
expanse, may be an explanation for the preservations
of the neuropil-cell body ratios.
The present results confirm those in a previous report demonstrating AD-related amygdaloid nerve cell
loss in which maximal declines were observed in the
medial, central, and cortical nuclei {ll].The packing
densities reported in that study were strikingly low,
however, with mean values ranging from 20 to 391
neurons/mm3 (in contrast to the 13,700 to 19,700 neurons/mm3 presently calculated). Use of a traditional
manual counting procedure 1281, compared with the
presently used automated method, may have resulted
in the detection of fewer neurons. The possibility of
AD-related neuronal shrinkage may also have been re5 6 0 Annals of Neurology Vol 32 No 4 October 1992
sponsible for the low values reported. Nevertheless,
both studies report severe neuronal loss in the amygdala in AD, iacluding similar magnitudes of loss.
In addition to cell and neuropil loss, the present
data support the concept that many large amygdaloid
neurons shrink in AD and are thus counted as small
neurons. Figure 1 graphically displays an AD-related
change in the distribution of cell size for both glia and
neurons, with a shift toward smaller neurons and larger
glia. These trends are supported by the data on cell
packing density which show that in both nuclei, AD-related reductions occur in the density of large neurons
while the density of small neurons is increased (see
Table 2). The shift in density from large to small neurons could result from a selective loss of large neurons,
neuronal shrinkage, or both 1341. However, the total
number of small neurons in each nucleus appears to be
stable in AD despite dramatic losses in nuclear volume.
Stability in their total numbers could only be maintained through shrinkage of larger neurons or loss of
large neurons exclusively. While it is very difficult to
ascertain exactly what percentage of the small neuronal
population results from such shrinkage, a previous
study of peptidergic neurons in the amygdala in dementia supports the presence of increases in small neuron numbers as a result of large neuron shrinkage 171.
Similar findings of neuronal shrinkage with AD have
been documented in other central nervous system
structures. Both the superior temporal and midfrontal
cortices showed a decrease in the density of large neurons while the overall neuronal density was unchanged
1161. Shrinkage has been observed in other neocortical regions 131, 351 as well as in nbM 1151 and in
epinephrine-containing nuclei of the upper spinal cord
1361. It is unknown whether neuronal shrinkage actually interferes with cellular functioning. Shrinkage of
the neuronal cell body in temporal cortex, however,
coincides with 20 to 40% atrophy of the cell nucleus
and up to 25% loss of cytoplasmic FWA 1371. One
may thus speculate that neuronal shrinkage is associated with diminished protein synthetic capability of the
cell and hence its ability to maintain functional integrity.
Neuron-Glia Size Cutoff
The present study employed a cross-sectional area of
50 p,m2 to differentiate neurons from glia. A similar
value (52 Fm2) was used as a neuron-glia cutoff for
subicular, superior temporal, and basofrontal cortices
in the study by Hubbard and Anderson [31J Size
ranges used to classify cells are extremely important in
studies employing automated techniques, particularly
when trying to distinguish neurons from glia. There is
no standard cell size to differentiate glia from neurons.
In fact, cutoff values ranging from 40 to 120 pm2 have
been used in studies of closely adjacent and even the
same cortices 116,22,23,38). The importance of accurate and empirical determination of such criteria is thus
readily appreciated. Use of an arbitrarily low value
would result in a disproportionate number of large glia
being counted as small neurons, whereas use of an
excessively high value would result in the misclassification of small neurons as glia. In the present study, the
slight overlap in size between the smallest morphologically identified neurons and the largest glia (see Fig 1)
allows for the possibility that during the actual counting, a few neurons were mistaken for glia and vice
versa. This degree of error had little effect, if any, on
the major findings of the investigation, since the overlaps were far smaller than the changes in calculated
numbers of cells.
Astroqte Hypertrophy
In the neocortex, numbers of glial fibrillary acidic protein-positive astrocytes have been found to be increased fourfold in AD while total glial counts remain
constant 1391. Marked astrogliosis in the amygdala in
AD has been described several times 113, 401, but
without quantitative assessment. In the present study,
the AD amygdala showed a significant increase in the
numerical density of large glia, with astrocytes appearing to comprise the majority of cells in this size
range. This relative increase probably resulted from
both gliosis and dial hypertrophy 141, 421. Another
interesting finding is the 7% AD-related increase in
the total number of large glia observed in the DCN.
Numerous studies have noted exceptionally high numbers of mature senile plaques within this region in AD
16, 13, 431, and astrocytes are commonly associated
with this type of plaque 144, 451.
