Directory UMM :Data Elmu:jurnal:B:Biological Psichatry:Vol48.Issue4.2000:

Hippocampal Volume in Geriatric Depression
David C. Steffens, Christopher E. Byrum, Douglas R. McQuoid,
Daniel L. Greenberg, Martha E. Payne, Timothy F. Blitchington,
James R. MacFall, and K. Ranga Rama Krishnan
Background: There is a growing literature on the importance of hippocampal volume in geriatric depression.
Methods: We examined hippocampal volume in a group
of elderly depressed patients and a group of elderly
control subjects (N 5 66 geriatric depressed patients and
18 elderly nondepressed control subjects) recruited
through Duke’s Mental Health Clinical Research Center
for the Study of Depression in the Elderly. The subjects
received a standardized evaluation, including a magnetic
resonance imaging scan of the brain. Patients had unipolar major depression and were free of comorbid major
psychiatric illness and neurologic illness. Differences
were assessed using t tests and linear regression
modeling.
Results: Accounting for the effects of age, gender, and
total brain volume, depressed patients tended to have
smaller right hippocampal volume (p 5 .014) and left
hippocampal volume (p 5 .073). Among depressed patients, age of onset was negatively but not significantly
related to right hippocampal volume (p 5 .052) and to left

hippocampal volume (p 5 .062). We noted that among
subjects with either right or left hippocampal volume of 3
mL or less, the vast majority were patients rather than
control subjects.
Conclusions: These results support a role for hippocampal dysfunction in depression, particularly in late-age
onset depression. Longitudinal studies examining both
depressive and cognitive outcomes are needed to clarify
the relationships between the hippocampus, depression,
and dementia. Biol Psychiatry 2000;48:301–309 © 2000
Society of Biological Psychiatry
Key Words: Depression, hippocampus, magnetic resonance imaging

From the Departments of Psychiatry and Behavioral Sciences (DCS, CEB, DRM,
DLG, MEP, TFB, KRRK) and Radiology (JRM), Duke University Medical
Center, Durham, North Carolina.
Address reprint requests to David C. Steffens, M.D., Assistant Professor of
Psychiatry, Duke University Medical Center, Box 3903, Durham NC 27710.
Received September 3, 1999; revised February 1, 2000; accepted February 7, 2000.

© 2000 Society of Biological Psychiatry

Introduction

T

he role of the hippocampus in mood disorders is
unclear. There are known neural connections between
the hippocampal formation and frontal and limbic areas
that are implicated in development of depressive symptoms, but how and to what extent the hippocampus may
modulate mood and the pathophysiology of the stress
response is not understood (McEwen 1997; Mongeau et al
1997). Neuroimaging, particularly magnetic resonance
imaging (MRI), has been used to study the hippocampus in
populations of patients with mood disorders (Axelson et al
1993; Coffey et al 1993; Krishnan et al 1991; O’Brien et
al 1997; Sheline et al 1996). These studies have yielded
conflicting results.
Focusing on hippocampal changes in geriatric depression allows investigators to test two major hypotheses
regarding age of first onset of depression and smaller
hippocampal volumes among elderly depressives. One

hypothesis follows the stress-induced glucocorticoid toxicity model (McEwen 1997; Sapolsky 1993), which links
the finding of hypothalamic-pituitary-adrenal axis dyscontrol leading to higher cortisol levels in depression (Carroll
et al 1981a) to subsequent glucocorticoid-induced hippocampal damage (Sheline 1996). In this model, greater
length of time of depressive symptoms would predict
greater hippocampal damage. Thus, older patients with
earlier first onset of depressive symptoms should have
smaller hippocampal volumes compared with those with
later first onset of depression.
The other hypothesis follows the observation (confirmed by our own clinical experience) that late-onset
depression often precedes onset of dementia, particularly
Alzheimer’s disease (Jorm et al 1991; Kokmen et al 1991;
Speck et al 1995; Steffens et al 1997), which is itself
associated with hippocampal degeneration (Jack et al
1998; Laasko et al 1998). Here, a depressive syndrome
may represent the heralding sign of incident dementia, and
thus we would expect that certain individuals with lateonset depression would have smaller hippocampal volumes. Complicating hypothesis linking geriatric with Alzheimer’s disease is another large body of literature linking
late-onset depression with vascular brain changes (Krish0006-3223/00/$20.00
PII S0006-3223(00)00829-5

302

D.C. Steffens et al

BIOL PSYCHIATRY
2000;48:301–309

nan et al 1997; Soares and Mann 1997); this evidence
would suggest that cognitive decline may be a vascular
phenomenon, possibly with limited hippocampal
involvement.
Both hypotheses would thus predict smaller hippocampal volumes in depressed patients versus control subjects
regardless of age of onset. If both hypotheses prove true,
then the effect of age of onset on hippocampal volume will
be determined by the relative effects of each condition,
and it is possible that one may not find any differences
when comparing early-onset with late-onset depressed
patients. Similarly, there will be no effect of age of onset
if neither model is operative. Yet it is also possible to
formulate hypotheses depending on the sample. If lateonset depression predominates, then a significant negative
association between age of onset and hippocampal volume

may exist. Such is the case with our sample. Thus, we
predicted the following: 1) older depressed patients will
have smaller hippocampal volumes than will nondepressed elderly control subjects and 2) hippocampal volume will be negatively correlated with age of onset among
the depressed group.
We sought to examine these issues in a heterogeneous
group of geriatric depressives and elderly nondepressed
controls.

