| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Departments of Psychiatry (M.J.deL., A.C., S.D.S., C.T., K.D.), Medicine (T.M.), Radiology (H.R.), Dermatology (N.O.), and Neurology (J.G.), New York University Medical Center, New York, New York; Departments of Chemistry and Medicine, Brookhaven National Laboratory (N.V.), Upton, New York; Orentreich Foundation for the Advancement of Science (N.O.), Coldspring, New York; Laboratory of Neuroendocrinology, Rockefeller University (B.M.), New York, New York; and Nathan Kline Institute for Psychiatric Research (M.J.deL., A.C.), Orangeburg, New York
Address all correspondence and requests for reprints to: Dr. Mony J. de Leon, New York University School of Medicine, Department of Psychiatry Neuroimaging Laboratory, 550 First Avenue, New York, New York 10016.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Considerable evidence exists to demonstrate that the hippocampus plays a major role in regulating the secretion of glucocorticoids via negative feedback to the hypothalamus and the pituitary (11, 12). Damage to the hippocampus can result in the overactivation of the hypothalamic-pituitary-adrenal (HPA) axis, which, in turn, has been associated with cognitive impairments and hippocampal damage (11, 13, 14, 15). Recent laboratory studies in the rat have shown that the degenerative hippocampal changes caused by overactivation of the HPA axis induced by chronic stress can be prevented or reduced by minimizing the brain’s exposure to glucocorticoids (i.e. adrenalectomy) (16) or with the administration of glucose (17). Overactivation of the HPA axis has been suggested to play a role in the hippocampal damage and memory deficits that have been observed in aging and Alzheimer’s disease (AD) (15, 18, 19, 20, 21, 22, 23, 24).
The hippocampal formation is affected early in the course of AD (25). Other human studies have shown that hippocampal integrity is necessary for normal memory functioning (26). Although exogenously administered glucocorticoids have been shown to adversely affect human memory consolidation and recall (27, 28), and glucose has been shown to improve memory performance in normal elderly (29) and elderly with probable AD (30, 31), little is known about the mechanisms mediating these effects or whether the hippocampus is a target site of action. In the present study of normal elderly subjects and patients with mild to moderate AD, we hypothesized that the response of the brain to exogenous administration of hydrocortisone (cortisol) would be to reduce hippocampal glucose utilization. We used positron emission tomography (PET) and 2-deoxy-2-[18F]fluoro-D-glucose (18FDG) to measure regional cerebral glucose utilization (CMRglu). The results, in support of our hypothesis, showed that hydrocortisone caused a selective hippocampal CMRglu reduction and raised serum glucose levels only in the normal group.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Seven normal elderly and eight mild to moderate AD patients,
matched for age and education, were selected for the study after
extensive diagnostic and screening evaluations at the New York
University School of Medicine, Aging and Dementia Research Center (see
Table 1
). All AD patients met both
NINCDS-ADRDA (32) and DSM III-R (33) diagnostic criteria for probable
AD, and the controls were all considered to be within normal
limits.
|
The screening evaluation was used to identify study patients who were free of potentially confounding medical conditions and/or therapies that could affect brain structure or function. The procedures included medical, neurological, psychiatric, neuropsychological, and neuroradiological examinations. Individuals with identifiable conditions associated with cognitive impairments were excluded: e.g. modified Hachinski Ischemia Score ratings (34) greater than 3, vitamin deficiency states, diabetics dependent on insulin or oral agents, thyroid disorders, history of significant head trauma, or substance abuse. Individuals with evidence of structural brain alterations that included mass lesions, stroke or lacunar infarcts, or evidence of hydrocephalus were also excluded. In addition, subjects were excluded if they had a prior psychiatric history, received a Hamilton Depression Scale Score (35) of 16 or greater, or had not discontinued any psychotropic or cognitively active medication at least 2 weeks before the evaluation period and the PET scans. For both patients and controls, all diagnostic and screening examinations were completed within a 3-month period.
