Psychiatry Research 93 Ž2000. 93]102
Low HDL cholesterol, aggression and altered central
serotonergic activity
Laure Buydens-Branchey U , Marc Branchey, Jeffrey Hudson,
Paul Fergeson
Department of Psychiatry, SUNY Health Science Center at Brooklyn, and VA New York Harbor Health Care System,
Brooklyn Campus, Brooklyn, NY, USA
Received 30 June 1999; received in revised form 7 December 1999; accepted 12 December 1999
Abstract
Many studies support a significant relation between low cholesterol levels and poor impulse, aggression and mood
control. Evidence exists also for a causal link between low brain serotonin Ž5-HT. activity and these behaviors.
Mechanisms linking cholesterol and hostile or self-destructive behavior are unknown, but it has been suggested that
low cholesterol influences 5-HT function. This study was designed to explore the relationship between plasma
cholesterol, measures of impulsivity and aggression, and indices of 5-HT function in personality disordered cocaine
addicts. Thirty-eight hospitalized male patients Žage 36.8" 7.1. were assessed with the DSM-III-R, the Buss]Durkee
Hostility Inventory ŽBDHI., the Barratt Impulsiveness Scale ŽBIS. and the Brown]Goodwin Assessment for Life
History of Aggression. Fasting basal cholesterol Žtotal, LDL and HDL. was determined 2 weeks after cocaine
discontinuation. On the same day 5-HT function was assessed by neuroendocrine Žcortisol and prolactin. and
psychological ŽNIMH and ‘high’ self-rating scales. responses following meta-chlorophenylpiperazine Ž m-CPP. challenges. Reduced neuroendocrine responses, ‘high’ feelings and increased ‘activation-euphoria’ following m-CPP have
been interpreted as indicating 5-HT alterations in a variety of psychiatric conditions. Significantly lower levels of
HDL cholesterol were found in patients who had a history of aggression Ž Ps 0.005.. Lower levels of HDL
cholesterol were also found to be significantly associated with more intense ‘high’ and ‘activation-euphoria’ responses
as well as with blunted cortisol responses to m-CPP Ž Ps 0.033, Ps 0.025 and Ps 0.018, respectively.. This study
gives further support to existing evidence indicating that in some individuals, the probability of exhibiting impulsive
U
Corresponding author. VA New York Harbor Health Care System, Brooklyn Campus, 800 Poly Place, Brooklyn, NY 11209,
USA. Tel.: q1-718-630-2967; fax: q1-718-630-2954.
E-mail address: [email protected] ŽL. Buydens-Branchey .
0165-1781r00r$ - see front matter Q 2000 Elsevier Science Ireland Ltd. All rights reserved.
PII: S 0 1 6 5 - 1 7 8 1 Ž 9 9 . 0 0 1 2 6 - 2
94
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
and violent behaviors may be increased when cholesterol is low. It also suggests that low cholesterol and alterations
in 5-HT activity may be causally related. Q 2000 Elsevier Science Ireland Ltd. All rights reserved.
Keywords: Violence; Impulsivity; Serotonin; Personality disorder; Cocaine-related disorders
1. Introduction
Recent studies suggest the existence of associations between low or lowered total cholesterol
levels and increased mortality from suicide, violence and accidents ŽMuldoon et al., 1990.. Associations have also been found between low
cholesterol and symptoms or behaviors that did
not result in final outcomes such as depression,
suicidal ideation, suicide attempts or violence.
Evidence for these associations is derived from
prevention trials aimed at lowering cholesterol
concentrations by diet, drugs or both, from
prospective epidemiological studies in which follow-up was done in individuals with varying levels
of plasma cholesterol who did not undergo specific cholesterol lowering interventions, and from
clinical studies in which cholesterol and psychopathology were assessed concomitantly Žfor
review, see Muldoon et al., 1993..
Behaviors associated with low cholesterol have
also been observed in patients with a deficient
serotonergic neurotransmission. This has
prompted some investigators to assess the existence of associations between low cholesterol and
indices of altered serotonergic function and psychopathology, and to wonder whether low cholesterol and reduced serotonergic activity could be
causally related. To date, evidence supporting the
existence of a link between reduced cholesterol
levels and alterations in 5-HT function is still
scant.
In order to examine the existence of relationships between plasma cholesterol, psychopathology and indices of serotonergic neurotransmission in humans, we measured plasma cholesterol in a group of abstinent personality disordered cocaine addicts whose serotonergic function had been assessed with the serotonergic
probe meta-chlorophenylpiperazine Ž m-CPP..