the MRI study (see MRI method section below). Within 60 days
of the MRI, a trained interviewer administered the Duke Depression Evaluation Schedule (DDES) to each enrolled subject. The
DDES, a composite diagnostic interview instrument, includes
sections of the NIMH Diagnostic Interview Schedule (Robins et
al 1981) assessing depression (enriched with items assessing
sleep problems and the clinical features of melancholia and
psychosis), dysthymia, mania, and alcohol abuse or dependence.
The DDES also includes the Montgomery–Asberg depression
rating scale (Montgomery et al 1979), the Mini-Mental State
Examination (MMSE; Folstein et al 1975), items assessing
self-reported physical health, four subscales of the Duke Social
Support Index (George et al 1989; Landerman et al 1989), and a

section assessing frequency and severity of stressful life events
during the year preceding the interview (Landerman et al 1989).
Other clinical data, including the Carroll Rating Scale for
Depression (Carroll et al 1981b), Hamilton Rating Scale for
Depression (Hamilton 1960), Consortium to Establish a Registry
for Alzheimer’s Disease test battery (Morris et al 1988), Hachinski Ischemia Scale (Hachinski et al 1975), Global Assessment
Scale (Endicott et al 1976), and Cumulative Illness Rating Scale
(Linn et al 1968) also were assessed at baseline.
DSM-IV diagnoses were assigned to all subjects by a consensus diagnostic conference, which included a board-certified or
board-eligible psychiatrist, using procedures conforming to the
Longitudinal, Expert and Available Data standard (Spitzer 1983)
and informed by screening data, the DDES, and the clinical data
listed above. Concordance between the consensus diagnoses and
other diagnostic tests was moderate to high (George et al 1989).

Methods and Materials
Design and Sample

Magnetic Resonance Imaging

This study used a cross-sectional design. All subjects were
participants in the NIMH Mental Health Clinical Research
Center (MHCRC) for the Study of Depression in Later Life,
located at Duke University, who were screened with the Center
for Epidemiologic Studies-Depression Scale (CES-D; Radloff
1977). Control subjects were obtained from the Duke Aging
Center Registry of elderly volunteers (60 years and older) who
were free of psychiatric and neurologic illness. After complete
description of the MHCRC and its procedures were explained to
each subject, written informed consent was obtained.
Eligibility for patients was limited to those with CES-D scores
$16 or a diagnosis of major depression, single (incident cases) or
recurrent (prevalent cases), and was restricted to patients aged 58
years or older who could speak and write English. Exclusion
criteria included 1) another major psychiatric illness, such as
bipolar disorder, schizophrenia, and schizoaffective disorder; 2)
active alcohol or drug dependence; 3) primary neurologic illness,
such as dementia, stroke, Parkinson’s disease, seizure disorder,
or multiple sclerosis; 4) presence of a medical illness or medication use that could affect cognitive function; 5) physical
disability that precludes cognitive testing; and 6) metal in the

body that precludes MRI. The MHCRC is particularly careful to
assess subjects for dementia; all subjects have a clinical examination by a geriatric psychiatrist, and all subjects with known or
suspected dementia are excluded from the MHCRC.
At baseline, all subjects underwent a standardized section of

MRI ACQUISITION. All subjects were screened for the
presence of cardiac pacemakers, neurostimulators, metallic implants, metal in the orbit, aneurysm clips, or any other condition
where MRI is contraindicated. Subjects were imaged with a 1.5
T whole-body MRI system (GE Medical Systems, Waukesha,
WI) using the standard head (volumetric) radiofrequency coil.
The scanner alignment light was used to adjust the head tilt and
rotation to ensure the median was the canthomeatal line.
HIGH-RESOLUTION IMAGING FOR VOLUME MEASURE-

Two dual-echo fast-spin echo acquisitions were obtained: one in the axial plane for morphometry for most cortical
structures and a second (fast spin echo) in a coronal oblique
plane for segmentation of the amygdala-hippocampus complex.
The pulse sequence parameters were: TR 5 4000 msec, TE 5
30, 135 msec, 32 KHz imaging bandwidth, echo train length 5
16, with 3-mm section thickness and 1 excitation per phaseencoding increment, 20-cm FOV. Saturation of spins outside the

imaging volume (standard gap 15 mm) was employed to minimize artifacts due to flowing blood and CSF. The images were
acquired in two separate acquisitions with a 3-mm gap between
sections for each acquisition. The second acquisition was offset
by 3 mm from the first so that the resulting data set consisted of
contiguous sections. For the near coronal acquisition, the localizer scan was used to identify the anterior commissure-posterior
commissure (AC-PC) line. Oblique, near-coronal FSE images

MENT.