Psychiatric and psychometric evaluations
Staging for the general level of functional ability was conducted using the seven-point Global Deterioration Scale (GDS) (36). With the GDS, elderly subjects receiving scores of 1 or 2 (differentiated only on the basis of subjective memory complaints) are considered to be normal. Individuals with scores of 3 have objective evidence for mild memory and cognitive dysfunctions. Patients with GDS scores of 4 or greater demonstrate functional deficits fulfilling criteria for the diagnosis of AD. In this project, all of the elderly controls had GDS scores of 1 or 2, and all AD patients had scores of 4 to 6. All GDS ratings were administered by clinicians blind to the neuropsychological and neuroimaging test results. In addition, the Mini Mental Status Examination (37) was included to further characterize the overall level of functioning and to provide reference to other published studies. We also used a psychometric test battery derived from the Guild Memory Test (38, 39). From this battery, we were able to derive a measure of both immediate and delayed recall by combining results on the paragraph recall and the verbal paired associates tests.
Magnetic resonance imaging (MRI) studies
MRI scans were performed using a 1.5-tesla Phillips Gyroscan imager (Phillips International, Einhoven, Netherlands). The MRI examination was designed to cover the whole brain, identify structural abnormalities, and permit an overall characterization of the extent of brain atrophy. To determine the planes of section for the diagnostic and research studies, we obtained 12 T1-weighted scout sagittal images (repetition time (TR) = 630 ms; echotime (TE) = 20 ms) that were 6 mm thick with 20% gaps. From visual inspection of the sagittal scout images, an axial plane was defined as parallel to the long axis of the hippocampus proper. The plane perpendicular to the axial plane defined the coronal plane. For the diagnostic examinations, 18 T1-weighted, axial spin echo images were obtained with a slice thickness of 4 mm and a 10% gap (TR = 630 ms; TE = 20 ms). Also obtained were 18 proton density-weighted and 18 T2-weighted axial slices with a 6-mm slice thickness and a 10% gap (TR = 2516 ms; TE = 29 and 80 ms). For research purposes, which included the anatomical sampling of the PET image data after MRI-PET registration, we obtained, through the anterior-posterior length of the hippocampus, a coronal set of 18 T1-weighted images with 4-mm slice thickness and 10% gaps (TR = 630 ms; TE = 20 ms; field of view = 230; number of excitations = 1; matrix, 256 x 256). The total MRI examination took approximately 45 min.
PET studies
PET scans measuring CMRglu were obtained using 18FDG
(half-life of 110 min) with a Siemens CTI scanner (model 931–08/12).
The scanner is an eight-ring whole body system, routinely generating 15
tomographic slices (eight direct and seven cross-slices). The axial
resolution is 6.2 mm FWHM (direct slices) and 6.7 mm for cross-slices.
The interslice distance is 6.75 mm, and the total coverage is 108 mm.
To improve counting statistics in relatively small brain subvolumes,
two interleaved scan sets (two bed positions) offset by one half-slice
thickness (
3.4 mm) were acquired. Before the PET-18FDG
study, transmission images were obtained for subsequent attenuation
correction. The 30 images of the three-dimensional (3-D) PET image data
set were displayed on a 128 x 128 matrix.
Patients were studied on 2 separate days. Examinations took place at approximately 1200 h. Subjects were requested to refrain from eating a large breakfast but, instead, to eat at 0600 h a small portion of toast and juice and, if customary, a small caffeinated beverage. To minimize motion and to ensure repositioning accuracy, a thermoplastic head holder was made for each subject. The order of the studies was not counterbalanced; in 13 of 15 cases the placebo study was the first study condition. One subject in each of the 2 groups had the hydrocortisone challenge as the first PET study. For both study conditions, approximately 1 h before the PET-18FDG study, a 21-gauge catheter was placed in the left radial artery, and a 21-gauge venous line was positioned in the contralateral antecubital region. The subjects received either an iv bolus of the saline-placebo or 35 mg hydrocortisone sodium succinate. Within approximately 2 min, a bolus of approximately 6 mCi 18FDG was injected through the same venous line. For the next 55 min, arterial blood samples were periodically obtained on a fixed schedule. These samples were used to determine serum 18F levels for quantitative modeling of radiotracer uptake (40) and to determine the time course for the serum glucose and cortisol levels. During the tracer uptake, subjects rested supine with eyes open and ears unplugged in a quiet and dimly lit scan room. Scanning was started 35 min after the 18FDG injection, and the PET image acquisition time was 20 min. The PET scans were acquired in an axial plane of section, 25° negative to the cantho-meatal line, a plane designed to parallel the plane of the hippocampus and to approximate the MRI plane of section.