The BNMC and DCN were selected for study because they differ in several aspects that are relevant to
their involvement in AD. In primates the DCN is one
of the few structures receiving both direct and indirect
olfactory input 1461, while the BNMC is heavily interconnected with neocortex 1461 and nbM {47}. These
anatomical pathways are each thought to be affected in
AD but possibly at different times in the course of
the disease 148). Choline acetyltransferase activity is
severely reduced in the BNMC in AD, indicating substantial disruption of its basal forebrain input 149-5 13.
However, the same nucleus is almost completely lacking in senile plaques, particularly in comparison with
the DCN [GI. Because of these differences, the two
nuclei were studied independently and compared.
Although the amygdaloid BNMC and DCN both
demonstrate cell loss in AD, the losses vary in both
degree and specificity. The variability results from (1)
greater overall shrinkage of the BNMC with AD and
(2) elevated gliosis in the DCN. Extensive volumetric
atrophy of the BNMC with AD [S, 121 coincides with
its greater degree of cell loss, whereas excessive plaque
formation in the DC N likely accounts for its relatively
increased density of large glia. These data suggest that
the regional specificity of neuron loss does not coincide with that of plaques in the AD amygdala 161,
which contradicts Herzog and Kemper’s 1111 finding
of greater AD-related cell loss in the conicomedial nuclei. Combined with earlier information on the distribution of tangles and plaques, their results suggested
that the developmentally and phylogenetically older
corticomedial nuclei were preferentially affected in AD
Clll. More recent data on the distribution of plaques,
tangles, and nuclear atrophy within the amygdala in
AD fail to support this concept. For example, numerous plaques can be found in the basolateral amygdala
by use of a modified silver stain 1521, and Brady and
Mufson {S] noted an abundance of “primitive” plaques
in the basal and lateral nuclei. Most neuritic plaques
are generally localized within the ventromedial region
of the amygdala 1531, which includes not only the cortical nucleus but also the paralaminar and parvocellular
regions of the basal nucleus, the accessory basal nucleus (both of the basolateral group), and the corticoamygdaloid transition area. The same study also reported that the distribution of A h 50-immunoreactive
tangles appears to include most if not all regions of the
amygdala in AD 1537.
These data suggest that AD affects the amygdala in
a topographically consistent manner involving both the
corticomedial and basolateral groups. However, ADrelated disruption of the more recently evolved basolateral nuclei would seem more relevant to cognitive
dysfunction. Anatomically the primate BNMC is characterized by prominent projections to neocortical areas, in particular the frontal and temporal lobes C541.
A decrease in this input resulting from cell loss could
contribute to the observed decline in synapse numbers
within those cortical regions in AD 155-581. Both the
cortical and basal amygdaloid nuclei have connections
with various regions of the medial temporal lobe such
as entorhinal cortex, subregions of the subiculum, Ammon’s horn, and the dentate gyrus 12, 541. The extensive and complex interconnections between these
regions underscore the fact that they are all pathologically disrupted in AD. In fact, the amygdala may
be effectively “disconnected” from key brain regions
in a manner analogous to that described for hippocampus 159, GO].
This work was supported in part by National Institutes of Health
grant AGO5144.
We thank Douglas Price for technical assistance.
1. Schwarct R, Ben-Ari Y, eds. Excitatory amino acids and epilepsy. New York: Plenum, 1986
2. Saunders RC, Rosene DL, Van Hoesen GW. Comparison of
Scott et al: Amygdala Cell Loss in AD
the efferents of the amygdala and hippocampal formation in the
rhesus monkey: 11. Reciprocal and non-reciprocal connections.
J Comp Neurol 1988;271:185-207
3. Grunthal E. Zur hirnpathologischen Analyse tier Alzheimerschen Kankheit. Psychiatry Neurol Wochenschr 1928;36:401407
4. Yamada M, Mehraein P. Verteilungstmunster der senilen veranderungen in gehirn. Die Beteiligung des limischen Systems
bei hirnatrophischen Prozessen des Seniums und beu Morbus
Alzheimer. Arc:h Psychiatr Zeit Neurol 1968;2 11:308-324
5. Kemper TL. Organization of the neuropatholoby of the m y g dala in Alzheimer’s disease. In: Report B, ed. Biological aspects
of Alzheimer’s disease, vol 15. New York: Cold Spring Harbor,
6. Brashear HR, Godec MS, Carlsen J. The distribution of neuritic
plaques and acetylcholinesterase staining in the amygdala in Alzheimer’s disease. Neurology 1988;38: 1694- 1699
7. Unger JW, McNeill T H , Lapham LL. Neuropeptides and neuropathology in the amygdala in Alzheimer’s disease: relationship
between somatostatin, neuropeptide Y and subregional distribution of neuritic plaques. Brain Res 1988;452:293-302
8. Brady DR, Mufson EJ. Amygdaloid pathology in Alzheimer’s
disease: qualitative and quantitative analysis. Dementia 1990;
9. Kromer-Vogt LJ, Hyman BT, Van Hoesen GW, Damasio AR.
Pathological alterations in the amygdala in Alzheimer’s disease.