Neuroendocrine and psychological changes following m-CPP administration were the topic of a
first study ŽBuydens-Branchey et al., 1997.. Because only a limited amount of research has been
undertaken to determine the possible role of
high-density lipoprotein ŽHDL. cholesterol and
low-density lipoprotein ŽLDL. cholesterol in psychiatric disorders and in serotonergic neurotransmission, we decided to measure plasma levels of
HDL and LDL cholesterol in addition to those of
total cholesterol in the patients who participated
in this investigation.
2. Methods
2.1. Patients
Patients selected for participation in the study
were 38 men whose age Žmean " S.D.. was 36.8"
7.1 years. They were studied while hospitalized on
a locked inpatient drug rehabilitation unit. They
did not receive any medication during their stay
in the hospital. Patients were excluded from the
study if they had a major physical illness or
abnormalities on liver function tests. Patients with
a history of intravenous use of any substance and
patients who had used opiates in any form during
the year preceding admission were also excluded
from the study. They were screened with the
SCID ŽSpitzer et al., 1990; First et al., 1997a,b.
and were enrolled in the study if they met DSMIII-R criteria for cocaine dependence but did not
meet criteria for any other Axis I disorder Žincluding dependence on any substance besides cocaine.. The presence of one or more Axis II
disorders was not an exclusionary criterion.
Patients were also administered the Barratt Impulsiveness Scale ŽBarratt, 1957. and the
Buss]Durkee Hostility Inventory ŽBuss and Durkee, 1957.. The Buss]Durkee score was arrived at
by adding the scores on seven hostility subscales
and one suspicion subscale. A modified version of
the Brown]Goodwin Assessment for Life History
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
of Aggression was also part of our test battery
ŽBrown et al., 1981.. This questionnaire was aimed
at eliciting information about the following behavioral categories: problems with discipline in
the armed forces; problems with discipline at
work; non-specific fighting; specific assaults on
other persons; property damage; antisocial behavior involving the police or not; and arrests or
incarceration for assaultive or other behaviors. In
each category, non-occurrence was scored as 0;
one event was scored as 1; two events were scored
as 2; three, ‘several,’ or ‘frequent’ events were
scored as 3; and four or more, ‘many,’ or ‘numerous’ events were scored as 4. Patients with a score
of 8 or more were considered to have a history of
aggression.
2.2. m-CPP challenge procedure
After having signed informed consent, each
subject participated in two test sessions during
which either m-CPP Ž0.5 mgrkg. or placebo was
given orally as identical-looking capsules. The order of m-CPP and placebo administration was
random and capsules were given according to a
double-blind design. The time interval between
the administration of m-CPP and placebo was 72
h. Testing took place from 12 to 19 days after
discontinuation of cocaine use. Subjects were instructed to fast after midnight on test days. During the test sessions, they remained in a semirecumbent position, stayed awake, abstained from
smoking, and continued to fast until the end of
the session. Blood samples were drawn for prolactin and cortisol determinations, 30 and 15 min
before and 30, 60, 90, 120, 180, and 210 min after
m-CPP and placebo administration. Total, HDL
and LDL cholesterol were determined in the
sample collected 15 min before m-CPP administration. Psychological changes occurring during
the challenge procedures were assessed with a
modified version of the National Institute of
Mental Health ŽNIMH. self-rating scale, which
comprises 24 items rated on a 4-point scale Žnonexistent s 0, milds 1, moderate s 2, markeds 3.
ŽMurphy et al., 1989.. This scale is divided into
the following six subscales: anxiety, activationeuphoria, depressive affect, dysphoria, altered
95
self-reality, and functional deficit. A ‘high’ scale,
which is not part of the NIMH scale but has been
utilized by other investigators, was used as well
ŽCharney et al., 1987.. The four-point rating system on this scale was the same as that used for
the NIMH scale. The NIMH and ‘high’ scales
were used 30 min before and 30, 90, 150, and 210
min after m-CPP and placebo administration.
2.3. Biochemical determinations
Plasma prolactin and cortisol levels were measured with radioimmunoassay kits from ‘ICM
Biomedical Inc.’ Intraassay and interassay coefficients of variance were 6.1% and 8.3% for prolactin, and 6.7% and 8.9% for cortisol, respectively. Plasma m-CPP levels were measured according to the method of Suckow et al. Ž1990..
Intraassay and interassay coefficients of variance
were 6.7% and 8.6%, respectively.
Total cholesterol was measured by an enzymatic method similar to that described by Allain
et al. Ž1974.. HDL cholesterol was separated by
precipitation of LDL and VLDL lipids, using
dextran sulfate and magnesium chloride ŽWarnick
et al., 1983. and measured in the supernatant by
the same enzymatic method as total cholesterol.
LDL cholesterol was calculated by subtracting
HDL cholesterol and triglycerides from total
cholesterol.