Image analyses
Two image analysis procedures were used for each patient to examine the effects of hydrocortisone on CMRglu. The first procedure was designed to provide a rapid exploratory survey of the entire brain for sites of cortisol activation. For all subjects, the 3-D PET image data set was transformed to a common coordinate system. This has the effect of aligning the anatomy across cases such that there is an approximate correspondence on a voxel by voxel basis. The transformation included 1) trilinear interpolation to cubic voxels, 2) reangulation to correct for the variable alignment of individual scans, and 3) linear scaling of each brain, using the maximum x-, y-, and z-axes measurements, to create correspondence in size to a specific control scan. After the transformations, the image data were resliced into a plane parallel to the long axis of the hippocampus with a slice thickness of 3.4 mm. No normalization procedures were used to adjust the absolute glucose metabolism.
For each of the four study factors (two groups x two conditions) we computed, on a voxel by voxel basis, the average metabolic rates and the SD’s. Subsequently, on a voxel by voxel basis, t tests were performed to test the null hypothesis that within each diagnostic group the placebo and the hydrocortisone conditions had equal effects on brain glucose utilization. The voxels that reached significance on these t tests were highlighted and overlaid on the transformed PET images to create so-called t-map images (see below).
The second procedure used operator-drawn anatomical regions of interest
to examine statistically the within- and between-group cortisol effects
on CMRglu. Here, for each patient, the PET images from each study
condition were interpolated from the 128 x 128 matrix from which
they were acquired into a 256 x 256 matrix with a pixel size of
0.9 mm so as to match the corresponding MRI data. Using an automated
surface-fitting algorithm, separately for each condition, the
interleaved sets of axial PET data were coregistered to the coronal MRI
(41, 42). Once coregistered, coronal PET images were created by
reslicing the PET data to conform with the 4-mm slice thickness and the
imaging plane of the MRI. Our prior work has shown that when two
interleaved PET data sets are acquired, reformatting the PET data from
the axial to the coronal plane reduces isotope recovery errors to only
4% even when sampling small structures with volumes comparable to that
of the hippocampus in AD (43). Using the coregistered MRI scans as the
anatomical guide, regions were drawn on the MRI scan and automatically
transferred to both the placebo and hydrocortisone PET scans (see Fig. 1). To improve accuracy, regions were
drawn on x3 zoomed images using our locally developed region of
interest software (MIDAS, New York University, New York, NY). Specific
regions were selected on the basis of the hypothesized site of action
of glucocorticoids, (i.e. hippocampus), sites known to be
affected by AD (i.e. lateral temporal and parietal lobes)
(44), regions known to be affected on the basis of normal aging, but,
less sensitive to mild AD changes (i.e. frontal lobe) (45, 46), and a site largely unaffected by age or AD (i.e.
midbrain) (47). The contribution of cerebrospinal fluid to the
estimated regional metabolic rates was minimized by using precise
anatomical drawings followed by a MRI tissue segmentation technique to
exclude cerebrospinal fluid from the region of the drawing. The CMRglu
for a structure was determined by computing over all slices the
weighted mean of all PET pixels in the outlined regions. These data
were then analyzed using software by SPSS, Inc. (Chicago, IL).
|
During the 1-h duration of each PET study, 1-mL blood samples were periodically drawn and centrifuged to extract the serum. One half milliliter of serum from each sampled time point was used in the quantification of the PET study. The serum remaining from each of the samples was subsequently combined into 4 vials. Vial 1 was comprised of 14 samples collected between 0–4 min of the study, vial 2 was made up of 4 samples between 6–15 min, vial 3 contained the 3 samples between 20–30 min, and vial 4 was comprised of the 2 samples between 45–55 min. All combined serum samples were kept on ice for approximately 2 h during transport to the laboratory, then frozen at -70 C until assayed. Glucose was measured using a fluorometric assay, and cortisol was measured using a RIA (Diagnostic Products Corp., Los Angeles, CA). We observed considerable individual variability in baseline serum glucose levels during both the placebo and hydrocortisone conditions. Consequently, glucose levels were normalized for each study condition separately, by dividing the values for each vial by the corresponding baseline (vial 1) measurement. Vial 1 was selected as the reference because not all subjects had blood drawn before the administration of hydrocortisone. Examination of the glucose time curves from vials 1–4 showed that over the first 2 time points, the glucose values were stable in both groups across both conditions. As the serum cortisol data immediately reflected the hydrocortisone injection, these data were not normalized.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Figure 2 shows both an axial MRI
slice through the long axis of the hippocampus and the t-maps overlaid
on the similarly oriented averaged PET images from the normal and AD
groups. For illustrative purposes, the t-maps for a brain slice through
the hippocampus are shown at two probability levels (P
< 0.05 and P < 0.001). At P < 0.001,
which controls for the number of voxel comparisons (48), we observed a
CMRglu reduction restricted to the left hippocampal region of the
normal group. In AD patients, no region showed significant changes at
the P < 0.001 level. Over the entire brain, in both
the normal and AD groups, at the less conservative P <
0.05 level, the largest areas of CMRglu reductions were observed in the
hippocampus. No brain region in either group, at either probability
level, showed an increased CMRglu related to the hydrocortisone
challenge.