Hippocampal Volume in Geriatric Depression

were then prescribed perpendicular to this line, covering the
entire brain from just anterior to the front of the temporal lobe to
posterior to the ventricles.
MR IMAGE PROCESSING. MrX Procedure (for Whole
Brain) The basic segmentation protocol is a modified version
of that developed by Kikinis et al (1992) and has been described
previously (Byrum 1996). Gray and white contrast is often poor
in MR scans of the elderly. Our seeding protocol, which
identified the range of signal intensities that characterize each

tissue type, was altered to address this problem. Gray matter
points were selected by moving the cursor from regions of
cerebrospinal fluid (CSF) on the outside of the brain to adjacent
cortical gray areas while observing a two-dimensional scatter
plot showing the image intensity for echo 1 graphed against the
image intensity for echo 2. This eliminated the need to distinguish the cortical gray regions from white matter, which can be
nearly isointense on some scans. When the cursor first enters the
cortical gray region, the location of the point in the scatter plot
changes distinctly, thus identifying the gray matter region. Points
for white matter seeding were selected from the anterior and
posterior corpus callosum, as well as the central region of white
matter tracts in each quadrant. The rationale was to exclude
points that were anywhere near gray matter, and therefore least
likely to be gray matter.
Once the brain was segmented into tissue types and the
nonbrain tissue stripped away through a masking procedure,
specific regions of interest (ROI) were assessed using tracing and
connectivity functions. The cerebral hemispheres and caudate
nuclei were traced and a mask was created that could be applied
to the segmented brain. A connectivity function was used for

quantifying the lateral ventricles.
The final step was to run a summarizing program that
calculated the volume of each tissue type within the specific ROI.
Volume was determined for the whole brain.

Grid Procedures (for Hippocampus) The Grid Program
(developed by one of the authors, TFB) was used to quantify the
left and right putamen and hippocampi. The Grid Program allows
for a highly reliable, semiautomated determination of ROI
volumes and is based on a manual point-counting method
(MacFall et al 1994).
Our definition of the hippocampus was as follows: on each
scan, we began with the most posterior coronal slice and then
moved anteriorly. We began measuring the hippocampus when
the pulvinar nucleus of the thalamus obscured the crura fornicis;
if the crus was only obscured on one side, we only measured that
side. The fimbria, which extends from the superior surface of the
hippocampus across the CSF into the white matter above, was
transected at its narrowest point. Along the medial border of the
hippocampus, the thin strip of gray matter was cut at its
narrowest point, and tracing then continued around the hippocampal body to the starting point. The amygdala-hippocampal
transition zone appeared as a diffuse area of gray matter between
the anterior portion of the hippocampus and the posterior portion
of the amygdala; as with the fimbria, this area was transected at
its narrowest point, which was usually found between the inferior
lateral ventricles and the cistern. Continuing anteriorly, the

BIOL PSYCHIATRY
2000;48:301–309

303

inferior lateral ventricles gradually shift from a vertical to a
horizontal orientation but remain superior to the hippocampus.
We defined the anterior border of the hippocampus as the slice
on which the inferior lateral ventricles appeared horizontally
without any body of gray matter visible below them. All
technicians received extensive training by experienced volumetric analysts. Reliability was established (kappa . 0.9) by repeat
measurements on multiple MR scans in younger subjects before
raters were approved to process study data. In addition, an
ongoing reliability study was conducted to insure that the quality
of volumetric analyses was maintained throughout the study.
When examining older subjects, our interrater reliability was
0.79 for left hippocampus and 0.69 for right hippocampus.

Statistical Analysis
INITIAL ANALYSES. Right and left hippocampal volumes
(HVs) between groups was examined using t tests for age (young
vs. old with median age as cut-off), gender, race, depression
versus control-subject status, and age of onset (patient group).
Continuous variables (age, MMSE score, and total brain volume)
were examined using analysis of variance (ANOVA) with the
General Linear Models procedure (SAS Institute, Cary, NC).
Differences between depressed patients and control subjects
were analyzed in a series of ANOVA models (including forward
selection models) with HV as the independent variable controlling for age, gender, and total brain volume. Among the
depressed group, a series of ANOVAs (including forward selection) was performed to examine right and left HVs with age,
gender, age of depression onset, number of previous depressive
episodes, and total brain volume as independent variables. For
age of onset analyses, we used an age cut-off of 45 years to
distinguish between early-onset and late-onset depression, an age
on which we have reported previously (Krishnan et al 1994).
Analysis of variance was also used to examine differences on
MMSE score within the patient group by age of onset.
POST HOC ANALYSES. We observed that in the sample of
both elderly control subjects and depressed subjects, the vast
majority with right or left HV below 3 mL (close to the median)
were depressed patients (Figures 1 and 2). We compared group
membership (control vs. patients) for subjects with small HV (3
mL or less) versus subjects with large HV (greater than 3 mL)
using a chi-square test. Right and left HVs, dichotomized into
small versus large groups, were then used as independent
variables in a series of logistic regression models controlling for
age, gender, race, and total brain volume.