2.4. Data analysis
The levels of total, HDL and LDL cholesterol
of aggressive and non-aggressive patients were
compared with analyses of covariance ŽANCOVAs., using age and weight as covariates.
Associations between continuous variables
Žsuch as Barratt and Buss]Durkee scores. and
levels of cholesterol were done with partial correlations, controlling for the effects of age and
weight.
GLM Žgeneral linear model. ANCOVAs, assessing separately the effect of individual continuous predictors Žcalled covariates. on dependent
variables, were used to assess the significance of
associations between total cholesterol and its subfractions and the neuroendocrine and psychologi-
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
96
cal responses to m-CPP ŽMcCullagh and Nelder,
1989.. More specifically, GLM repeated measures
ANCOVAs were used to test the effect of either
total, HDL or LDL cholesterol Žcovariates. on
sequential baseline- and placebo-corrected mCPP responses with age, weight and peak m-CPP
level as additional covariates.
The following indices of the intensity of psychological and neuroendocrine responses to m-CPP
were also calculated: delta max Žmaximum change
from baseline following m-CPP. and double delta
Ždifference between delta max and the change
from baseline at the corresponding time following
placebo..
3. Results
3.1. Pre-challenge assessments
3.1.1. History of Aggressionr Barratt and
Buss]Durkee scales
Eleven patients were determined to have had a
life history of aggression on the basis of their
answers to the modified version of the Brown]
Goodwin questionnaire. Patients who had a positive score in one of the aggression categories
usually had positive scores in other categories. Of
the 11 patients determined to be aggressive, 10
had a history of assaultive behavior. These behav-
iors included murder Žone patient., attempted
murder Žone patient., shooting and injuring others Žthree patients., stabbing Žtwo patients.,
threatening drug dealers with a gun to obtain
cocaine Žone patient., and repeated physical fights
without weapons Žnine patients.. Other behaviors
included armed robbery Žsix patients. and repeated episodes of property damage Žeight
patients.. Three patients had been jailed for
crimes involving violence. Patients who did not
meet criteria for a life history of aggression had
not engaged in the behaviors described above
with the exception of assaults and property damage, but far fewer patients in this group had
exhibited these behaviors and, when they did,
they reported fewer episodes than did individuals
categorized as having had a history of aggression.
As could be expected, the majority of patients in
both groups had committed crimes that did not
involve violence, such as burglaries, thefts without
confrontation of the victim, and receiving stolen
property.
The mean Buss]Durkee inventory and Barratt
scale scores for the entire population were 34.2"
11.3 and 81.0 " 13.6, respectively. Aggressive
patients had significantly higher Buss]Durkee
scores than non-aggressive patients Ž43.4" 9.7 vs.
30.5" 9.8; Ps 0.001.. They also had significantly
higher Barratt scale scores Ž92.5" 14.9 vs. 76.3"
9.8; Ps 0.001.. These data are shown in Table 1.
Table 1
Characteristics of patients with and without a history of aggressiona
Aggression history
Age
Weight Žkg.
Borderline personality disorder
Nr. of patients Ž%.
Antisocial personality disorder
Nr. of patients Ž%.
Barratt scale score
Buss]Durkee scale score
a
Positive
Ž n s 11.
Negative
Ž n s 27.
36.45" 6.56
79.45" 14.62
36.96" 7.39
78.70" 11.66
Between-group
comparisons
Žsignificance level.
N.S.
N.S.
6 Ž54.5%.
2 Ž7.4%.
Ps 0.001
11 Ž100%.
92.55" 14.93
43.45" 9.73
9 Ž33.3%.
76.26" 9.82
30.48" 9.76
Ps 0.001
Ps 0.001
Ps 0.001
The age, weight and Barratt and Buss]Durkee scores of aggressive and non-aggressive patients were compared with Student’s
t-tests. The numbers of patients meeting criteria for personality disorders were compared with Fisher’s exact tests.
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
3.1.2. Personality disorders
Cocaine addicts selected for this study did not
meet DSM-III-R criteria for any Axis I disorder
with the exception of psychoactive substance use
disorder but were not excluded if they met criteria for Axis II disorders. A sizable percentage
Ž65.8%. of patients met DSM-III-R criteria for at
least one of these Axis II disorders. The percentages of patients meeting criteria for the various
personality disorders were as follows: antisocial,
52.6%; passive-aggressive, 28.9%; borderline,
21.1%; narcissistic, 15.8%; paranoid, 13.2%;
avoidant, 10.6%; histrionic, 5.3%; self-defeating,
5.3%; schizotypal, 2.6%. No one met criteria for
dependent, obsessive]compulsive, or schizoid
personality disorders. Thirty-four percent of the
patient population Ž13 patients. met criteria for
two or more personality disorder diagnoses.