|
). We used a two-way repeated measures
ANOVA to examine whether exposure of the brain to hydrocortisone would
have a differential regional effect on the CMRglu of AD and normal
groups. In these analyses, CMRglu was the dependent variable,
diagnostic group was the independent variable, and one repeated
measures factor was used (placebo-hydrocortisone conditions). With this
statistical design, a significant diagnostic group by repeated measures
factor interaction for the CMRglu of any brain region would support the
main hypothesis. The results show that only the hippocampal regions
bilaterally demonstrated significant group by condition interactions
[F(1, 13) = 24.32; P < 0.001].
|
Although not the focus of this study, as anticipated, we found baseline
(placebo) regional CMRglu differences between normal elderly and
patients with AD (see Table 2
). The
placebo condition, which is comparable to the resting conditions
typically used in PET studies, showed significant AD-related CMRglu
reductions in the hippocampi (30%), lateral temporal lobes (25%), and
inferior parietal lobule (26%; P < 0.05 for all). No
differences were found for the frontal or midbrain regions. Examining
the between-group t tests under the hydrocortisone condition
showed that both the inferior parietal and the lateral temporal lobe
CMRglu in the AD group continued to show significant reductions from
the normal group (P < 0.05). Under the hydrocortisone
condition, the hippocampal region CMRglu in the normal group was no
longer significantly different from that in the AD group.
Serum cortisol and glucose measures
In a two-way repeated measures ANOVA design, there were no group,
condition, or group by condition interactions (P >
0.05) for any of the cortisol measures (area under the curve, mean of
the four samples, minimum to maximum, or baseline to last sample).
Statistical examination for each condition separately showed that the
area under the cortisol curve during the placebo condition was
significantly higher in the AD group [
normal =
291.6 ± 101.6; AD = 549.9 ± 295.0;
t(13) = -2.20; P < 0.05]. The arithmetic
mean of the four time points was consistently higher in the AD group,
but did not reach significance (normal = 5.7 ±
2.0 µg/dL; AD = 10.9 ± 6.3 µg/dL;
t(13) = -2.06; P = 0.06). With
hydrocortisone administration, the AD patients and the controls showed
statistically equivalent cortisol clearance (area under the curve;
normal = 3338.7 ± 616.2; AD
= 3153.1 ± 575.5; t(13) = 0.61; P
> 0.05). By the first sample, the levels reached a peak of 87.6
± 15.1 µg/dL in the normal group and 79.8 ± 12.9 µg/dL in
the AD patients. By 50 min, the cortisol value was 49.9 ± 9.9
µg/dL in the normal group and 46.5 ± 17.1 µg/dL in the AD
group.
The area under the curve for the normalized serum glucose values,
however, did demonstrate, by two-way repeated measures ANOVA, a
significant condition effect [F(1, 13) = 5.2; P <
0.05] and a trend toward a significant group by condition interaction
[F(1, 13) = 3.5; P < 0.10]. Follow-up paired
t tests comparing study conditions within each of the two
groups indicated that for the third and fourth sample points studied
(approximately 25 and 50 min after the hydrocortisone injection), the
control group showed a significant increase in the normalized serum
glucose values in the hydrocortisone condition relative to the placebo
condition [sample 3: t(6) = -2.96; P <
0.05; sample 4: t(6) = -3.07; P < 0.05].