Neuroscience 1990;37:377-385
10. Sarter M, Markowitsch HJ. The amygdala’s role in human mnemonic processing. Cortex 1985;21:7-24
11. Henog AG, Kemper TL. Amygdaloid changes in aging and
dementia, Arch Neurol 1980;37:625-629
12. Scott SA, DeKosky ST, Scheff SW. Volumetric atrophy of the
amygdala in Alzheimer’s disease: quantitative serial reconstruction. Neurology 1991;41:35 1-356
13. Brockhaus H. Zur normalen und pathologischen Anatomic des
mandelkerngebietes. J Psycho1 Neurol 1938;49:1-1 36
14. Corsellis JAN. The limbic areas in Alzheimer’s disease and in
other conditions associated with dementia. In: Wolstenholme
GEW, ed. Alzheimer’s disease and related conditions. A Ciba
Foundation symposium. London: Churchill, 1970:37-50
15. Vogels OJM, Broere CAJ, ter Laak HJ, et al. Cell loss and
shrinkage in the nucleus basalis of Meynert complex in Alzheimer’s disease. Neurobiol Aging 1990;11:3-13
16. Terry RD, Peck A, DeTeresa R, et al. Some morphometric
aspects of the brain in senile dementia of the Alzheimer’s type.
Ann Neurol 1981;10:184-192
17. Buell S, Coleman P. Quantitative evidence for selective dendritic growth in normal human aging but not in senile dementia.
Brain Res 1981;214:23-41
18. Sparks DL, Hunsaker JC, Scheff SW, et al. Cortical senile
plaques in coronary artery disease, aging and Alzheimer’s disease. Neurobiol Aging 1990;11:601-607
19. McKhann G, Drachman D, Folstein M, Katzman D. Clinical
diagnosis of Alzheimer’s disease. Neurology 1984;34:939944
20. Khachaturian ZD. Diagnosis of Alzheimer’s disease. Arch Neurol 1985;42:1097-1105
21. Crosby EC, Humphrey T. Studies on the vertebrate telencephaIon: 11. The nuclear pattern of the anterior olfactory nucleus,
tuberculum olfactorium and amygdaloid complex in man. J
Comp Neurol 1941;74:309-352
22. Henderson G , Tomlinson BE, Gibson PH. Cell counts in human cerebral cortex in normal adults throughout life using an
image analyzing computer. J Neurol Sci 1980:,46:113-136
23. Mountjoy CQ, Roth M, Evans NJR, Evans HM. Cortical neuronal counts in normal elderly controls and demented patients.
Neurobiol Aging 1983;4:1-11
24. Henderson G, Tomlinson BE, Weightman D. Cell counts in
562 Annals of Neurology Vol 32 No 4 October 1992
human cerebral cortex using a traditional and an automatic
method. J Neurol Sci 1975;25:129-144
Corsellis JAN. Cell counting in the human brain: traditional and
electronic methods. Postgrad Med J 1975;51:722-726
Glees P. Neuroglia. Morphology and function. Oxford: Blackwell Scientific, 1955
Terry RD, DeTeresa R. The importance of video editing in
automated image analysis in studies of the cerebral cortex.’J
Neurol Sci 1982;53:413-421
Konigsmark BW. Contemporary methods in neuroanatomy. In:
Nauta WJH, ed. Methods for the counting of neurons. New
York: Springer, 1970:315-340
Winer BJ. Statistical principles in experimental design. New
York: McGraw-Hill, 1971
Coleman P, Flood D. Neuron numbers and dendritic extent
in normal aging and Alzheimer’s disease. Neurobiol Aging
Hubbard BM, Anderson JM. Age-related variations in the neuron content of the cerebral cortex in senile dementia of Alzheimer type. J Neuropathol Appl Neurobiol 1985;11:369-382
de la Monte SM. Quantitation of cerebral atrophy in preclinical
and end-stage Alzheimer’s disease. Ann Neurol 1989;25:45&
Tsuchiya K, Kosaka K. Neuropathological study of the amygdala
in presenile Alzheimer’s disease. J Neurol Sci 1990;lOO:165173
Terry RD, DeTeresa R, Hansen LA. Neocortical cell counts in
normal human adult aging. Ann Neurol 1987;21:530-539
Nakamura S, Vincent SR. Somatostatin- and neuropeptide Yimmunoreactive neurons in the neocortex in senile dementia of
Alzheimer’s type. Brain Res 1986;370:11-20
Burke WJ, Chung HD, Huang JS, et al. Evidence for retrograde
degeneration of epinephrine neurons in Alzheimer’s disease