Results
The Sample
Table 1 shows demographic characteristics of the sample.
Patient subjects were significantly older than control
subjects, and there were significantly more women among
patient subjects compared with control subjects. In analyses that used total brain volume, the total number was
reduced to 75 because seven patient subjects and two

304

D.C. Steffens et al

BIOL PSYCHIATRY
2000;48:301–309

in the table, white subjects had larger hippocampal volumes than African American subjects, and this difference
reached significance for left HV. Further analyses revealed
that African American subjects tended to be older than
white subjects, and that the proportion of women was
higher among African Americans. In subsequent analyses
that controlled for age and gender, the race variable lost
significance.

Analyses Performed within Patient and
Control Groups

Figure 1. Numbers of subjects for each of 10 0.250-mL ranges
of right hippocampal volume. ■, control subjects; h, patients.

control subjects were missing that variable. Figures 1 and
2 show the distribution of right and left HVs for patient
and control subjects. We determined that a cut-off below
3 mL for both right and left HVs identified a subgroup of
subjects comprised almost entirely of patient subjects.
This observation serves as the basic for post hoc analyses
(see below). There was no difference between left and
right HV scores for the sample (p 5 .268, t 5 21.112).

Bivariate Analyses
Table 2 demonstrates mean values for right and left HVs.
In bivariate analyses, uncontrolled for age, gender, and
race, depressed subjects had significantly smaller HV than
did control subjects for both right HV (p 5 .003, t 5
23.088) and left HV (p 5 .014, t 5 22.509). As shown

In analyses using ANOVA, there was an association
among patient subjects between right HV and total brain
volume (p 5 .007) and a nonsignificant association
between right HV and MMSE score (p 5 .053), whereas
there was no association between right HV and age (p 5
.097). In control subjects, there was a significant association between right HV and total brain volume (p 5 .033),
but no association between right HV and age (p 5 .411) or
MMSE score (p 5 .577). Among patients, left HV was
associated with MMSE score (p 5 .039) and to a nonsignificant degree with total brain volume (p 5 .058), but not
with age (p 5 .143). In the control group, left HV was
associated with total brain volume (p 5 .048), but there
were no associations between left HV and age (p 5 .680)
or MMSE score (p 5 .880).

Age of Onset
Among depressed patients, age of onset was negatively but
not significantly related to right HV (p 5 .052) and to left
HV (p 5 .062). Number of reported episodes was unrelated to either right or left HV (p 5 .999 and .942,
respectively). In the patient group, we also dichotomized
age of onset into early onset (,45 years, n 5 28, mean age
of onset 5 25.6 years) and late onset ($45 years, n 5 38,
mean age of onset 5 65.2 years). There were no gender
differences by age of onset, with women comprising 75%
of early-onset cases and 79% of late onset cases (x2 5
0.143, p 5 .705). Mini-Mental State Examination score
was significantly associated with right HV (p 5 .016) and
left HV (p 5 .018) for the late-onset depressed patients,
but not for the group with early-onset depression (p 5 .358
for right HV and p 5 .543 for left HV).

Initial Regression Analyses

Figure 2. Numbers of subjects for each of 10 0.250-mL ranges
of left hippocampal volume. ■, control subjects; h, patients.

Table 3 demonstrates the results of linear regression
models with right and left HVs as the independent variables. Accounting for the effects of age, gender, and total
brain volume, depressed patients tended to have smaller
right and left HVs. In forward selection models, the

Hippocampal Volume in Geriatric Depression

BIOL PSYCHIATRY
2000;48:301–309

305

Table 1. Characteristics of the Sample

Age, mean (SD)
Women, n (%)
Race (%)
White
African American
Other
MMSE score, mean (SD)

Depressed subjects (n 5 66)

Control subjects (n 5 18)

p value

71.74 (8.42)
51 (77.27)

67.11 (5.04)
9 (50.00)

.005a
.023b

56 (84.85)
9 (13.64)
1 (1.52)
27.16 (3.10)

17 (94.44)
1 (5.56)
0 (0)
28.83 (0.99)

.683c
.0004d

a

t 5 22.938.
x 5 5.155.
Fisher’s exact test, 2-tailed, for white vs. African-American subjects.
d
Baseline Mini-Mental State Examination (MMSE) score was available on 63 depressed subjects; t 5 3.684.
b 2
c

negative association between depressed patients and HV
was significant for right HV (p 5 .014) and close to
significance for left HV (p 5 .073). Table 4 contains
results of linear regression models using forward selection
models demonstrating a nonsignificant negative relationship between age of onset and right HV (p 5 .153) and left
HV (p 5 .076).