All patients with a life-long history of aggression met criteria for at least one personality disorder. The criteria they met most frequently were
those of Borderline Personality Disorder and Antisocial Personality Disorder. Six of the eight
Borderline Personality Disorder patients and 11
of the 20 Antisocial Personality Disorder patients
fell into the aggressive patients category Žsee
Table 1..
3.1.3. Cholesterol and beha¨ ioral characteristics
An ANCOVA, with age and weight as covariates, comparing the HDL values of aggressive and
non-aggressive patients revealed that the aggressive patients had significantly lower levels of HDL
cholesterol than patients without such a history
Ž30.6" 5.5 mgrdl vs. 40.3" 10.2 mgrdl; Ps
0.005.. These data are shown in Fig. 1.
Although total and LDL cholesterol were lower
in patients with a history of aggression, differences assessed with ANCOVAs did not reach
conventional levels of significance Ž141.6" 34.4
mgrdl vs. 158.0" 32.2 mgrdl for total cholesterol, Ps 0.086; and 84.8" 27.0 mgrdl vs. 100.2
" 30.8 mgrdl for LDL cholesterol, N.S...
Partial correlations controlling for the effects
of age and weight revealed a significant association between HDL cholesterol and the Buss]
Durkee score Ž r s y0.35; Ps 0.038. and a nonsignificant association between HDL cholesterol
97
Fig. 1. HDL cholesterol levels in cocaine addicts with and
without a history of aggression. Circles represent individual
patients. A comparison of the cholesterol levels of the two
patient groups with an ANCOVA with age and weight as
covariates revealed significantly lower levels of HDL cholesterol in the aggressive patients Ž Ps 0.005..
and the Barratt score Ž r s y0.24; Ps 0.153.. Associations between total and LDL cholesterol and
psychological scale scores were not significant.
3.2. Cholesterol and indices of serotonergic function
The statistical associations between HDL
cholesterol levels and the neuroendocrine and
psychological responses to m-CPP were tested
with GLM repeated measures ANCOVAs, using
HDL cholesterol, age, weight and m-CPP peak
levels as covariates.
Low HDL cholesterol levels were associated
with blunted cortisol responses. A GLM repeated
measures ANCOVA revealed a significant association between HDL cholesterol and the intensity
of the cortisol response Ž F1,33 s 6.18, Ps 0.018.
but not between HDL cholesterol and the prolactin response. The trivariate relationship
between the cortisol response double deltas, HDL
cholesterol levels and presence or absence of
aggressive tendencies is depicted in Fig. 2. This
figure was divided into four quadrants by two
lines intercepting the abscissa and ordinate at
points representing the mean cortisol response
and HDL cholesterol values. The figure shows
that eight of the 11 aggressive patients Ž72.7%.
98
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
fell into the lower left quadrant whereas only six
of the 27 non-aggressive patients Ž22.2%. did.
Lower HDL cholesterol levels were associated
with more intense ‘activation-euphoria’ and ‘high’
responses. GLM repeated measures ANCOVAs
revealed a significant association between HDL
cholesterol and the ‘activation-euphoria’ response
Ž F1,33 s 5.52, Ps 0.025. as well as between HDL
cholesterol and the ‘high’ response Ž F1,33 s 4.93,
P s 0.033.. The relationship between HDL
cholesterol and the ‘high’ double deltas is illustrated in Fig. 3. This figure also indicates that
aggressive patients experienced a moderate or
marked ‘high’ more frequently than non-aggressive patients Ž6r11 or 54.5% vs. 6r27 or 22.2%..
Using the same statistical analyses, total and
LDL cholesterol were not found to be associated
with any of the neuroendocrine and psychological
responses to m-CPP.
Fig. 3. Relationship between HDL cholesterol levels and ‘high’
responses to m-CPP expressed as double deltas. Solid circles
represent patients with a history of aggression, open circles,
patients without such a history. Lower HDL cholesterol levels
were associated with more intense ‘high’ responses. A GLM
repeated measures ANCOVA, revealed a significant association between HDL cholesterol and the intensity of the ‘high’
response Ž P s 0.033.. Six of the 11 aggressive patients Ž54.5%.
experienced a moderate or marked ‘high’ whereas only six of
the 27 non-aggressive patients Ž22.2%. did.
4. Discussion
4.1. Cholesterol and psychopathology
Fig. 2. Relationship between HDL cholesterol levels and cortisol responses to m-CPP expressed as double deltas. Solid
circles represent patients with a history of aggression, open
circles, patients without such a history. The solid line is the
regression line. Lower HDL cholesterol levels were associated
with blunted cortisol responses. A GLM repeated measures
ANCOVA revealed a significant association between HDL
cholesterol and the intensity of the cortisol response Ž Ps
0.018.. The dotted lines intercept the abscissa and ordinate at
points corresponding to the mean HDL cholesterol and cortisol values. Eight of the 11 aggressive patients Ž72.7%. fell into
the lower left quadrant whereas only six of the 27 non-aggressive patients Ž22.2%. did.