Ann Neurol 1988;24:532-536
Mann DMA, Neary D, Yates PO, et al. Alterations in protein
synthetic capability of nerve cells in Alzheimer’s disease. J Neurol Neurosurg Psychiatry 1981;44:96-102
Hof PR, Cox K, Morrison JH. Quantitative analysis of a vulnerable subset of pyramidal neurons in Alzheimer’s disease: I. Superior frontal and inferior temporal cortex. J Comp Neurol
1990;30 1:44-54
Schecter R, Yen S-H, Terry RD. Fibrous astrocytes in senile
dementia of the Alzheimer type. J Neuropathol Exp Neurol
1981;40:95- 101
Tomlinson B, Blessed G, Roth M. Observations on the brains
of demented old people. J Neurol Sci 1970;11:205-242
Olejniczak P. Increased GFAP reactivity in septum and hippocampus from patients with Alzheimer’s disease resembles regionally specific GFAP response in rats subjected to selective
denervation. Neurology 1988;38(suppl 1):287
Vijayan VK, Geddes JW, Anderson KJ, et al. Astrocyte hypertrophy in the Alzheimer’s disease hippocampal formation. Exp
Neurol 1991;112:72-78
Hooper MW, Vogel FS. The limbic system in Alzheimer’s disease. Am J Pathol 1976;85:1-19
Beach TG, McGeer EG. Lamina-specific arrangement of astrcrcytic gliosis and senile plaques in Alzheimer’s disease visual COItex. Brain Res 1988;463:357-361
Mandybur TI, Chuirazzi CC. Astrocytes and the plaques of Alzheimer’s disease. Neurology 1990;40:635-639
Russchen FT. Cortical and subcortical afferenrs of the amygdala
complex. In: Schwarcz R, Ben-Ari Y, eds. Excitatory amino
acids and epilepsy. New York: Plenum, 1986:35-52
Amaral DG, Bassett JL. Cholinergic innervation of the monkey
amygdala: an immunohistochemical analysis with antisera to choline acetyltransferase. J Comp Neurol 1989;281:337-361
Pearson RCA, Powell TPS. The neuroanatomy of Alzheimer’s
disease. Rev Neurosci 1989;2:101-122
49. Davies P. Neurotransmitter-related enzymes in senile dementia
of the Alzheimer type. Brain Res 1979;171:319-327
50. Rossor MN, Garrett NJ, Johnson AL, et al. A postmortem study
of the cholinergic and GABA systems in senile dementia. Brain
51. Ferrier IN, Cross AJ, Hohnson JA, et al. Neuropeptides in
Akzheimer-type dementia. J Neurol Sci 1983;62:159-170
52. Hedreen JC, Broadhead JC, Price DL. Senile plaques in the
amygdala in Alzheimer’s disease. Neurology 1988;38(suppl 1):227
53. Unger JW, Lapham LW, McNeill T H , et al. The amygdala in
Alzheimer’s disease: neuropathology and Alz 50 immunoreactivity. Neurobiol Aging 1991;12:389-399
54. Amaral DG. Amygdalohippocampal and amygdalocortical projections in the primate brain. In: Schwarcz R, Ben-Ari Y, eds.
Excitatory amino acids and epilepsy. New York: Plenum,
55. Davies CA, Mann DMA, Sumpter PW, Yates PO. A quantitative morphometric analysis of the neuronal and synaptic content
of the frontal and temporal cortex in patients with Alzheimer’s
disease. J Neurol Sci 1987;78:151-164
Scheff SW, DeKosky ST, Price DA. Quantitative assessment
of cortical synaptic density in Alzheimer’s disease. Neurobiol
Aging 1990;11:29-37
DeKosky ST, Scheff SW. Synapse loss in frontal cortex biopsies
in Alzheimer’s disease: correlation with cognitive severity. Ann
Neurol 1990;27:457-464
Terry RD, Masliah E, Salmon DP, et al. Physical basis of cognitive alterations in Alzheimer’s disease: synapse loss is the major
correlate of cognitive impairment. Ann Neurol 1991;30:572580
Hyman BT, Van Hoesen GW, Damasio AR, Barnes CL. Alzheimer‘s disease: cell-specific pathology isolates the hippocampal
formation. Science 1984;225:1168-1170
Hyman B, Van Hoesen G , Kromer L, Damasio A. Perforant
pathway changes and memory impairment of Alzheimer’s disease. Ann Neurol 1986;20:472-481
Scott et al: Amygdala Cell Loss in AD
Без категории
Размер файла
883 Кб
loss, atrophy, amygdalar, disease, alzheimers, cells
Пожаловаться на содержимое документа