No difference between study conditions was found for the AD group (see
Fig. 3).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Across species, numerous responses to acute stress (glucocorticoid challenge) have been detected peripherally and are referred to as the fight or flight response. In part, these responses include hormonal secretions (e.g. CRH, ACTH, cortisol, epinephrine, glucagon, and norepinephrine), conserving glucose by decreasing sensitivity to insulin, decreasing the rate of enzymatic phosphorylation of glucose, increasing the availability of stored sugars, and releasing the neurotransmitters serotonin and noradrenaline to increase arousal and blood pressure. Munck reported that the inhibition of glucose uptake in peripheral tissues was among the consequences of elevated glucocorticoid levels (2). He observed that the mechanism affecting glucose metabolism was "a block between glucose and glucose-6-phosphate" that is, inhibition of the transport of glucose rather than glucose phosphorylation (2). Studies of the effects of corticosterone in rat brain have consistently shown evidence for a similar mechanism in the hippocampus (5, 6, 7, 8). Our study is the first to extend those observations to the human, demonstrating that hydrocortisone selectively reduces hippocampal glucose metabolism in healthy elderly subjects. Moreover, our observation that elevations in serum glucose levels are provoked by hydrocortisone administration are also consistent with prior rat and human studies (3, 8, 9).
Although the genomic effects of glucocorticoids on peripheral
carbohydrate metabolism have been subject to extensive investigations
(4, 49), there are few studies showing more rapid membrane effects,
particularly those affecting the brain. Munck’s early studies of
peripheral tissues showed that the inhibitory effects of cortisol on
glucose transport varied by the type of tissue studied. Thymus,
adipose, and skin tissue showed faster responses than muscle tissue. In
thymus tissue, which showed the fastest inhibition of glucose
transport, inhibition could be detected in less than 20 min (2).
Similarly, Shamoon et al. showed, using
[3-3H]glucose, that after raising cortisol levels from 11
to 38 µg/dL in young healthy humans, whole body glucose uptake was
reduced relatively early, whereas plasma glucose levels demonstrated a
more gradual and delayed increase (3). Glucose uptake decreased by
about 5% in the 30-min period after cortisol infusion, by 11% after
60 min, and by 15% after 5 h. Plasma glucose levels showed small
elevations from baseline by 30 min (2 mg/dL) and 60 min (4
mg/dL), a magnitude of change similar to what we observed in our normal
control group. By 5 h, glucose levels were elevated by
approximately 15 mg/dL over baseline. Acute effects have also been
shown in brain. In an in vivo study of the rat brain,
15 s after corticosterone administration, a reduction in
[14C]glucose uptake was demonstrated in several regions,
including the hippocampus (7). These researchers, by demonstrating in
another group of rats that corticosterone had no effect on cerebral
blood flow, concluded that the decreased brain glucose uptake was due
to a corticosterone-induced reduction in glucose permeability at the
blood-brain barrier. A study of single unit recording from the in
vivo rat brain showed consistent corticosterone decreases in
hippocampal neuronal activity from 10–40 min after ip corticosterone
injection; the effect lasted a minimum of 2 h (50). In an in
vitro study, after 10–40 min of incubation in corticosterone, the
excitability of hippocampal pyramidal neurons decreased (51).
In our study, we could not resolve the time course of brain glucose metabolism after hydrocortisone administration in greater detail than a single averaged 35-min epoch. With 18FDG as the precursor, during a 35-min uptake period to allow for transport of 18FDG into the brain and intracellular phosphorylation through hexokinase, most of the label is trapped as 18FDG-6-phosphate. There are negligible losses of the phosphorylated product for 60 min, and therefore, after the uptake period there is a sufficient period of time to image the brain with a PET camera (40). Consequently, our data suggest that in the human brain, there is also an early and preferential decrease in glucose uptake in the hippocampus. However, because our PET data do not permit independent measures of transport and glucose phosphorylation, we cannot conclude that the transport is differentially affected.