This study, conducted in personality disordered
cocaine addicts withdrawn from cocaine for 12]19
days, revealed that patients who had a past history of aggression had significantly lower levels of
HDL cholesterol. Our data confirm the previously
reported association between low cholesterol levels and violence. In several meta-analyses of
cholesterol-lowering trials, more violent deaths
were found among patients whose cholesterol had
been reduced ŽGolomb, 1998.. An association
between low cholesterol levels and violence was
also found in individuals who did not undergo
cholesterol-lowering interventions but had low
cholesterol values such as forensic patients ŽSpitz
et al., 1994., criminals with an antisocial personality disorder ŽVikkunen, 1979. and adolescent boys
with aggressive conduct disorder ŽVikkunen and
Pentinnen, 1984.. In addition, low cholesterol was
found in groups of patients whose aggression was
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
directed against the self instead of others
ŽLindberg et al., 1992; Modai et al., 1994; Sullivan
et al., 1994; Golier et al., 1995; Kunugi et al.,
1997..
In most studies levels of total cholesterol were
examined and not those of the HDL and LDL
fractions. A few studies assessed HDL and LDL
cholesterol levels in addition to total cholesterol.
Glueck et al. Ž1994., who studied patients with
affective disorders and controls, found that the
patients had lower plasma total, HDL and LDL
cholesterol levels than controls. Maes et al. Ž1997.
also examined total, HDL and LDL cholesterol in
patients with major depression and control subjects and found significantly lower serum HDL
cholesterol in the depressed subjects. They also
observed that total cholesterol was lower in the
patients but that the difference between patients
and controls was less significant. HDL cholesterol
was significantly lower in depressed patients who
had made serious suicide attempts than in those
who had not. These data led Maes et al. to
speculate that the most important change in
serum lipid composition in depressed subjects
could occur in HDL cholesterol rather than in
total or LDL cholesterol. In the present study, we
made similar observations in patients who exhibited problems in aggression and impulse control.
4.2. Cholesterol and 5-HT acti¨ ity
Our second finding is that of a significant association between low HDL cholesterol and indices
of reduced 5-HT function. We observed that low
HDL cholesterol was significantly associated with
an increased intensity of some psychological responses to m-CPP, namely the ‘high’ and ‘activation-euphoria’ responses. m-CPP is considered to
be a predominantly post-synaptic agonist and the
increase in 5-HT activity that follows its administration leads to post-synaptically mediated events
which include acute psychological and neuroendocrine changes. Several investigators who studied the effects of m-CPP in groups of psychiatric
patients with hypothesized serotonin deficits reported psychological effects similar to those we
observed. Following the administration of m-CPP,
patients diagnosed as having a Borderline Person-
99
ality Disorder were found to have increased ‘high,
happy and depersonalization’ ratings ŽHollander
et al., 1994., alcoholics were observed to experience ‘ethanol-like’ feelings ŽBenkelfat et al., 1991;
by Krystal et al., 1994. and patients with a seasonal affective disorder were reported to experience
an increase in ‘activation-euphoria’ ŽJacobsen et
al., 1993.. In these studies psychological changes
produced by m-CPP were interpreted as being
indicative of alterations in 5-HT function.
We observed a significant association between
low HDL cholesterol and a blunted cortisol response to m-CPP. Blunted neuroendocrine responses to m-CPP have been observed in various
psychiatric conditions and interpreted as indicative of 5-HT alterations. A decreased neuroendocrine responsivity to m-CPP has been found by
ourselves ŽBuydens-Branchey et al., 1997. and
others ŽHollander et al., 1994. to coexist with an
increased psychological responsivity and points to
the existence of variations in the nature of 5-HT
alterations in different brain regions. A recent
study using positron emission tomography to explore whether a reduced serotonergic function
occurs in brain regions that may play a role in the
modulation of aggression Žsuch as the orbital,
frontal and cingulate cortex. was conducted in
impulsive-aggressive patients and healthy volunteers ŽSiever et al., 1999.. In this study, regional
glucose metabolism was assessed following the
administration of the 5-HT releaserrreuptake inhibitor, D,L-fenfluramine Ž60 mg p.o... Compared
with controls, patients showed significantly
blunted metabolic responses in the orbital, frontal
and adjacent ventral medial and cingulate cortex
but not in the inferior parietal lobe. This study
provides direct evidence that differences in the
metabolic responsivity of healthy subjects and impulsive-aggressive patients to a serotonergic probe
can be found in some but not in all brain regions.