Post Hoc Analyses
We report analyses that have not undergone statistical
correction for post hoc comparison. From Figure 1, 39 of
66 (59.1%) depressed subjects and 3 of 18 (16.7%) control
subjects had a right HV less than or equal to 3 mL
[x2(1) 5 10.182, p , .001]. In Figure 2, 44 of 66 (66.7%)
depressed subjects and 5 of 18 (27.8%) control subjects
had a left HV less than or equal to 3 mL [x2(1) 5 8.800,
p 5 .003]. In a logistic regression model (Table 5) with

right HV as a dichotomous independent variable, when we
controlled for age, gender, and total brain volume, depression status (patient vs. control subject) was a significant
predictor (odds ratio 5 7.945, CI 5 1.455– 43.400, p 5
.017). Similarly, in a logistic regression model for left HV
(Table 5), depression status remained significant (odds
ratio 5 4.413, CI 5 1.133–17.184, p 5 .032). C values for
the models of right HV and left HV were 0.803 and 0.779,
respectively.
We were also interested in characterizing the group of
patients with right and left HVs less than 3 mL. Bivariate
analysis using ANOVA demonstrated that for this group,
age of onset was negatively associated with right HV (p 5
.193) with a similar trend for left HV (p 5 .079). In a
linear regression model with age of onset, age, gender, and
total brain volume, none of the independent variables were
significant predictors of either right or left HV.

Table 2. Mean Values (SD) for Right and Left Hippocampal Volumes (HVs; mL)

Age (median age 5 70.5)
Less than 70.5 (n 5 42)
70.5 or older (n 5 42)
Gender
Men (n 5 24)
Women (n 5 60)
Male patients (n 5 15)
Female patients (n 5 51)
Male control subjects (n 5 9)
Female control subjects (n 5 9)
Race
White (n 5 73)
African American (n 5 10)
Other (n 5 1)
Depression status
Patients (n 5 66)
Control subjects (n 5 18)
Age of onset (patients)
Onset , 45 years (n 5 28)
Onset $45 years (n 5 38)
a

Right HV

t score, pa

3.12 (0.41)
2.97 (0.42)

1.750, .084

3.33 (0.42)
2.93 (0.36)
3.24 (0.39)
2.90 (0.35)
3.49 (0.46)
3.11 (0.34)
3.06 (0.42)
2.83 (0.31)
3.78

4.420, ,.001
3.183, .002
2.022, .060

21.744, .085

Left HV

t score, pa

3.03 (0.36)
2.92 (0.42)

1.251, .214

3.23 (0.40)
2.87 (0.34)
3.15 (0.34)
2.86 (0.34)
3.38 (0.46)
2.97 (0.34)
3.01 (0.39)
2.72 (0.29)
3.27

4.195, .0001
2.911, .005
2.153, .047

22.243, .028

2.98 (0.39)
3.30 (0.44)

23.088, .003

2.92 (0.36)
3.17 (0.44)

22.509, .014

3.06 (0.38)
2.92 (0.39)

1.467, .147

3.00 (0.35)
2.87 (0.36)

1.437, .156

t test was used for all comparisons; for race the test compared whites and African Americans.

306

D.C. Steffens et al

BIOL PSYCHIATRY
2000;48:301–309

Table 3. Linear Regression Models for Depressed Subjects vs. Control Subjects Controlling for
Age, Gender, and Total Brain Volume (Model 1) and Using Forward Selection for Significant
Variables
Right hippocampal volume
Variable
Model 1 (N 5 75 subjects)
Age
Gendera
Depression statusb
Total brain volume
Model R2
Model 2 (N 5 75 subjects)
Gendera
Depression statusb
Total brain volume
Model R2
a
b

Parameter
estimate (SE)

p value

20.0085 (0.0058)
20.1306 (0.1217)
20.2097 (0.1092)
0.0008 (0.0003)
.294

.149
.287
.059
.037

20.2673 (0.1062)
0.0010 (0.0003)
.256

.014
.0004

Left hippocampal volume
Parameter
estimate (SE)

p value

20.0055 (0.0057)
20.1752 (0.1202)
20.1557 (0.1079)
0.0005 (0.0004)
.211

.338
.149
.153
.203

20.2855 (0.0943)
20.1908 (0.1048)

.003
.073

.182

Female subject 5 0, male subject 5 1.
Control subject 5 0, depressed subject 5 1.