We interpret our data as preliminary, because we cannot rule out that some uncontrolled factors might have adversely affected the validity of our result. These other factors, although unlikely, include serum glucose levels not at a steady state, major alterations in the kinetic parameters that are used to determine the glucose utilization, and changes in the tissue for the relative affinity to glucose and to 18FDG (lumped constant) (40). With respect to the first point, serum glucose levels were essentially unchanged during the first 10 min, and even by the end of the PET-18FDG uptake period only small elevations were noted in the control group, a serum glucose result comparable to that reported by Shamoon et al. (3). Moreover, the predicted consequence of increasing serum glucose levels would be to decrease the CMRglu for the entire brain, and we did not find any evidence of this. The other points cannot be directly addressed by our study, but there, too, we would expect to see CMRglu changes over a widespread brain topography. The possibility also exists that the failure of the AD brain to respond is due to a floor effect in the CMRglu. In the present study, our AD subjects, under placebo conditions, had reductions of 30% in the hippocampal CMRglu relative to normal subjects. However, among other more severely impaired patients, we observed CMRglu reductions in excess of 60% relative to normal values. Therefore, the simple explanation of a floor effect is not adequate. Moreover, in the present study we did not investigate the extent to which failure to respond was reflective of a smaller structure size. However, this also would be too simple a solution, as we have determined in prior studies (52) that the hippocampus in mild to moderate severity AD patients was sufficiently large in size (reduced 22% in volume from control) to accurately study with PET when the images were coregistered to MRI scans.
Our study did not address the issue of the dose-response curve. We used
a single pharmacological dose to challenge the subjects, one that
elevated serum cortisol levels beyond physiological levels (80
µg/dL). Although it has been reported in the rat that corticosterone
inhibits brain glucose uptake in a dose-dependent fashion (7), it has
also been shown that the relationship between glucocorticoid dose and
measures of hippocampal activity-primed burst potentiation may be an
inverted U function (53). That is, intermediate doses of
glucocorticoids were reported to have a stimulatory effect, whereas
larger doses suppress hippocampal activity. Consequently, our data are
consistent with results in the rat indicating that a relatively large
dose of cortisol would have an inhibitory effect on hippocampal MRglu.
It is intriguing that our results demonstrated this effect only in the
normal group, and the brain effect was specific to the hippocampus.
The mechanism explaining the failure of AD patients to mount a hydrocortisone-related hippocampal CMRglu response is unknown. We speculatively offer that in AD, elevated basal cortisol levels alter functions mediated by glucocorticoids, such as the glucose transport mediated by glucose transporters. Considerable data exist to support the conclusion that with increasing age, there is an increase in circulating glucocorticoid levels, resulting in a preferential insult to hippocampal neurons (11, 15). This may in part be due to age-related alterations in the function of glucocorticoid receptors on hippocampal neurons, resulting in inadequate feedback. Studies by Landfield et al. in the rat have demonstrated that with aging, hippocampal neurons show reductions in their capacity to up-regulate their glucocorticoid receptors after adrenalectomy (54) and to down-regulate receptors under conditions of stress (elevated glucocorticoid levels) (55). Moreover, an increase in the affinity between the hippocampal glucocorticoid receptor and its ligand has also been reported (56). This may increase the vulnerability of neurons to the damaging effects of glucocorticoids. Recent human studies suggest that although there is relatively preserved basal control of cortisol levels with increasing age, there is also overresponsivity of the HPA axis to stressors and reduced HPA axis sensitivity to glucocorticoid feedback (24). This pattern may be exaggerated in AD, as a recent study demonstrated that compared with age-matched controls, glucocorticoid receptor messenger ribonucleic acid is not reduced in AD brain (57), and this study and others observed elevated basal cortisol levels in AD patients relative to those in age-matched controls (22, 58, 59). Moreover, other AD studies have reported excess cortisol responses to diverse challenges: dexamethasone suppression (60), insulin (18), and glucose (21).