Few studies have investigated relationships
between cholesterol concentrations and 5-HT activity. In an in vitro study, Heron et al. Ž1980.
demonstrated that the incubation of mouse brain
membrane preparations with cholesterol hemisuccinate increased their viscosity and was accompanied by an increase in the binding of 5-HT.
Steegmans et al. Ž1996. found that plasma sero-
100
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
tonin concentrations were lower in men found to
have persistently low serum cholesterol concentrations. Muldoon et al. Ž1992. observed that the
prolactin response of monkeys fed a low-fat and
low-cholesterol diet to the serotonergic probe
fenfluramine was significantly lower than the response of animals fed a high-fat and highcholesterol diet. In a study of relationships
between hormonal responses to m-CPP and serum
cholesterol conducted in 10 healthy Japanese subjects, Terao et al. Ž1997. observed a significant
correlation between low cholesterol and blunted
cortisol and prolactin responses. The authors interpreted their findings as indicating that cholesterol levels may be associated with serotonergic
receptor function. In the present study, an association was found between low HDL cholesterol
and a blunted cortisol response to m-CPP. This
was not the case for the prolactin response. One
can only speculate about the reasons for this
discrepancy. Whereas some of the mechanisms
mediating the secretion of prolactin and cortisol
following m-CPP administration are similar, other
mechanisms contribute more specifically to the
release of each hormone separately. In a previous
study, involving 31 of the 38 patients who are the
topic of the present report, we reported that
there was no difference between the prolactin
responses to m-CPP of aggressive and non-aggressive patients, although the responses of the
two patient groups were blunted by comparison
with those of controls ŽBuydens-Branchey et al.,
1997.. By contrast, the aggressive patients’ cortisol response was significantly blunted by comparison with that of both non-aggressive patients and
controls. Thus, a significant association between
low HDL cholesterol and a blunted hormonal
response was found only for cortisol whose responsivity to m-CPP differentiated aggressive
from non-aggressive patients.
4.3. Cholesterol, psychopathology and 5-HT acti¨ ity
Although a number of studies have reported
associations between low cholesterol and problems in anger, mood and impulse control as well
as between low cholesterol and 5-HT function,
there is still a dearth of evidence concerning the
existence of links between low cholesterol, reduced 5-HT activity and psychopathology. The
most compelling evidence for the possible existence of such links can be derived from a study
performed by Kaplan et al. Ž1994. in monkeys.
These investigators artificially manipulated the
amounts of cholesterol consumed by cynomolgus
macaques. The animals were given diets high in
fat and either rich or low in cholesterol. After
they had consumed these diets for an average of 4
months, behavioral observations started and continued for 8 months. Plasma cholesterol and cerebrospinal fluid ŽCSF. metabolites of 5-HT,
norephinephrine and dopamine were assessed 4
and 5.5 months after the initiation of behavioral
observations. Monkeys fed the low-cholesterol
diet were found to have a significantly lower
plasma cholesterol level and to be more aggressive and less affiliative. They also had lower CSF
concentrations of the 5-HT metabolite, 5-hydroxyindol acetic acid. No relationship was found
between cholesterol and the norepinephrine and
dopamine metabolites. These data show that the
use of a low-cholesterol diet affects simultaneously plasma cholesterol levels, behavior and
serotonergic activity. Our finding of simultaneous
associations between low HDL cholesterol and
indices of altered serotonergic function could be
understood in the context of Kaplan et al.’s study.
Mechanisms underlying the association among
cholesterol, behavior and 5-HT activity have not
yet been elucidated. It has been suggested that
reduced plasma cholesterol could depress the
cholesterolrphospholipid ratio in neuronal membranes with consequent alterations in membrane
fluidity, viscosity and function, including the function of 5-HT receptors and of the 5-HT transporter. It has also been suggested that low cholesterol may be accompanied by a decrease in plasma
levels of the 5-HT precursor tryptophan ŽEngelberg, 1992; Salter, 1992..
In light of the large number of individuals who
attempt to lower their blood cholesterol by diet,
drugs or both, additional studies are needed to
confirm preliminary findings of a functional association between hostile or self-destructive behavior, cholesterol and serotonin. If existing findings
were confirmed, a number of questions would
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
need to be answered. For example, one could
wonder whether it is the total cholesterol or one
of its subfractions that plays the most important
role in behavioral problems. More importantly,
one could also wonder whether individuals who
exhibit aggression, impulsiveness or suicidal tendencies following cholesterol-lowering interventions present characteristics antedating these
treatments that might predict adverse outcomes.