Discussion
Our study supports the hypothesis that depressed elderly
patients have smaller HV than a group of nondepressed
elderly control subjects. In some of our models controlling
for age, gender, and total brain volume, we were able to
demonstrate differences between depressed patients and
control subjects. Additionally, among depressed patients,
those with older age of onset had smaller HV. Our study
reports on 66 depressed patients and 18 elderly control
subjects; therefore, sample size considerations suggest
caution when interpreting results. Number of subjects may
also explain why some analyses yielded significant results
(particularly the logistic regression models), whereas others demonstrated statistical trends.
Two previous studies that reported decreased HV in
major depression (Axelson et al 1993; Sheline et al 1996)
examined a wide range of age of subjects; our study

focused on geriatric patients. Axelson et al (1993) in
measuring the amygdala-hippocampal complex (AHC)
found a negative relationship between AHC volume and
age of onset of depression (significant on the left, trending
on the right) and a significant negative correlation between
left AHC volume and number of hospitalization with a
trend for trend right AHC volume. Sheline et al (1996)
reported that days of depression were negatively and
significantly associated with HV, and Axelson et al (1993)
found similar trends between duration of illness and AHC
volume. Sheline et al (1996) concluded that this was
consistent with the glucocorticoid toxicity hypothesis put
forward by Sapolsky (1993). Our findings are only mildly
supportive of that hypothesis (with a trend toward smaller
HV in patients with early-onset depression compared with
control subjects) and are rather more supportive of the
observation that late-onset depression is a risk factor for
Alzheimer’s disease (Jorm et al 1991; Kokmen et al 1991;

Table 4. Linear Regression Models for Right and Left Hippocampal Volumes Using Forward
Selection with Age of Depression Onset, Age, Gender, and Total Brain Volume as Potential
Variables
Right hippocampal volume
Variable
Age
Gendera
Age of depression onset
Total brain volume
Model R2

Parameter
estimate (SE)
20.0061 (0.0064)
20.2080 (0.1364)
20.0033 (0.0023)
0.0005 (0.0004)
.218

p value
.341
.133
.153
.269

Left hippocampal volume
Parameter
estimate (SE)

p value

20.2803 (0.1063)
20.0036 (0.0020)

.011
.076

.149

Model of left hippocampal volume contains only gender and age of onset because in four-variable forward selection model,
age and total brain volume both had p . .5 and were excluded.
a
Female subject 5 0, male subject 5 1.

Hippocampal Volume in Geriatric Depression

BIOL PSYCHIATRY
2000;48:301–309

307

Table 5. Logistic Regression Models for Controlling for Age, Gender, Depression Status, and
Total Brain Volume (Model 1) and with Forward Regression (Model 2)
Variable
Model 1: full model (n 5 75 subjects)
Age
Gendera
Depression statusb
Total brain volume
Model c value
Model 2: Forward selection (n 5 75 subjects)
Gendera
Depression statusb
Total brain volume
Model c value

Right hippocampal
volume

Left hippocampal
volume

1.021 (0.951–1.096)
2.900 (0.647–13.006)
7.945 (1.455– 43.400)
0.996 (0.991–1.001)
.803

1.023 (0.952–1.099)
7.145 (1.545–33.042)
4.413 (1.133–17.184)
1.002 (0.997–1.006)
.779

9.282 (1.777– 48.493)
0.995 (0.990 – 0.999)
.780

5.263 (1.662–16.660)
4.736 (1.255–17.870)
.739

Odds ratios with confidence intervals for right and left hippocampal volumes dichotomized at , 3 mL and $ 3 mL.
a
Female subject 5 0, male subject 5 1.
b
Depressed subject 5 0, control subject 5 1.

Speck et al 1995; Steffens et al 1997), particularly with our
finding of an association between MMSE score and HV
among depressed individuals. Although we did not measure directly days of depression, our finding of a significant negative relationship between age of onset and HV, as
well as a lack of association between HV and number of
depressive episodes, did not support those findings.
How do our findings relate to those of Sapolsky
(1993) and Sheline et al (1996)? Although they only
mildly support their findings, glucocorticoid function in
geriatric depression may express itself in a different
manner from other clinical populations. For example,
the severe, late-onset depression experienced by many
of our subjects may be associated with greater hypothalamic-pituitary-adrenal axis dysregulation or with more
comorbid medical illness. Such differences may relate
to severity of illness or the underlying pathophysiology
of illness. Thus, higher glucocorticoid exposure may be
related to the intensity and biology of the illness in
addition to its duration. In support of this hypothesis (in
results not shown), we found that patients with severe
depression had smaller HV compared with other depressed individuals, with the difference on the left being
greater, but these results were not statistically significant. Axelson et al (1993) found no relationship between severity of depression and AHC volume.
Alternatively, it may be that depressed subjects in
this sample are comprised of patients with either lateonset preclinical dementia or early-onset recurrent major depression. Those with preclinical dementia may not
be experiencing any glucocorticoid-induced neurotoxicity, but they may have experienced a much larger
decrease in hippocampal volume as a consequence of
the dementing process. Simply put, the changes in
hippocampal volume may be greater in late-onset depression compared with early-onset depression because