In AD there are reduced numbers of hippocampal glucose transporter sites (61, 62). This is consistent with recent PET data showing reductions in the kinetics for brain glucose transport (63, 64). Over all, these findings suggest two nonindependent explanations for our observation of a lack of response to hydrocortisone in AD. First, there is reduced metabolic demand from deafferentiated or dying hippocampal cells. As reviewed above, ample evidence exists documenting severe pathology in the hippocampus and its projection areas. It appears reasonable to assume, but it remains untested, that such damage can diminish the magnitude of the response of the hippocampus to hydrocortisone or other pharmacological or behavioral challenges. Second, under conditions of elevated basal cortisol levels, glucocorticoid-regulated glucose transport may have been already reduced in the AD group, thus impairing the inhibitory effects of hydrocortisone. As recent studies have identified both insulin-dependent (65, 66, 67) and noninsulin-dependent actions of glucocorticoids on glucose transporters (68), it is possible that the higher basal (endogenous) cortisol levels we observed in the AD group desensitized the remaining glucose transporters and inhibited further glucocorticoid-mediated reductions of glucose transport. Consequently, it is conceivable that in AD, where there are reduced numbers of brain glucose transporters, preserved glucocorticoid receptor messenger ribonucleic acid, and possibly even higher receptor affinity, there is a functional impairment of glucocorticoid receptor regulation and glucocorticoid-mediated responses. It also is unknown whether a similar mechanism may apply peripherally and thus account for the observed failure of AD patients to mount a serum glucose elevation in response to the hydrocortisone challenge. Further complicating this issue is the possibility that AD patients, as a consequence of impaired reasoning, suffer both more frequent and more severe stress responses to perceived danger. As a result, we do not know to what extent the findings we observed are a consequence of either the experimental conditions or a trait-like feature of AD.
In vivo MR studies of AD consistently show hippocampal atrophy (52, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81), and the current PET study adds to our understandings of AD by showing that in addition to the well known progressive neocortical CMRglu reductions (44, 82), there are hippocampal CMRglu reductions that can be detected in the unchallenged (baseline) state. It remains to be determined whether the hippocampal CMRglu reductions are of greater diagnostic value than the volume reductions determined by MRI. However, because it has been suggested that the earliest stages of AD might most clearly reveal altered HPA axis functioning (83, 84), and because prior AD studies show that there are very large losses of hippocampal neurons (85, 86, 87) with marked hippocampal formation neuropathology beginning early in the course of the disease (25, 88), the potential for hydrocortisone challenge as an early diagnostic test or therapeutic outcome measure for AD should be explored.
An extensive literature indicates that excess exposure to glucocorticoids is detrimental to memory performance, presumably for their effects on the hippocampus (11, 19, 89). Furthermore, recent clinical data show that experimental elevation of cortisol is associated with electroencephalogram slowing in young normal humans (90) and with hippocampal damage and memory changes in several clinical populations (91, 92, 93, 94). Therefore, it would be of interest to determine whether the hydrocortisone challenge would have a differential impact on the brain metabolism and neuropsychological test performance in groups of patients with hippocampal damage or in the comparison of young and old normal subjects. We did not observe any relationships between measures of immediate or delayed recall performance and the hydrocortisone-induced change in the hippocampal CMRglu. However, in the present study, because the memory measures were obtained before the PET studies, the neuropsychological data do not directly address the hypothesized relationship between changes in hippocampal glucose utilization and memory performance. Future brain-imaging studies will be able to examine neuropsychological performance during the time of the scan.
In conclusion, the data suggest that a pharmacological dose of hydrocortisone will reduce the hippocampal CMRglu in normal subjects as well as raise their serum glucose levels. In AD patients, there is a failure of these responses. The mechanism behind this lack of response in AD remains unknown. Based on other work, showing damage to the hippocampus and its neuronal connections along with reduced numbers of hippocampal glucose transporters, we speculate that functional dysregulations of glucocorticoid receptors and altered glucocorticoid-mediated glucose transporter function may play a role. We emphasize that our results, which are based on small samples, a single dose of hydrocortisone, and limited time-course sampling, should be considered preliminary.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received January 27, 1997.
Revised May 14, 1997.
Accepted June 26, 1997.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. L. Wright and C. Persad Distinguishing Between Depression and Dementia in Older Persons: Neuropsychological and Neuropathological Correlates J Geriatr Psychiatry Neurol, December 1, 2007; 20(4): 189 - 198. [Abstract] [PDF]
|
|
|
|
 |
|
 R. GROSSMAN, R. YEHUDA, J. GOLIER, B. MCEWEN, P. HARVEY, and N. STA. MARIA Cognitive Effects of Intravenous Hydrocortisone in Subjects with PTSD and Healthy Control Subjects Ann. N.Y. Acad. Sci., July 1, 2006; 1071(1): 410 - 421. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Mosconi, W. -H. Tsui, S. De Santi, J. Li, H. Rusinek, A. Convit, Y. Li, M. Boppana, and M. J. de Leon Reduced hippocampal metabolism in MCI and AD: Automated FDG-PET image analysis Neurology, June 14, 2005; 64(11): 1860 - 1867. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