In the present study, some subjects classified as
non-aggressive had cholesterol levels as low as
those of the aggressive patients. Moreover, normal HDL cholesterol values range from 27 mgrdl
to 67 mgrdl. Patients classified as aggressive had
mean HDL cholesterol values in the lower range
of values found in normal subjects. Low cholesterol values are thus not necessarily associated
with an increase in aggression or impulsiveness.
On the other hand, evidence reviewed above indicates that in some individuals, low or lowered
cholesterol and behavioral problems might be
causally related. In these individuals, the suffering due to adverse psychological effects may outweigh the beneficial effects of lowered cholesterol
on heart disease. Although it is not presently
known who might be at risk for developing behavioral problems in the context of cholesterol-lowering interventions, caution might have to be exercised in the treatment of individuals who have a
history of psychiatric disorders.
Acknowledgements
This study was supported by the Department of
Veterans Affairs and by research grant DA-09468
from the National Institute on Drug Abuse,
Rockville, MD.
References
Allain, C.C., Poon, L.S., Chan, C.S., Richmond, W., Fu, P.C.,
1974. Enzymatic determination of total serum cholesterol.
Clinical Chemistry 20, 470]475.
Barratt, E.S., 1957. Factor analysis of some psychometric
measures of impulsiveness and anxiety. Psychological Reports 21, 343]349.
Benkelfat, C., Murphy, D.L., Hill, J.I., George, D.T., Nutt, D.,
101
Linnoila, M., 1991. Ethanol-like properties of the serotonergic partial agonist m-chlorophenylpiperazine in chronic
alcoholic patients. Archives of General Psychiatry 48, 383.
Brown, G.L., Ballenger, J.C., Minichiello, M.D., Goodwin,
F.K., 1981. Human aggression and its relationship to CSF
5-HIAA, MHPG, and HVA. In: Sandler, M. ŽEd.., Psychopharmacology of Aggression. Raven Press, New York,
pp. 131]147.
Buss, A.H., Durkee, A., 1957. An inventory for assessing
different kinds of hostility. Journal of Consulting and Clinical Psychology 21, 343]349.
Buydens-Branchey, L., Branchey, M., Fergeson, P., Hudson, J.,
McKernin, C., 1997. The meta-chlorophenylpiperazine
challenge test in cocaine addicts: hormonal and psychological responses. Biological Psychiatry 41, 1071]1086.
Charney, D.S., Woods, S.W., Goodman, W.K., Heninger, G.R.,
1987. Serotonin function in anxiety. II. Effects of the serotonin agonist m-CPP in panic disorder patients and healthy
subjects. Psychopharmacology 92, 14]24.
Engelberg, H., 1992. Low serum cholesterol and suicide.
Lancet 339, 727]729.
Glueck, C.J., Tieger, M., Kunkel, R., Hamer, T., Tracy, T.,
Speirs, J., 1994. Hypocholesterolemia and affective disorders. American Journal of the Medical Sciences 308,
218]225.
Golier, J.A., Marzuk, P.M., Leon, A.C., Weiner, C., Tardiff,
K., 1995. Low serum cholesterol level and attempted suicide. American Journal of Psychiatry 152, 419]423.
Golomb, B.A., 1998. Cholesterol and violence: is there a
connection? Annals of Internal Medicine 128, 478]487.
Heron, D.S., Shinitzky, M., Hershkowitz, M., Samuel, D., 1980.
Lipid fluidity markedly modulates the binding of serotonin
to mouse brain membranes. Proceedings of the National
Academy of Sciences of the United States of America 77,
7463]7467.
Hollander, E., Stein, D.J., DeCaria, C.M., Cohen, L., Saoud,
J.B., Skodol, A.E., Kellman, D., Rosnick, L., Oldham, J.M.,
1994. Serotonergic sensitivity in borderline personality disorder: preliminary findings. American Journal of Psychiatry
151, 277]280.
Jacobsen, F.M., Mueller, E.A., Rosenthal, N.E., Rogers, S.,
Hill, J.L., Murphy, D.L., 1993. Behavioral response to intravenous m-CPP in patients with seasonal affective disorder
and control subjects before and after phototherapy. Psychiatry Research 52, 181]197.
Kaplan, J.R., Shively, C.A., Fontenot, M.B., Morgan, T.M.,
Howell, S.M., Manuck, S.B., Muldoon, M.F., Mann, J.J.,
1994. Demonstration of an association among dietary
cholesterol, central serotonergic activity, and social behavior in monkeys. Psychosomatic Medicine 56, 479]484.
Krystal, J.H., Webb, E., Cooney, E., Kranzler, H.R., Charney,
D.S., 1994. Specificity of ethanollike effects elicited by
serotonergic and noradrenergic mechanisms. Archives of
General Psychiatry 51, 898]911.