the pathologic processes associated with late-onset
depression exert greater hippocampal effects than does
glucocorticoid neurotoxicity in patients with early-onset
depression. Clearly, more investigations using larger
samples of patients are needed to clarify the relationship
between duration of illness, severity of illness, stress
level, and hippocampal volume.
One potential limitation of our study was the timing of
the MRI scans in relation to entry into the study. With our
protocol, patients had a 2-month window upon entry into
the study to have a MRI scan. If HV is associated with
state effects of depression (e.g., higher cortisol levels),
then patients who underwent an MRI scan when they were
most acutely depressed might have smaller HV than those
who were unable to have an MRI upon entry to the study.
Coffey et al (1991), however, found no difference between
baseline, 2-week and 6-month volumes of the amygdalahippocampal complex in depressed patients receiving
electroconvulsive therapy. Thus, it is doubtful that over
the course of 2 months, much change in HV should occur.
We also may face limitations from our scanning
method. We used 3-mm MRI slices that may have limited
our ability to define the hippocampus or differentiate it
from surrounding structures. It also may have led to partial
volume artifacts. Lack of clarity between of hippocampal
boundaries in these older subjects may have contributed to
the suboptimal interrater reliabilities.
Caution is also warranted in the interpretation of our
post hoc results. We did not perform a statistical correction
to account for post hoc analysis. It became clear upon
inspection of the data presented in Figures 1 and 2 that
further analyses using 3 mL as a cutoff might yield
interesting results. We then performed four sets of analyses for right and left HVs. The initial chi-square analyses
of depression status by volume were highly significant and
remain significant after post hoc statistical correction. The

308

BIOL PSYCHIATRY
2000;48:301–309

subsequent logistic models (p 5 .017 for right HV, p 5
0.032 for left HV) may lose significance. Future studies
that dichotomize the volume of the hippocampus should
employ larger samples of depressed patients to overcome
post hoc correction.
Clinical longitudinal research on hippocampal volume and function is needed to address the role of the
hippocampus in affective disorders. Such studies will
need to focus on both depression and cognitive outcomes. In particular, studies will need to examine HV in
older depressed patients who subsequently develop
dementia. Alzheimer’s disease is associated with hippocampal atrophy, whereas vascular dementia may not
be. Thus, the study of hippocampal volume in geriatric
depression may inform the relationship between depression and Alzheimer’s disease.

This study was supported by NIMH Grants Nos. P 30 MH40159, R01
MH54846, and K07 MH01367. Presented at the Annual Meeting of the
American Association for Geriatric Psychiatry, March 13–17, 1999.

References
Axelson DA, Doraiswamy PM, McDonald WM, Boyko OB, Tupler
LA, Patterson LJ, et al (1993): Hypercortisolemia and hippocampal changes in depression. Psychiatry Res 47:163–173.
Byrum CE, MacFall JR, Charles HC, Chitilla VR, Boyko OB,
Upchurch L, et al (1996): Accuracy and reproducibility of
brain and tissue volumes using a magnetic resonance segmentation method. Psychiatry Res 67:215–234.
Carroll BJ, Feinberg M, Greden JF, Tarika J, Albala AA, Haskett
RF, et al (1981a): A specific laboratory test for the diagnosis
of melancholia. Standardization, validation, and clinical utility. Arch Gen Psychiatry 38:15–22.
Carroll BJ, Feinberg M, Smouse PE, Rawson SG, Greden JF
(1981b): The Carroll Rating Scale for Depression. I. Development, reliability, and validation. Br J Psychiatry 138:194 –200.
Coffey CE, Weiner RD, Djang WT, Figiel GS, Soady SA,
Patterson LJ, et al (1991): Brain anatomic effects of electroconvulsive therapy. A prospective magnetic resonance imaging study. Arch Gen Psychiatry 48:1013–1021.
Coffey CE, Wilkinson WE, Weiner RD, Parashos IA, Djang WT,
Webb MC, et al (1993): Quantitative cerebral anatomy in
depression. A controlled magnetic resonance imaging study.
Arch Gen Psychiatry 50:7–16.
Endicott J, Spitzer RL, Fleiss JL, Cohen J (1976): The Global
Assessment Scale: A procedure for measuring overall severity of psychiatric disturbance. Arch Gen Psychiatry 33:766 –
771.
Folstein MF, Folstein SE, McHugh PR (1975): “Mini-mental
state.” A practical method for grading the cognitive state of
patients for the clinician. J Psychiatr Res 12:189 –198.
George LK, Blazer DG, Hughes DC, Fowler N (1989): Social
support and the outcome of major depression. Br J Psychiatry
154:478 – 485.
Hachinski VC, Iliff LD, Zilhka E, Du Boulay GH, McAllister