Kunugi, H., Takei, N., Aoki, H., Shinichiro, N., 1997. Low
serum cholesterol in suicide attempters. Biological Psychiatry 41, 196]200.
102
L. Buydens-Branchey et al. r Psychiatry Research 93 (2000) 93]102
Lindberg, G., Rastam, L., Gullberg, B., Eklund, G.A., 1992.
Low serum cholesterol concentration and short term mortality from injuries in men and women. British Medical
Journal 305, 277]279.
McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models.
Chapman & Hall, New York.
Maes, M., Smith, R., Christophe, A., Vandoolaeghe, E., Van
Gastel, V., Neels, H., Desmedts, P., Wauters, A., Meltzer,
H.Y., 1997. Lower serum high-density lipoprotein cholesterol ŽHDL-C. in major depression and in depressed men
with serious suicidal attempts: relationship with immuneinflammatory markers. Acta Psychiatrica Scandinavica 95,
212]221.
Modai, I., Valevski, A., Dror, S., Weizman, A., 1994. Serum
cholesterol levels and suicidal tendencies in psychiatric
patients. Journal of Clinical Psychiatry 55, 252]254.
Muldoon, M.F., Manuck, S.B., Mathews, K.M., 1990. Lowering
cholesterol concentration and mortality: a quantitative review of primary prevention trials. British Medical Journal
301, 309]314.
Muldoon, M.F., Kaplan, J.R., Manuck, S.B., Mann, J.J., 1992.
Effects of a low-fat diet on brain serotonergic responsivity
in Cynomolgus monkeys. Biological Psychiatry 31, 739]742.
Muldoon, M.F., Rossouw, J., Manuck, S.B., Glueck, C.J., Kaplan, J.R., Kaufmann, P.G., 1993. Low or lowered cholesterol and risk of death from suicide and trauma. Metabolism
42 ŽSupplement 1., 45]56.
Murphy, D.L., Mueller, E.A., Hill, J.L., Tolliver, T.J., Jacobsen, F.M., 1989. Comparative anxiogenic, neuroendocrine and other physiologic effects of m-CPP given intravenously or orally to healthy volunteers. Psychopharmacology 98, 275]282.
Salter, M., 1992. Low serum cholesterol and suicide. Lancet
339, 1169.
Siever, L.J., Buchsbaum, M.S., New, A.S., Spiegel-Cohen, J.,
Wei, T., Hazlett, E.A., Sevin, E., Nunn, M., Mitropoulou,
V., 1999. D,L-Fenfluramine response in impulsive personal-
ity disorder assessed with w 18 Fxfluorideoxyglucose positron
emission tomography. Neuropsychopharmacology 20,
413]423.
Spitz, R., Hillbrand, M., Foster Jr., H.J., 1994. Serum cholesterol levels and frequency of aggression. Psychological Reports 74, 622.
Spitzer, R.L., Williams, J.B., Gibbon, M., First, M.B., 1990.
User’s Guide for the Structured Clinical Interview of the
DSM-III-R. American Psychiatric Press, Washington, DC.
Steegmans, P.H.A., Fekkes, D., Hoes, A.W., Bak, A.A.A., van
der Does, E., Diederick, E.G., 1996. Low serum cholesterol
concentration and serotonin metabolism in men. British
Medical Journal 312, 221.
Suckow, R.F., Cooper, T.B., Kahn, R.S., 1990. High-performance liquid chromatographic method for the analysis of
plasma m-chlorophenylpiperazine. Journal of Chromatography 528, 228]234.
Sullivan, P.F., Joyce, P.R., Bulik, C.M., Mulder, R.T., OakleyBrowne, M., 1994. Total cholesterol and suicidality in depression. Biological Psychiatry 36, 472]477.
Terao, T., Yoshimura, R., Ohmori, O., Takano, T., Takahashi,
N., Iwata, N., Suzuki, T., Kazuhiko, A., 1997. Effect of
serum cholesterol levels on meta-chlorophenylpiperazineevoked neuroendocrine responses in healthy subjects. Biological Psychiatry 41, 974]978.
Vikkunen, M., 1979. Serum cholesterol in antisocial personality. Neuropsychobiology 5, 27]30.
Vikkunen, M., Pentinnen, H., 1984. Serum cholesterol in
aggressive conduct disorder: a preliminary study. Biological
Psychiatry 19, 435]439.
Warnick, G.R., Benderson, J., Albers, J.J., 1983. Dextran
sulfate-Mg 2q precipitation procedure for quantitation of
high-density lipoprotein cholesterol. In: Cooper, G.R. ŽEd..,
Selected Methods of Clinical Chemistry, Vol. 10. American
Association for Clinical Chemistry, Washington, DC, pp.
91]99.