D.C. Steffens et al

VL, Marshall J, et al (1975): Cerebral blood flow in dementia.
Arch Neurol 32:632– 637.
Hamilton M (1960): A rating scale for depression. J Neurol
Neurosurg Psychiatry 23:56 – 61.
Jack CR Jr, Petersen RC, Xu YC, O’Brien PC, Waring SC,
Tangalos EG, et al (1998): Hippocampal atrophy and apolipoprotein E genotype are independently associated with
Alzheimer’s disease. Ann Neurol 43:303–310.
Jorm AF, van Duijn CM, Chandra V, Fratiglioni L, Graves AB,
Heyman A, et al (1991): Psychiatric history and related
exposures as risk factors for Alzheimer’s disease: A collaborative re-analysis of case-control studies. Int J Epidemiol
20(suppl 2):S43–S47.
Kikinis R, Shenton ME, Gerig G, Martin J, Anderson M, Metcalf
D, et al (1992): Routine quantitative analysis of brain and
cerebrospinal fluid spaces with MR imaging. J Magn Reson
Imaging 2:619 – 629.
Kokmen E, Beard CM, Chandra V, Offord KP, Schoenberg BS,
Ballard DJ (1991): Clinical risk factors for Alzheimer’s
disease: A population-based case-control study. Neurology
41:1393–1397.
Krishnan KR, Doraiswamy PM, Figiel GS, Husain MM, Shah
SA, Na C, et al (1991): Hippocampal abnormalities in
depression. J Neuropsychiatry Clin Neurosci 3:387–391.
Krishnan KR, Hays JC, Blazer DG (1997): MRI-defined vascular
depression. Am J Psychiatry 154:497–501.
Krishnan KRR, Ritchie JC Jr, Tupler LA, McDonald WM,
Knight D, Nemeroff CB, et al (1994): Apolipoprotein E
epsilon 4. Neurology 44:2420 –2421.
Laakso MP, Soininen H, Partanen K, Lehtovirta M, Hallikainen
M, Hanninen T, et al (1998): MRI of the hippocampus in
Alzheimer’s disease: sensitivity, specificity, and analysis of
the incorrectly classified subjects. Neurobiol Aging 19:23–31.
Landerman R, George LK, Campbell RT, Blazer DG (1989):
Alternative models of the stress buffering hypothesis. Am J
Community Psychol 17:626 – 642.
Linn BS, Linn MW, Gurel L (1968): Cumulative illness rating
scale. J Am Geriatr Soc 16:622– 626.
MacFall JR, Byrum CE, Parashos I, Early B, Charles HC,
Chittilla V, et al (1994): Relative accuracy and reproducibility
of regional MRI brain volumes for point-counting methods.
Psychiatry Res 55:167–177.
McEwen BS (1997): Possible mechanisms for atrophy of the
human hippocampus. Mol Psychiatry 2:255–262.
Mongeau R, Blier P, de Montigny C (1997): The serotonergic
and noradrenergic systems of the hippocampus: Their interactions and the effects of antidepressant treatments. Brain Res
Brain Res Rev 23:145–195.
Montgomery SA, Asberg M (1979): A new depression scale
designed to be sensitive to change. Br J Psychiatry 134:382–
389.
Morris JC, Mohs RC, Rogers H, Fillenbaum G, Heyman A
(1988): Consortium to Establish a Registry for Alzheimer’s
Disease (CERAD) clinical and neuropsychological assessment of Alzheimer’s disease. Psychopharmacol Bull 24:641–
652.
O’Brien JT, Desmond P, Ames D, Schweitzer I, Chiu E, Tress B
(1997): Temporal lobe magnetic resonance imaging can
differentiate Alzheimer’s disease from normal ageing, depres-

Hippocampal Volume in Geriatric Depression

sion, vascular dementia and other causes of cognitive impairment. Psychol Med 27:1267–1275.
Radloff LS (1977): The CES-D scale: A self-report depression
scale for research in the general population. Appl Psychol
Measures 1:385– 401.
Robins LN, Helzer JE, Croughan J, Ratcliff KS (1981): National
Institute of Mental Health Diagnostic Interview Schedule: Its
history, characteristics, and validity. Arch Gen Psychiatry
38:381–389.
Sapolsky RM (1993): Potential behavioral modification of glucocorticoid damage to the hippocampus. Behav Brain Res
57:175–182.
Sheline YI (1996): Hippocampal atrophy in major depression: A
result of depression-induced neurotoxicity. Mol Psychiatry
1:298 –299.

BIOL PSYCHIATRY
2000;48:301–309

309

Sheline YI, Wang PW, Gado MH, Csernansky JG, Vannier MW
(1996): Hippocampal atrophy in recurrent major depression.
Proc Natl Acad Sci U S A 93:3908 –3913.
Soares JC, Mann JJ (1997): The anatomy of mood disorders—
review of structural neuroimaging studies. Biol Psychiatry
41:86 –106.
Speck CE, Kukull WA, Brenner DE, Bowen JD, McCormick
WC, Teri L, et al (1995): History of depression as a risk factor
for Alzheimer’s disease. Epidemiology 6:366 –369.
Spitzer RL (1983): Psychiatric diagnosis: Are clinicians still
necessary? Compr Psychiatry 24:399 – 411.
Steffens DC, Plassman BL, Helms MJ, Welsh-Bohmer KA,
Saunders AM, Breitner JC (1997): A twin study of late-onset
depression and apolipoprotein E epsilon 4 as risk factors for
Alzheimer’s disease. Biol Psychiatry 41:851– 856.